| |
Last updated on October 21, 2024. This conference program is tentative and subject to change
Technical Program for Wednesday October 30, 2024
|
WeAT1 |
Avenue Ballroom E/W |
System Identification and Estimation Techniques |
Regular Session |
Co-Chair: Abdollahi Biron, Zoleikha | University of Florida |
|
09:15-09:18, Paper WeAT1.1 | |
State of Health Estimation Using Temperature Sensors During Charging in Li-Ion Batteries |
|
Ahuja, Nitisha | The Pennsylvania State University |
Bhaskar, Kiran | The Pennsylvania State University |
Rackaitis, Mindaugas | Avery Dennison |
Collin, Moore | Avery Dennison |
Rahn, Christopher D. | Penn State Univ |
Keywords: Power and Energy Systems, Estimation, Modelling, Identification and Signal Processing
Abstract: Accurate prediction of State of Health (SOH) is crucial to ensuring battery system performance and safety. Unlike traditional SOH estimation methods that are based on battery voltage and current, we propose using only surface temperature measurements. In this way, SOH can be estimated without requiring an interface to each cell’s power terminals, using, for example, a temperature-sensing Radio Frequency Identification (RFID) tag that is bonded to the cell surface. An RFID-enabled charger or puck inside the pack can read the RFID tags at a high rate to measure the surface temperature and time history of every cell in the pack during constant current charging. SOH is estimated from the increase in surface temperature over the life of the battery due to the increase in resistance associated with aging. The proposed SOH estimation algorithm is validated with an open-source battery cycling dataset. We found a robust correlation between estimated and actual battery internal resistance, particularly under well-controlled ambient temperatures and fast charging.
|
|
09:18-09:21, Paper WeAT1.2 | |
a Time Series Stackelberg Game for the Load Management Considering Unexpected EV Behaviors |
|
Hu, Miaomiao | University of Florida |
Kushwaha, Dhruv | University of Florida |
Abdollahi Biron, Zoleikha | University of Florida |
Keywords: Control of Smart Buildings and Microgrids, Power and Energy Systems, Transportation Systems
Abstract: Unexpected Electric Vehicle (EV) behaviors can cause severe power flow fluctuations in the real world. By encouraging EV users to discharge at peak hours and charge during off-peak hours can largely relieve the load pressure caused by sudden EV integration. Considering the uncertainty of user behavior, this paper proposes a time series Stackelberg game with the sliding window method to deal with sudden changes in EV user behaviors and realize load management in the microgrid. Simulation results show that with our proposed time series method, the mean-square error between the ideal load and actual load can be reduced by 42.54% and the maximum social welfare can be increased by 65.47% compared to the traditional Stackelberg game.
|
|
09:21-09:24, Paper WeAT1.3 | |
Black-Box Modelling of Non-Stationary N2O Dynamics in a Full-Scale Wastewater Treatment Plant |
|
Hansen, Laura Debel | Aalborg University |
Stentoft, Peter Alexander | Technical University of Denmark |
Ortiz Arroyo, Daniel | Aalborg University |
Durdevic, Petar | Aalborg University Esbjerg |
Keywords: Machine Learning in modeling, estimation, and control, Modeling and Validation, Modelling and Control of Biomedical Systems
Abstract: Using a data-driven approach, we present and compare linear and nonlinear methods for system identification of the potent greenhouse gas, nitrous oxide (N2O), which is produced during the biological treatment of wastewater. N2O is challenging to estimate, as the full understanding of its production process is yet to be determined. Therefore, data-driven approaches hold promise in advancing our understanding and offering solutions for model-based control, fault detection, and analysis. We present two methods for modelling the N2O in a full-scale wastewater treatment plant; the long short-term memory (LSTM) and a linear ARX model and discuss the performance of these well-established models on real-world implementations. Results indicate that the nonlinear LSTM model has enhanced performance when compared to the linear ARX. While single-step predictions exhibit minimal mean squared error (MSE), the time-invariant models struggle to capture the production mechanisms over multi-step predictions due to the excessive need of multi-year data and non-stationarity and non-normality of the predicted variable.
|
|
09:24-09:27, Paper WeAT1.4 | |
Synchronization of Multiple Non-Identical Fractional-Order Flexible Link Manipulators Via Adaptive Neuro-Fuzzy Sliding Mode Control |
|
Kharabian, Behrouz | Kent State University |
Mirinejad, Hossein | Kent State University |
Keywords: Control Design, Nonlinear Control Systems, Robotics
Abstract: This paper presents a novel method utilizing adaptive neuro-fuzzy sliding mode control (ANFSMC) to synchronize multiple non-identical fractional-order flexible link manipulators. The synchronization framework employs two flexible manipulators with unknown parameters as slave systems, in parallel with two master flexible manipulators having distinct known parameters. The slave flexible link manipulators (SFLMs) achieve synchronization with each master flexible link manipulator (MFLM) both before and after the switching mode—a phase where the MFLM transitions. A multi-layer perceptron neural network is used to adjust the continuous part of the sliding mode control (SMC), while a neuro-fuzzy system, eliminating the need for predetermined knowledge of upper bounds of unknown parameters, estimates the SMC’s discontinuous part. The ANFSMC approach includes adaptive estimation of unknown parameters and disturbances. Additionally, the unknown exponential term of the SFLMs is determined using a Takagi-Sugeno (T-S) fuzzy observer. The control system’s stability is verified using a Lyapunov approach. Comparative studies highlight the superiority of our proposed method over state-of-the-art approaches such as pure SMC, active control, and backstepping control, which are typically applied to synchronize highly nonlinear systems.
|
|
09:27-09:30, Paper WeAT1.5 | |
The Effect of Different Tyres on the Performance of an Anti-Lock Braking System for Continuous Modulation Actuators |
|
Taglione, Luca | Politecnico Di Milano |
Bernal, Martina | Dipartimento Di Elettronica Informazione E Bioingegneria, Polite |
Corno, Matteo | Politecnico Di Milano |
Savaresi, Sergio | Politecnico Di Milano |
Keywords: Automotive Systems, Control Applications, Control Design
Abstract: The Anti-lock Braking System (ABS) plays a role of paramount importance in the safety of ground-wheeled vehicles. It prevents the wheels from locking up during panic braking maneuvers, minimizing the braking distance while ensuring vehicle driveability. Recent technological advancements involving the braking actuation unit offer the opportunity for enhancing traditional braking active safety systems. This work proposes an analysis -- from a controller designer perspective -- focused on how the tyre characteristics (compound, geometry, inflation pressure, temperature, wear) affect the performance of a state-of-the-art ABS for continuous modulation braking actuators, with the objective of determining how the latter can be improved by adapting the control logic parameters according to the tyre mounted. To this aim, an experimental campaign is carried out to identify the friction characteristics of a heterogeneous set of tyres that can be mounted on a passenger car; the resulting friction curves are then used in a high-fidelity simulation environment (VI-grade CarRealTime) to develop a tyre-adaptive and a standard versions of the aforementioned logic. The braking performance is then compared in panic braking maneuver scenarios exploiting suitable key performance indicators.
|
|
09:30-09:33, Paper WeAT1.6 | |
System Identification Applied to a Shale Shaker Using Linear and Non-Linear Models |
|
Estevao, Gedraite | University of Sao Paulo |
Barbosa, Vinícius Pimenta | Universidade Federal De Uberlândia |
Barbosa, Rafael | Federal University of Uberlândia |
Waldmann, Alex | PETROBRAS |
Gedraite, Rubens | UFU |
Garcia, Claudio | Polytechnic School of the Univ of Sao Paulo |
Keywords: Modeling and Validation, Electromechanical systems, Control Design
Abstract: This work presents one analysis of linear and nonlinear techniques to model the shale shaker. Due to the absence of literature on this regard, the authors tested multiple models and evaluated the results to present a viable model to be used as a reference for further studies. The results shows that non-linear models outperform liner models and that measuring a disturbance variable of the fluid flow improves the quality of models, as these aspects are a source of non-linearities and deeply affects the performance of the model. At the conclusion, future work is pointed out based on these results.
|
|
09:33-09:36, Paper WeAT1.7 | |
Energy-Based Data Sampling for Traffic Prediction with Small Training Datasets |
|
Yang, Zhaohui | Google LLC |
Jerath, Kshitij | University of Massachusetts Lowell |
Keywords: Machine Learning in modeling, estimation, and control, Transportation Systems
Abstract: Despite extensive domain knowledge of traffic flow dynamics, existing deep learning methodologies are still unable to make accurate fine-scale predictions. To address this issues, we propose an innovative solution combining Convolutional Neural Networks (CNNs) with Long Short-Term Memory (LSTM) architecture to improve traffic flow prediction. Large deep learning models are computationally intensive and memory-hungry, making them impractical for resource-constrained environments. Instead of model compression, a data-compression or data-sampling approach could fundamentally solve these challenges. A key finding of our research is the feasibility of sampling training data for large traffic systems using simulations on smaller systems. This insight suggests the potential for referencing a macroscopic-level energy distribution to inform the sampling of microscopic data. Such sampling is enabled by the observed scale invariance in the normalized energy distribution of the energy-driven dynamical models, streamlining the data generation process for large-scale traffic systems. Simulations demonstrate promising agreement between predicted and actual traffic flow dynamics, highlighting the efficacy of our approach.
|
|
09:36-09:39, Paper WeAT1.8 | |
Robust Auto-Tuning Control of a Delivery Quadcopter with Motor Faults, Mass and Inertia Estimation |
|
Mafaz, Mohamed Ahsan | Coventry University |
Horri, Nadjim Mehdi | University of Leicester |
Lu, Qian | Coventry University |
England, Matthew | Coventry University |
Keywords: Control Applications, Unmanned Ground and Aerial Vehicles, Uncertain Systems and Robust Control
Abstract: Unstable position and attitude caused by severe motor faults and varying payloads complicates the design of a reliable flight controller for delivery quadcopters during last-mile deliveries. To ensure safe and efficient operation, a novel, robust and adaptive control strategy is employed using a two-stage recursive least square based estimation, combined with an auto-tuning mixed-sensitivity H∞ controller. A numerical simulation demonstrates fast and accurate real-time estimation of actuator loss of effectiveness and changes in the mass and inertia due to payload pickup/drop off during delivery missions. The combined and uneven effects of multiple faults with initial mass and inertia uncertainties are efficiently compensated by this novel fault tolerant flight control strategy. The controller also improves the rejection of these uncertainties during the transient stage before the convergence of the parameter estimation. The proposed approach maintains the operational reliability of delivery quadcopters under a range of challenging uneven faults and significant mass and inertia uncertainties.
|
|
09:39-09:42, Paper WeAT1.9 | |
Real Time Camera-Based Sideslip Angle Estimation: Design and Experiments |
|
Serena, Leonardo | Università Di Padova |
Bruschetta, Mattia | University of Padova |
Righetti, Giovanni | Università Di Padova |
de Castro, Ricardo | University of California, Merced |
Lenzo, Basilio | University of Padua |
Keywords: Modelling, Identification and Signal Processing, Estimation, Automotive Systems
Abstract: Vehicle stability controllers and car modeling significantly rely on the knowledge of the vehicle velocity and sideslip angle, i.e. the angle between the velocity vector and the vehicle's heading. While various estimation techniques have been proposed over time, their reliability across generic driving scenarios remains unsatisfactory. This study investigates a novel methodology to directly measure velocity and sideslip angle, using computer vision techniques. A real-time sensing framework is put in place on a full-scale passenger vehicle, with the camera setup allowing to send signals over the vehicle Controller Area Network (CAN). Experiments, along several manoeuvres, confirm the real-time applicability and effectiveness of the proposed approach against the state-of-the-art sensor Kistler S-Motion (ground truth).
|
|
09:42-09:45, Paper WeAT1.10 | |
A New Simple-To-Configure Self-Perturbing Multivariable Extremum-Seeking Controller |
|
Salsbury, Timothy | Pacific Northwest National Laboratory |
Yu, Min Gyung | Pacific Northwest National Laboratory |
Keywords: Adaptive and Learning Systems, Nonlinear Control Systems, Control Design
Abstract: This paper presents a new stochastic relay-based extremum-seeking controller (ESC) for multi-input-single-output (MISO) systems. The goal of this work was to create an algorithm that is much simpler to configure than alternative approaches making deployment to real-world problems easier. A solution is developed first for a static map and then adapted for a general class of dynamic systems. The number of configurable parameters is one per input channel for the static case and only one additional parameter is needed for the dynamic version. The problem of gradient identification is solved via the use of stochastic relay gains and a simple stability proof for the static case is presented. Simulation tests demonstrate the performance of the strategy for optimizing both static and dynamic systems.
|
|
09:45-09:48, Paper WeAT1.11 | |
Comparative Evaluation of Control-Oriented Heavy Duty Vehicle Air Drag Coefficient Models |
|
Best, Micah | Texas A&M University |
Sujan, Vivek | Oak Ridge National Laboratory |
Zhou, Anye | Oak Ridge National Laboratory |
Cook, Adian | Oak Ridge National Laboratory |
Wang, Zejiang | The University of Texas at Dallas |
Keywords: Automotive Systems
Abstract: Heavy-duty vehicles (HDVs) are a significant source of fuel consumption and greenhouse gas emissions, prompting solutions such as HDV platooning to mitigate these negative impacts through air drag reduction. The intervehicle distance in an HDV platoon needs to be carefully selected, such that the platoon-level energy efficiency and safety considerations can be well balanced. Underlying this problem lies in accurately modeling the relationship between HDV air drag coefficient and intervehicle distance. Through comprehensive evaluation and comparison, we analyze five control-oriented HDV air drag coefficient models, including the polynomial model, rational polynomial model, rational model, semi-quadratic model, and ridge model. Leveraging Scipy Curve-Fit toolbox and our previously compiled air drag coefficient datasets, we optimally identify the parameters inside each model. The calibrated models are then thoroughly evaluated via five complementary metrics. The comparison results reveal that the semi-quadratic model has the highest overall performance, while the widely adopted rational model only exhibits suboptimal performance.
|
|
09:48-09:51, Paper WeAT1.12 | |
Improving Battery Pack SOC Estimation through Multi-Chemistry Hybridization |
|
Casten, Casey | University of Maryland, College Park |
Fathy, Hosam K. | University of Maryland |
Keywords: Estimation, Power and Energy Systems
Abstract: This paper examines the accuracy with which one can estimate the state of charge (SOC) of a series string of battery cells with distinct chemistries. Accurate SOC estimation is an integral function of any battery management system (BMS): it helps in avoiding cycling a battery beyond its capacity limits, which has the potential to accelerate degradation and failure. Previous work in the literature quantifies limitations in SOC estimation accuracy, and attempts to address them through improved battery modeling, improved estimation algorithms, and the creation of series battery-capacitor packs. However, to the best of the authors' knowledge, this is the first body of work quantifying and demonstrating the degree to which the series hybridization of distinct battery chemistries can help improve SOC estimation accuracy. The paper derives Cramér-Rao bounds for the error variance with which one can estimate SOC, with and without series hybridization. This is followed by a Monte Carlo simulation of SOC estimation for series-connected commercial LiFePO4 (LFP) and LiNiMnCoO2 (NMC) battery cells, based on laboratory-characterized models of these cells. Both of the above analytic study and Monte Carlo simulation study show a significant potential for improving battery pack SOC estimation accuracy through series hybridization.
|
|
09:51-09:54, Paper WeAT1.13 | |
Permanent Magnet Synchronous Motor Speed and Position Estimation Using Reduced-Order Extended Kalman Filter |
|
Su, Jiayi | Marquette University |
Yaz, Edwin | Marquette University |
Schneider, Susan | Marquette University |
Keywords: Estimation, Modelling, Identification and Signal Processing, Stochastic Systems
Abstract: Permanent Magnet Synchronous Motor (PMSM) plays a pivotal role in many applications. Precise control of the PMSM necessitates accurate estimation of its states, particularly the motor's speed and position. Conventionally, a full-order Extended Kalman filter (EKF) is employed for state estimation. However, since the winding currents are directly measurable, deploying a full-order observer to estimate all states becomes unnecessary. In this work, a novel Reduced-order Extended Kalman filter (ROEKF) is introduced to estimate the speed and position of a two-phase PMSM. Simulation results show that the proposed ROEKF brings equivalent estimation accuracy compared to the full-order EKF, while it reduces the computation cost significantly. In addition, the proposed reduced-order filtering approach can be easily adapted to other applications to reduce the computation cost as well.
|
|
09:54-09:57, Paper WeAT1.14 | |
Analytical Gradient and Hessian Evaluation for System Identification Using State-Parameter Transition Tensors |
|
Saha, Premjit | University at Buffalo |
Singh, Tarunraj | State Univ. of New York at Buffalo |
Keywords: Modelling, Identification and Signal Processing, Estimation, Education for modeling, estimation and control
Abstract: In this work, the Einstein notation is utilized to synthesize state and parameter transition matrices, by solving a set of ordinary differential equations. Additionally, for the system identification problem, it has been demonstrated that the gradient and Hessian of a cost function can be analytically constructed using the same matrix and tensor metrics. A general gradient-based optimization problem is then posed to identify unknown system parameters and unknown initial conditions. Here, the analytical gradient and Hessian of the cost function are derived using these state and parameter transition matrices. The more robust performance of the proposed method for identifying unknown system parameters and unknown initial conditions over an existing conventional quasi-Newton method-based system identification toolbox (available in MATLAB) is demonstrated by using two widely used benchmark datasets from real dynamic systems. In the existing toolbox, gradient and Hessian information are derived using a finite difference method, which are susceptibility to numerical errors compared to the analytical approach presented.
|
|
09:57-10:00, Paper WeAT1.15 | |
Estimation Sample Complexity of a Class of Nonlinear Continuous-Time Systems |
|
Kuang, Simon | UC Davis |
Lin, Xinfan | University of California, Davis |
Keywords: Modelling, Identification and Signal Processing, Machine Learning in modeling, estimation, and control, Estimation
Abstract: We present a method of parameter estimation for large class of nonlinear systems, namely those in which the state consists of output derivatives and the flow is linear in the parameter. The method, which solves for the unknown parameter by directly inverting the dynamics using regularized linear regression, is based on new design and analysis ideas for differentiation filtering and regularized least squares. Combined in series, they yield a novel finite-sample bound on mean absolute error of estimation.
|
|
10:00-10:03, Paper WeAT1.16 | |
Numerical Implementation of Deep Neural PDE Backstepping Control of Reaction-Diffusion PDEs with Delay |
|
Wang, Shanshan | University of Shanghai for Science and Technology |
Diagne, Mamadou | University of California San Diego |
Krstic, Miroslav | Univ. of California at San Diego |
Keywords: Machine Learning in modeling, estimation, and control, Distributed Parameter Systems, Control Design
Abstract: Deep neural networks that approximate nonlinear function-to-function mappings, i.e., operators, which are called DeepONet, have been demonstrated in recent articles to be capable of encoding entire PDE control methodologies, such as backstepping, so that, for each new functional coefficient of a PDE plant, the backstepping gains are obtained through a simple function evaluation. These initial results have been limited to single PDEs from a given class, approximating the solutions of only single-PDE operators for the gain kernels. In this paper, we expand this framework by numerically illustrating the approximation of multiple (cascaded) nonlinear operators. Multiple operators arise in the control of PDE systems from distinct PDE classes, such as the system in this paper: a reaction-diffusion plant, which is a parabolic PDE, with input delay, which is a hyperbolic PDE. The DeepONet-approximated nonlinear operator is a cascade/composition of the operators defined by one hyperbolic PDE of the Goursat form and one parabolic PDE on a rectangle, both of which are bilinear in their input functions and not explicitly solvable. For the delay-compensated PDE backstepping controller, which employs the learned control operator, namely, the approximated gain kernel, exponential stability is guaranteed in our companion manuscript, in the L2 norm of the plant state and the H1 norm of the input delay state. In this paper, learning of a neural operator is presented, along with PDE backstepping simulations that illustrate closed-loop stabilization
|
|
10:03-10:06, Paper WeAT1.17 | |
Nonlinear System Identification with Gaussian Processes Using Laguerre and Kautz Filters |
|
Illg, Christopher | University of Siegen |
Balar, Nishilkumar | University of Siegen |
Nelles, Oliver | University of Siegen |
Keywords: Machine Learning in modeling, estimation, and control, Modeling and Validation
Abstract: System identification can be used to determine data-driven mathematical models of dynamic processes. For nonlinear processes, model architectures that are as flexible as possible are required. One possibility is to utilize Gaussian processes (GPs) as a universal approximator with an external dynamics realization, leading to highly flexible models. Novel Laguerre and Kautz filter-based dynamics realizations in GP models are proposed. The Laguerre/Kautz pole(s) are treated as hyperparameters with the GPs' standard hyperparameter for the SE-ARD kernel. The two novel dynamics realizations in GP models are compared to different state-of-the-art dynamics realizations such as finite impulse response~(FIR) or autoregressive with exogenous input (ARX). The big data case is handled via support points. Using Laguerre and Kautz regressor spaces allows both the dimensionality of the regressor space to be kept small and achieve superior performance. This is demonstrated through numerical examples and measured benchmark data of a Wiener-Hammerstein process.
|
|
10:06-10:09, Paper WeAT1.18 | |
The Case for DeepSOH: Addressing Path Dependency for Remaining Useful Life in Li-Ion Batteries |
|
Movahedi, Hamidreza | University of Michigan |
Weng, Andrew | University of Michigan |
Pannala, Sravan | University of Michigan |
Siegel, Jason | University of Michigan |
Stefanopoulou, Anna G. | Univ of Michigan |
Keywords: Power and Energy Systems, Automotive Systems, Estimation
Abstract: The battery state of health (SOH) based on capacity fade (SOH-C), and resistance increase (SOH-R) is not sufficient for predicting the Remaining Useful life (RUL). The electrochemical community blames the path dependency of the battery degradation mechanisms for our inability to forecast future degradation. The control community knows that the path dependency is addressed by full state estimation. In this work, we demonstrate the inadequacy of popular definitions of State of Health (SOH). We demonstrate that even electrode-specific SOH (eSOH) and resistance estimations are not sufficient to completely characterize degradation even in a simplified model based on a single particle model (SPM) which includes SEI, plating, and mechanical fracture. We illustrate that it is possible to simulate different possible degradation trajectories and RULs from the same eSOH and second-life duty cycles. Therefore, the lifetime battery degradation of more complex models will be even less predictable, let alone in the real world. We finally define the deepSOH (states of the degradation mechanisms) that capture the individual contributions of each degradation mechanism to the total loss of lithium inventory. We show that adding other battery measurements, such as cell expansion, allows us to identify the deepSOH and, therefore, predict the RUL.
|
|
10:09-10:12, Paper WeAT1.19 | |
Accurate Co-Estimation Methods for Second-Life Battery Management Systems (BMS-2): Integrating State and Parameter Estimations |
|
Nguyen, Nhat | University of California, Los Angeles (UCLA) |
Liu, Zexiang | University of Michigan |
Cui, Xiaofan | UCLA |
Keywords: Power and Energy Systems, Estimation
Abstract: In the next ten years, over 200 GWh lithium-ion batteries from electric vehicles are projected to reach the end of their first life. The reuse of these batteries for stationary energy storage offers economic and environmental benefits. However, the heterogeneity and complicated aging behaviors of retired batteries demand a specialized second-life battery management system (BMS-2) that can accurately monitor their status. An online adaptive estimator is needed to address this challenge. In this study, we investigated and improved classic state-and-parameter co-estimation methods and propose a new approach integrating Newton’s method and Extended Kalman Filter (EKF) for co-estimation. Results demonstrate that the proposed method outperforms the parameter-augmented EKF and dual EKF methods, especially in addressing complicated aging behaviors in retired batteries.
|
|
10:12-10:15, Paper WeAT1.20 | |
Fine-Tuning Hybrid Physics-Informed Neural Networks for Vehicle Dynamics Model Estimation |
|
Fang, Shiming | Binghamton University |
Yu, Kaiyan | State University of New York at Binghamton |
Keywords: Machine Learning in modeling, estimation, and control, Automotive Systems, Estimation
Abstract: Accurate dynamics modeling is crucial for autonomous racing vehicles, especially during high-speed maneuvers. Traditional methods often rely on initial guesses, lengthy fitting processes, and complex testing setups. Purely data-driven machine learning approaches face challenges in capturing internal physical constraints and require abundant data to perform optimally. This paper introduces the Fine-Tuning Hybrid Dynamics (FTHD) method, which uses a hybrid approach for model training. FTHD requires a smaller dataset for training and demonstrates superior results compared to recent methods such as Deep Dynamics Model (DDM) or Deep Pacejka Model (DPM). The Bayesrace Physics-based Simulator is employed for data collection and training. The performance of the proposed FTHD is evaluated against the state-of-the-art method, DDM. The comparison includes assessing the root-mean-square-error (RMSE) of velocities, maximum errors, estimated Pacejka coefficients, and lateral forces against the ground truth. The results illustrate that, with identical hyperparameter configurations (training epochs, layer size, etc.), the proposed FTHD method outperforms recent PINN methods in accurately describing vehicle dynamics properties while utilizing less data.
|
|
WeAT3 |
Prime 1 |
Modeling, Estimation, and Control of Energy-Harvesting Systems |
Invited Session |
Chair: Hajj, Muhammad | Stevens Institute of Technology |
Co-Chair: Li, Perry Y. | Univ. of Minnesota |
|
09:15-09:30, Paper WeAT3.1 | |
Co-Design for Real-Time Adaptability: Methodology and Wind Energy Case Study (I) |
|
Fine, Jacob | University of Michigan |
Holbrook, Ian | University of Michigan |
Vermillion, Christopher | University of Michigan |
Keywords: Adaptive and Learning Systems, Stochastic Systems, Power and Energy Systems
Abstract: This work presents a unique control co-design formulation that explicitly optimizes the level of real-time adaptability in both the physical design and control parameters. Adaptability is particularly important for large-scale energy-harvesting systems, which operate in highly variable environments and are subject to long design and manufacturing cycles. Here, physical components and software often must be “frozen” relatively early, sometimes well-before the system’s dynamics been fully characterized. The proposed co-design framework performs a maximization of expected profit, accounting for a low-complexity surrogate model of the system’s performance, a statistical model of the environment, a statistical characterization of how modeling uncertainty diminishes over the design cycle, and cost models that consider the price of adaptability. This co-design framework is coupled with an online control strategy that performs the real-time adaptation, subject to constraints. To evaluate this approach, we focus on the segmented ultralight morphing rotor (SUMR) described in Noyes et al. (2020), Zalkind et al. (2017), Kianbakht et al. (2022), and Ananda et al. (2018). Applying the co-design framework to the SUMR, and utilizing the aforementioned performance surrogate model, the expected lifetime profit is shown to increase by 6.5% when the level of adaptability is optimized. Over a 24-hour dynamic simulation, the predictions based on the simplified surrogate model are shown to deviate by less than 1% relative to the results of the higher-fidelity dynamic model (which is suitable for 24-hour simulations but not for long-term profitability projections).
|
|
09:30-09:45, Paper WeAT3.2 | |
Parameter Discovery for Optimal Magnetopiezoelastic Energy Harvesters Using Neural Optimization Approach (I) |
|
Ayyad, Mahmoud | Stevens Institute of Technology |
Alqaleiby, Hossam | Stevens Institute of Technology |
Hajj, Muhammad | Stevens Institute of Technology |
Keywords: Motion and Vibration Control, Machine Learning in modeling, estimation, and control, Modeling and Validation
Abstract: Piezoelectric transduction of vibrational energy to electric power has attracted interest because of its potential to reduce the dependence on depletable batteries currently used to power micro sensors and devices. Assessment of variations in the output power based on varying the harvester's parameters may not yield an optimal design. For that purpose, we implement a neural optimization approach to optimize the performance of a magnetopiezoelastic energy harvester under specific constraints. The data set used in training the neural networks and optimization approach are generated using simulations of an experimentally validated numerical model. The results demonstrate the usefulness of this approach in the design optimization of piezoelectric energy harvesters.
|
|
09:45-10:00, Paper WeAT3.3 | |
Experimental Investigation of Using Vision-Based Wave Height Sensors and a Kalman Filter to Estimate Ocean Wave Field (I) |
|
Chen, Zihao | University of Minnesota |
Li, Perry Y. | Univ. of Minnesota |
Keywords: Marine Systems, Linear Control Systems, Sensors and Actuators
Abstract: The vision of this research is to use a small array of discrete low-cost wave height sensors to estimate the 2D wave field in the vicinity of the sensors. This will provide useful information for wave energy converters (WECs) within a wave farm to cooperate with each other to maximize total energy production. To this end, a Kalman Filter Observer has previously been developed to synthesize the sensor information and to estimate the temporal and spatial evolution of the wave field. In this paper, this concept is validated experimentally in a flume equipped with a wave maker. The tests utilize a simple wave height sensor consisting of a buoy and a camera connected to a single-board computer. Wave height information is obtained in real-time at a rate of 40 frames per second using a custom-designed vision-based localization algorithm. The information is then processed by the Kalman Filter Observer to estimate the wave profile. The estimated wave heights at locations away from the sensor are found to be accurate under various wave conditions. The experiments also provided information on how best to tune the parameters of the Kalman Filter Observer.
|
|
10:00-10:15, Paper WeAT3.4 | |
Model Predictive Control for Active Cooling of Photovoltaic Panels (I) |
|
Ouedraogo, Asmaou | Texas Tech University |
Docimo, Donald | Texas Tech University |
Keywords: Power and Energy Systems, Optimal Control, Control Applications
Abstract: This paper studies control of photovoltaic (PV) panels as related to energy efficiency and thermal dynamics. Solar-based electricity generation is instrumental to modern and future grid systems, supporting cleaner energy production and energy independence. However, PV power output is inversely correlated to the panel temperature, which is sensitive to environmental conditions such as ambient air temperature and wind velocity. To mitigate the impact of these variables, studies explore passive and active cooling to reduce PV temperature and increase efficiency. However, these emphasize design changes, neglecting the use of control to extract maximum value from the cooling system and PV. This paper introduces optimization-based control as a foundation for active cooling. After determining environmental conditions under which active cooling can increase PV power output, a novel cooling configuration using phase change material (PCM) and a thermal transducer is presented. A model predictive control (MPC) algorithm is developed to manage the cooling system and take advantage of forecasts. The cooling configuration and controller are validated using multiple scenarios, showing potential to reduce PV temperature by 20℃ and increase daily energy produced by 8%. An exploration into the ability of control to reduce the amount of PCM is presented, showing the tradeoff between daily energy production increases and cooling system size.
|
|
10:15-10:30, Paper WeAT3.5 | |
Multivariable Control Design for Load Reduction on Wind Turbines ⋆ |
|
Mohsin, Kazi | University of Central Florida |
Odeh, Mohammad | University of Central Florida |
Ngo, Tri | University of Central Florida |
Das, Tuhin | University of Central Florida |
Keywords: Control Design, Modeling and Validation, Motion and Vibration Control
Abstract: This article presents a multivariable control design to improve performance and reduce fatigue loads in wind turbines using torque control, collective pitch control (CPC), and individual pitch control (IPC). The Relative Gain Array (RGA) is used to quantify the level of interactions between inputs and outputs in the wind turbine control system. This analysis helps determine the control structure and input-output pairings, identifying inputs that have a dominant effect on each output. Various control structures are proposed, implemented, and evaluated in this study. Extensive simulations show that the RGA-based controller outperforms other controllers in reducing loads on the blade root bending moment, tower side-to-side, and tower fore-aft bending moments at the frequencies of interest, in comparison to other control structures. Moreover, it has no detrimental effects on the rotor speed and power generation, which are regulated by the CPC and torque controller. A Control-oriented, Reconfigurable, and Acausal Floating Turbine Simulator (CRAFTS), developed in-house, is used for the control design, implementation, and evaluation.
|
|
10:30-10:45, Paper WeAT3.6 | |
Fault-Tolerant Decentralized Control for Large-Scale Inverter-Based Resources for Active Power Tracking |
|
Vedula, Satish | Florida State University |
Olajube, Ayobami | Florida State University |
Anubi, Olugbenga | Florida State University |
Keywords: Power and Energy Systems, Control Design, Control Applications
Abstract: Integration of Inverter Based Resources (IBRs) which lack the intrinsic characteristics such as the inertial response of the traditional synchronous-generator (SG) based sources presents a new challenge in the form of analyzing the grid stability under their presence. While the dynamic composition of IBRs differs from that of the SGs, the control objective remains similar in terms of tracking the desired active power. This letter presents a decentralized primal-dual-based fault-tolerant control framework for the power allocation in IBRs. Overall, a hierarchical control algorithm is developed with a lower level addressing the current control and the parameter estimation for the IBRs and the higher level acting as the reference power generator to the low level based on the desired active power profile. The decentralized network-based algorithm adaptively splits the desired power between the IBRs taking into consideration the health of the IBRs transmission lines. The proposed framework is tested through a simulation on the network of IBRs and the high-level controller performance is compared against the existing framework in the literature. The proposed algorithm shows significant performance improvement in the magnitude of power deviation and settling time to the nominal value under faulty conditions as compared to the algorithm in the literature.
|
|
WeAT4 |
Prime 2 |
Motion Planning |
Regular Session |
Chair: Diaz-Mercado, Yancy | University of Maryland |
Co-Chair: Radisavljevic-Gajic, Verica | Ajman Univeristy |
|
09:15-09:30, Paper WeAT4.1 | |
An Approach to Realize Generalized Optimal Motion Primitives Using Physics Informed Neural Networks |
|
Slightam, Jonathon | Sandia National Laboratories |
Steyer, Andrew | Sandia National Laboratories |
Beaver, Logan | Old Dominion University |
Young, Carol | Sandia National Lab |
Keywords: Path Planning and Motion Control, Robotics, Machine Learning in modeling, estimation, and control
Abstract: Autonomous manipulation is a challenging problem in field robotics due to uncertainty in object properties, constraints, and coupling phenomenon to robot control systems. Humans learn motion primitives over time to effectively interact with the environment. We postulate that autonomous manipulation can be enabled by basic sets of motion primitives as well, but do not necessitate mimicking human motion primitives. This work presents an approach to generalized optimal motion primitives using physics-informed neural networks. Our simulated and experimental results demonstrate that optimality is notionally maintained through the generalization of our physics-informed machine learning approach which enables real-time adaptation of primitive motion profiles.
|
|
09:30-09:45, Paper WeAT4.2 | |
NMPC for Collision Avoidance by Superellipsoid Separation |
|
Moran, Ruairi | The Queen's University, Belfast |
Bagley, Sheila | Equipmentshare |
Kasmann, Seth | Equipment Share |
Martin, Rob | Equipmentshare |
Pasley, David | Equipment Share |
Trimble, Shane | EquipmentShare |
Dianics, James | Equipmentshare.com Inc |
Sopasakis, Pantelis | Queen’s University Belfast |
Keywords: Intelligent Autonomous Vehicles, Path Planning and Motion Control, Nonlinear Control Systems
Abstract: This paper introduces a novel NMPC formulation for real-time obstacle avoidance on heavy equipment by modeling both vehicle and obstacles as convex superellipsoids. The combination of this approach with the separating hyperplane theorem and Optimization Engine (OpEn) allows to achieve efficient obstacle avoidance in autonomous heavy equipment and robotics. We demonstrate the efficacy of the approach through simulated and experimental results, showcasing a skid-steer loader's capability to navigate in obstructed environments.
|
|
09:45-10:00, Paper WeAT4.3 | |
Avoiding Internal Gaps with Heterogeneous Circle Coverings Via Optimal Power Diagrams |
|
Frommer, Andrew | University of Maryland |
Diaz-Mercado, Yancy | University of Maryland |
Keywords: Multi-agent and Networked Systems, Nonlinear Control Systems
Abstract: In this work, we present a strategy for distributing a collection of heterogeneous circles over a convex domain such that there are no gaps between circles. We find optimal power diagram weights to partition the domain and repeatedly update the location of the circles to ensure there are no gaps between circles. Results presented demonstrate the algorithm's effectiveness and comparisons are provided with two other naive coverage algorithms. We show an improvement in coverage over naive Voronoi diagram coverage and demonstrate no internal gaps for feasible configurations.
|
|
10:00-10:15, Paper WeAT4.4 | |
Multi-Agent Motion Planning with Non-Uniform Sensors and Intermittent Q-Learning in Urban Air Mobility Scenarios |
|
Staatz, Dylan | Georgia Institute of Technology |
Vamvoudakis, Kyriakos G. | Georgia Tech |
Netter, Josh | Georgia Institute of Technology |
Keywords: Unmanned Ground and Aerial Vehicles, Cyber physical systems, Optimal Control
Abstract: In this paper, we develop a learning-based sampling motion planning framework for multi-agent systems with non-uniform sensor distributions. This is done by placing constraints on the search space for agents that do not have environmental sensing abilities of their own. Instead, these agents rely on a primary agent to have the sensors needed to be knowledgeable of the surrounding environment. Additionally, deep intermittent Q-learning is used to control the unknown system dynamics of the agents. This controller is intermittent to reduce the amount of communication from the primary agent to the other agents. An approach for linear agents in 2D and 3D is given as an example application with results from simulated environments.
|
|
10:15-10:30, Paper WeAT4.5 | |
Control Barrier Function Based Energy Optimal Obstacles Avoidance for Point-To-Point Maneuvers |
|
Kim, Youngjin | University at Buffalo |
He, Chaozhe | University at Buffalo |
Singh, Tarunraj | State Univ. of New York at Buffalo |
Keywords: Optimal Control, Unmanned Ground and Aerial Vehicles, Transportation Systems
Abstract: This paper focuses on developing a motion planning algorithm for static obstacle avoidance for a kinematic unicycle robot undergoing an energy-optimal point-to-point maneuver. The standard kinematic model is redefined in the geometric center space, motivated by the feedback linearization technique, resulting in a reduced order kinematic model. The proposed optimal motion planning approach is decomposed into two {sequential} stages: pre-planning and re-planning. In the pre-planning stage, an obstacle-free point-to-point optimal control problem is formulated and solved. Utilizing the solution from the optimal control problem, a perturbation controller is introduced which incorporates the nominal optimal control as a feedforward controller and a feedback tracking controller. In the second stage, the control barrier function method is employed to account for safety requirements, resulting in a minimum intervention control and solved in a point-wise optimization framework that accounts for the obstacles. {The safety constraints are used as a quantitative metric to trigger trajectory re-planning, ultimately resulting in a nearly optimal control and trajectory.
|
|
10:30-10:45, Paper WeAT4.6 | |
SODA-RRT: Safe Optimal Dynamics-AwareMotion Planning |
|
Niknejad, Nariman | Michigan State University |
Esmzad, Ramin | Michigan State University |
Modares, Hamidreza | Michigan State University |
Keywords: Path Planning and Motion Control, Optimal Control, Linear Control Systems
Abstract: This paper presents a performance-aware motion planning approach that generates collision-free paths with guaranteed performance using invariant sets. Specifically, the presented planner connected conflict-free invariant sets inside which closed-loop trajectories respect safety and optimality. Waypoints are randomly generated and invariant sets are formed for them and are connected to create a sequence of invariant sets from the initial to the target point. For each waypoint, an optimization problem finds the largest conflict-free zone and a safe-optimal controller. The presented algorithm, called safe optimal dynamics-aware motion planning SODA-RRT, accounts for the performance-reachability of connected waypoints, removing the need for frequent re-planning. Its effectiveness is demonstrated through spacecraft motion planning to avoid debris.
|
|
WeAT5 |
Streeterville E |
Process and Manufacturing Control |
Regular Session |
Chair: Chen, Dongmei | UT Austin |
Co-Chair: Devasia, Santosh | Univ of Washington |
|
09:15-09:30, Paper WeAT5.1 | |
Optimal Control of an R2R Dry Transfer Process with Bounded Dynamics Convexification |
|
Martin, Christopher | University of Texas at Austin |
Zhao, Qishen | University of Texas at Austin |
Bakshi, Soovadeep | University of Texas at Austin |
Velasquez, Enrique | The University of Texas at Austin |
Li, Wei | University of Texas at Austin |
Chen, Dongmei | UT Austin |
Keywords: Manufacturing Systems, Control Design, Mechatronic Systems
Abstract: For efficient roll-to-roll (R2R) production of flexible electronic components, a precise R2R transfer peeling process is essential, requiring accurate modeling and control. This paper introduces a novel approach to confining the dynamics of a nonlinear R2R mechanical peeling system within a convex set known as a norm-bounded linear differential inclusion (NLDI). This method utilizes constraints on uncertain system variables to create a tighter NLDI representation compared to other convexification techniques. Moreover, it offers drastically reduced computational cost compared to previous methods applied to convexify the R2R peeling system. The NLDI is employed to generate an H_∞-optimal controller for the R2R peeling system, and both simulations and experiments demonstrate better dynamic performance compared to other controllers for R2R transfer.
|
|
09:30-09:45, Paper WeAT5.2 | |
Modeling and Control of Strip Transport in Metal Peeling |
|
Yalamanchili, Aditya | Texas A&M University |
Sagapuram, Dinakar | Texas A&M University |
Pagilla, Prabhakar R. | Texas A&M University |
Keywords: Modeling and Validation, Control Design, Manufacturing Systems
Abstract: Metal peeling refers to the process of forming a thin metal strip from the surface of a rotating feedstock using controlled material removal -- machining under an applied strip tension. Conventional metal strip and sheet production relies on multi-stage hot and cold rolling. In this process, cast slabs are progressively reduced in thickness through numerous deformation passes, necessitating multiple intermediate heating, cooling, annealing, and finishing steps to produce strips of desired quality and properties. In contrast, metal peeling is a single-step process that offers an energy efficient alternative to traditional rolling and eliminates the need for preheating. This study presents a comprehensive approach to modeling and controlling the transport behavior of material from the peeling edge to a coiler, considering the relationship between strip thickness and tension. The mechanics of strip formation process is described, while emphasizing the role of strip tension in ensuring uniformity and quality of the peeled strip. This includes an analysis of the deformation history in the peeling zone and the transport dynamics of the strip. Using conservation laws, governing equations for strip tension and velocity that incorporate dynamic spatiotemporal interactions between peeling and transport processes are developed. A cascaded control approach for regulating strip velocity and tension is developed employing feedforward equilibrium control combined with feedback control. The feedforward control component, derived from the governing equations, is designed to maintain the system at the ideal forced equilibrium, while feedback control provides real-time adjustments to correct any deviations from the forced equilibrium state and provide robustness to handle real-world variations and disturbances. Peeling experiments are performed with steel to evaluate the proposed control approach. Comparisons between two control strategies, with and without tension feedback, are presented and discussed. The importance of real-time tension control for maintaining desired strip tension, mitigating strip thickness variations and improving other dimensional features of the strip is also briefly discussed.
|
|
09:45-10:00, Paper WeAT5.3 | |
Multiresolution Dynamic Mode Decomposition Based Modeling of Wastewater Treatment Process |
|
Prakash, Om | University of Alberta |
Huang, Biao | Univ. of Alberta |
Keywords: Modeling and Validation, Modelling, Identification and Signal Processing
Abstract: The wastewater treatment process (WWTP) has now become an increasingly important process in order to accommodate sustainable goals and mitigate the water crisis. Modeling of WWTP is crucial to performing estimation, control, and optimization. Since WWTP displays large-scale, multiscale behavior, a data-driven multiresolution dynamic mode decomposition (MrDMD) based approach fits well to uncover spatiotemporal structure at high resolution. In the current work, we propose incorporating exogenous variables (control) in the MrDMD framework. Further, we illustrate how slow mode reconstructions at different decomposition levels uniquely capture the process's multiscale behavior. Additionally, we demonstrate the efficacy of the proposed approach on benchmark WWTP. We also compare its performance with the dynamic mode decomposition-based model.
|
|
10:00-10:15, Paper WeAT5.4 | |
Digital Twin Design and Cross Process Model Transfer for Additive Manufacturing |
|
Lai, Frank Y. | University of Michigan |
Shen, Eric | University of Michigan |
Fonda, James | Boeing |
Salour, Al | Boeing |
Tilbury, Dawn M. | Univ of Michigan |
Barton, Kira | University of Michigan |
Keywords: Modelling, Identification and Signal Processing, Machine Learning in modeling, estimation, and control, Manufacturing Systems
Abstract: Digital twins serve as powerful tools for monitoring the status and anticipating errors within manufacturing processes. However, the creation of these digital twins entails a significant investment in time and resources. This paper delves into the process of crafting digital twins for material extrusion additive manufacturing. Furthermore, it investigates the application of cross model transfer learning to such digital twins, and assesses the feasibility in transferring digital twins rather than building them from the ground up.
|
|
10:15-10:30, Paper WeAT5.5 | |
Constrained State Estimation Using Lie Theory: Application to Directional Drilling |
|
Karvinen, Kai | Baker Hughes INTEQ GmbH |
Keywords: Estimation
Abstract: Downhole estimation of inclination, azimuth, toolface angle, and rotational speed is vital to ensure precise steering and wellbore monitoring in directional drilling applications. This paper outlines the application of Lie theory to the real-time estimation of downhole tool orientation and demonstrates how both fault-tolerant operation and bias estimation can be achieved. Furthermore, real-time estimation of the measurement variances allows the proposed algorithm to adapt to the constantly changing downhole conditions. Testing with field data shows that the proposed algorithm can outperform current state-of-the-art directional algorithms, which rely strictly on conventional signal processing techniques.
|
|
10:30-10:45, Paper WeAT5.6 | |
Active Data-Enabled Robot Learning of Elastic Workpiece Interactions |
|
McCann, Lance | University of Washington |
Yan, Leon (Liangwu) | University of Washington |
Hassan, Sarmad | University of Washington |
Garbini, Joseph | University of Washington |
Devasia, Santosh | Univ of Washington |
Keywords: Machine Learning in modeling, estimation, and control, Manufacturing Systems, Robotics
Abstract: During manufacturing processes, such as clamping and drilling of elastic structures, it is essential to maintain tool-workpiece normality to minimize shear forces and torques, and thereby preventing damage to the tool or the workpiece. The challenge arises in making precise model-based predictions of the relatively large deformations that occur as the applied normal force (e.g., clamping force) is increased. However, precision deformation predictions are essential for selecting the optimal robot pose that maintains force normality. Therefore, recent works have employed force displacement measurements at each work location to determine the robot pose for maintaining tool normality. Nevertheless, this approach, which relies on local measurements at each work location and at each gradual increment of the applied normal force, can be slow and consequently, time prohibitive. The main contributions of this work are: (i) to use Gaussian Process methods to learn the robot-pose map for force normality at unmeasured workpiece locations; and (ii) to use active learning to optimally select and minimize the number of measurement locations needed for accurate learning of the robot-pose map. Experimental results show that the number of data points needed with active learning is 77.8% less than the case with a benchmark linear positioning learning for the same level of model precision. Additionally, the learned robot-pose map enables a rapid increase of the normal force at unmeasured locations on the workpiece, reaching force-increment rates up to eight times faster than the original force-increment rate when the robot is learning the correct pose.
|
|
WeAT6 |
Streeterville W |
Thermal Process Modeling and Control |
Regular Session |
Chair: Chen, Pingen | Tennessee Technological University |
Co-Chair: Nazari, Shima | UC Davis |
|
09:15-09:30, Paper WeAT6.1 | |
Control of Combustion Phasing Using Accelerometer-Based Non-Intrusive Sensing |
|
Govind Raju, Sathya Aswath | University of Minnesota - Twin Cities |
Reisetter, Mitchell | University of Wisconsin-Madison |
Miganakallu, Niranjan | Southwest Research Institute |
Stafford, Jacob | University of Wisconsin - Madison |
Sun, Zongxuan | University of Minnesota |
Rothamer, David | University of Wisconsin-Madison |
Kim, Kenneth | DEVCOM Army Research Laboratory |
Kweon, Chol-Bum | DEVCOM Army Research Laboratory |
Keywords: Automotive Systems, Control Applications, Modelling, Identification and Signal Processing
Abstract: Measuring the combustion phasing of an engine using in-cylinder pressure sensors is well established. However, pressure sensors need to be directly exposed to the in-cylinder environment, requiring changes to the cylinder head. Several methods have been proposed for sensing combustion phasing non-intrusively by mounting an accelerometer on the engine block. This paper presents real-time control of combustion phasing in a compression-ignition engine using non-intrusive accelerometer-based sensing during a dynamic fuel switch. A systematic data-driven control framework capable of handling fuel switching in real-time is used. The control is designed based on the pressure data, and the real-time implementation is performed using the accelerometer signal. Results from combustion phasing tracking experiments performed on a compression-ignition engine are presented.
|
|
09:30-09:45, Paper WeAT6.2 | |
LiFePO4 Battery Thermal Modeling: Bus Bar Thermal Effects |
|
Haas, Meridian | UC Davis |
Nemati, Alireza | University of California-Davis |
Moura, Scott | UC Berkeley |
Nazari, Shima | UC Davis |
Keywords: Power and Energy Systems, Modeling and Validation, Modelling, Identification and Signal Processing
Abstract: Lithium iron phosphate (LFP) batteries are ideal for electrification of off-road heavy-duty vehicles with less concerns on the system weight. However, the limited battery life aggravated by the long working hours is a primary concern for some off-road applications such as construction equipment. Temperature is one of the main influencing factors in battery aging. Therefore, accurate prediction of temperature dynamics with fast lumped parameter models is essential and can be used for long-term analysis. This paper introduces a thermal model for pack of cells, with each cell represented by surface and core temperature states. We derived model parameters from experimental thermal cycling data, emphasizing the significance of reversible entropic heat generation for capturing faster dynamics. Furthermore, our work highlights the errors introduced by neglecting the bus bar thermal effects when extending a single-cell model to a cell pack. Our proposed solution incorporates the conduction between cell cores via the bus bar and accounts for heat dissipation through convection from the bus bar to surrounding air.
|
|
09:45-10:00, Paper WeAT6.3 | |
Optimizing Cooling Load in a Central Chiller Plant: A Data-Driven Approach |
|
Souza, Diego | The Ohio State University |
Stockar, Stephanie | The Ohio State University |
Keywords: Machine Learning in modeling, estimation, and control, Modeling and Validation, Power and Energy Systems
Abstract: This work presents an advanced optimization strategy for the operation of the individual chillers in a plant that reduces the power demand while meeting the cooling load. This is achieved by first developing a hybrid model combining energy-based and data-driven methods to describe the energy demand of a central chiller plant for a given cooling load and environmental conditions. The model is calibrated and validated on The Ohio State University operation data. The validated model is then used in an optimization algorithm based on particle swarm optimization to find the optimal load of each chiller for different weather and operation conditions. The optimized strategy is tested in simulation considering one year of operation in Central Ohio and compared against the baseline strategy. The new strategy achieves, on average, a 4% reduction in daily peak power consumption for four mild weather months in the year, with instances of up to 12% reduction.
|
|
10:00-10:15, Paper WeAT6.4 | |
Modeling Thermal-Mechanical Dynamics in an Electrothermally-Actuated Micro-Origami Systems with Large Deflection |
|
Yang, Yiwei | University of Michigan - Ann Arbor |
Zhu, Yi | University of Michigan - Ann Arbor |
Ghrayeb, Anan | University of Michigan - Ann Arbor |
Yu, Joonyoung | University of Michigan |
Filipov, Evgueni | University of Michigan - Ann Arbor |
Oldham, Kenn | University of Michigan |
Keywords: Modeling and Validation, Electromechanical systems, Robotics
Abstract: While the electrothermal actuators are popular among micro-scale devices, the heat transfer process in electrothermally-actuated micro-origami systems can be geometry-dependent due to the large deflections that occur with folding. This paper proposes a thermal model based on thermal conduction for an electrothermally-actuated micro-origami system with structure-dependent thermal resistances and capacitances. The model is compared to experimental results for both static and step responses. After calibration, comparison in the static response suggests that the model is able to capture non-linear behaviors due to large folding of the micro-origami system. Several additional nonlinear behaviors are observed from transient step dynamics that are hypothesized to arise from temperature/strain rate-dependent material nonlinearities. After calibration of material properties and introduction of additional residual stress that is related to the heating rate, the model can closely match the experimental step response behavior. This model may help to control the transient folding process of future micro-origami systems.
|
|
10:15-10:30, Paper WeAT6.5 | |
Modeling and Experimental Validation of Degreened and Aged Diesel Selective Catalytic Reduction Systems |
|
Joshi, Sachin | Tennessee Tech University |
Kocher, Lyle | Purdue Univeristy |
Schmidt, David | Cummins Inc |
Chen, Pingen | Tennessee Technological University |
Keywords: Automotive Systems, Modeling and Validation
Abstract: Selective Catalytic Reduction (SCR) and Ammonia Slip Catalysts (ASC) are important components in Diesel aftertreatment systems for reducing tailpipe NOx emissions while minimizing tailpipe ammonia (NH3) slip. A significant challenge faced by Diesel SCR system is the degradation or aging of the catalyst over time, which leads to an increase in tailpipe NOx and NH3 emissions. Therefore, an accurate mathematical model is needed to differentiate a fresh (degreened) catalyst and an aged catalyst for future diagnosis. However, modeling such a combined system (SCR+ASC) can be challenging, due to complex chemical reactions and system dynamics in real-world operation. Since oxidation of the NH3 can only occur at higher temperatures (>220℃), effects of the ASC can be neglected. Using this assumption, this paper proposes single-cell and two-cell diagnostic-oriented SCR system models under lower temperature operating conditions. The aged end-of-life (EUL) SCR models incorporates aging factor(s) to the SCR’s storage capacity to capture the impacts of aging on SCR performance. The proposed SCR models are calibrated using the experimental data from degreened and aged catalysts. The model validation results demonstrate that high accuracy of the proposed SCR models in predicting the behaviors of fresh and aged SCR systems in FTP cycles. Furthermore, this paper shows difference between single-cell SCR model and two-cell SCR model capability in terms of accuracy and reliability.
|
|
10:30-10:45, Paper WeAT6.6 | |
Using Flexibility Setback in the Demand-Side Control of District Heating |
|
Blizard, Audrey | The Ohio State University |
Stockar, Stephanie | The Ohio State University |
Keywords: Control of Smart Buildings and Microgrids, Multi-agent and Networked Systems, Power and Energy Systems
Abstract: This paper examines the effect time-varying flexibility has on the demand-side performance of a district heating network. A hierarchical control framework is used to control heat supplied to users, and the setback flexibility is considered via time-varying flexibility envelopes. The results are presented for an 18-user district heating network, with realistic network parameters and four occupancy profiles for the connected buildings: residential, commercial, retail, and medical. Overall, the time-varying flexibility method results in a 3.8% decrease in total losses as compared to the nominal case.
|
|
WeP2L |
Avenue Ballroom E/W |
Plenary Talk: The Central Role of Modeling, Estimation and Control in
Sustainable, Connected and Automated Mobility – Some Reflections on 30
Years of Innovation by Our Community, and Some Thoughts on the Future |
Plenary Session |
Chair: Onori, Simona | Stanford University |
Co-Chair: Fathy, Hosam K. | University of Maryland |
|
11:00-12:00, Paper WeP2L.1 | |
The Central Role of Modeling, Estimation and Control in Sustainable, Connected and Automated Mobility – Some Reflections on 30 Years of Innovation by Our Community, and Some Thoughts on the Future |
|
Rizzoni, Giorgio | Ohio State Univ |
|
WeCT3 |
Prime 1 |
Batteries and Fuel Cells |
Regular Session |
Chair: Shahbakhti, Mahdi | University of Alberta |
Co-Chair: Soudbakhsh, Damoon | Temple University |
|
13:30-13:45, Paper WeCT3.1 | |
Onboard Health Estimation Using Distribution of Relaxation Times for Lithium-Ion Batteries |
|
Khan, Muhammad Aadil | Stanford University |
Thatipamula, Venkata Saicharan | Stanford University |
Onori, Simona | Stanford University |
Keywords: Modeling and Validation, Machine Learning in modeling, estimation, and control
Abstract: Real-life batteries tend to experience a range of operating conditions, and undergo degradation due to a combination of both calendar and cycling aging. Onboard health estimation models typically use cycling aging data only, and account for at most one operating condition e.g., temperature, which can limit the accuracy of the models for state-of-health (SOH) estimation. In this paper, we utilize electrochemical impedance spectroscopy (EIS) data from 5 calendar-aged and 17 cycling-aged cells to perform SOH estimation under various operating conditions. The EIS curves are deconvoluted using the distribution of relaxation times (DRT) technique to map them onto a function g which consists of distinct timescales representing different resistances inside the cell. These DRT curves, g, are then used as inputs to a long short-term memory (LSTM)-based neural network model for SOH estimation. We validate the model performance by testing it on ten different test sets, and achieve an average RMSPE of 1.69% across these sets.
|
|
13:45-14:00, Paper WeCT3.2 | |
All-Solid-State Lithium Sulfur Cell Safety Characterization |
|
Cleary, Timothy | Pennsylvania State University |
Wang, Daiwei | Pennsylvania State University |
Wang, Donghai | Pennsylvania State University |
Fathy, Hosam K. | University of Maryland |
Rahn, Christopher D. | Penn State Univ |
Keywords: Electromechanical systems
Abstract: Motivated by high theoretical specific capacity and inherent safety, this work investigates the performance of all-solid-state lithium-sulfur batteries (ASLBs) across increasing applied pressure and under electrical and mechanical abuse conditions. An all-PEEK (polyether ether ketone) split-cell holder with embedded temperature sensors was developed to provide a near-adiabatic test environment for measuring anode and cathode thermal response. Cells were fabricated, mechanically loaded, and electrically characterized, and then test-to-failure conditions were explored, including overcharge, short circuit, and nail puncture. The laboratory-fabricated cell and newly developed split-cell holder with a sulfur loading of 3 g/cm^2 demonstrated areal capacities greater than 3 mAh/cm^2 at current densities of 0.5 mA/cm^2, validating the cell holder's ability to operate across a range of applied pressures up to and including 60MPa. Increasing applied pressure showed recovery of lost cyclic capacity during constant pressure testing. Also, minimum anode/cathode thermal responses under test-to-failure conditions were measured, supporting a promising early look at the potential of this battery cell for use in safety-critical environments with low cycle life and high energy density requirements.
|
|
14:00-14:15, Paper WeCT3.3 | |
Short Circuit Estimation in Lithium-Ion Batteries Using Moving Horizon Estimation |
|
Moon, Jihoon | The Pennsylvania State University |
Bhaskar, Kiran | The Pennsylvania State University |
Rahn, Christopher D. | Penn State Univ |
Keywords: Control Applications, Power and Energy Systems, Estimation
Abstract: This paper proposes rapid and accurate short circuit estimation under resting condition using joint Moving Horizon Estimation (MHE). The use of lithium-ion batteries (LiBs) in electric vehicles (EVs) has been increasing, leading to heightened concerns regarding the safety of LiBs. Detecting a short circuit, which is a major cause of safety incidents, is challenging when it is in its early stages. Therefore, short circuits should be detected swiftly and accurately to prevent thermal runaway and potential fires, property damage, injuries and mortalities. During leak testing of new cells or often an EV crash, applied current may be zero and parameters unknown. The presented work addresses these challenges through the application of a joint MHE approach, to estimate both short circuit current and battery capacity. The proposed approach is evaluated through extensive simulations involving various short circuit scenarios and is compared to a Dual Extended Kalman Filter (EKF) and Dual Unscented Kalman Filter (UKF). Experimental data is also used to validate the effectiveness in of states and parameters estimation.
|
|
14:15-14:30, Paper WeCT3.4 | |
SOC-Dependency of the Time Constants and Polarizations of Li-Ion Batteries |
|
Derakhshan, Mohsen | Temple University |
Soudbakhsh, Damoon | Temple University |
Keywords: Power and Energy Systems, Modeling and Validation, Machine Learning in modeling, estimation, and control
Abstract: This paper presents an analysis of the time constants of Li-ion batteries. Our experiments were conducted on batteries with positive electrode material of NMC811. The time constants were determined using a distribution function of relaxation times (DRT) from the impedance spectra. Batteries' internal processes can be characterized using their time constants and associated polarizations. Therefore, they allow for the evaluation of Li-ion batteries for safety and performance, as well as other energy storage technologies. In this study, we investigated the effect of temperature and State of Charge (SOC) on the time constants and polarizations of eight cylindrical cells. After the initial cycling of the cells, their EIS (Electrochemical Impedance Spectroscopy) data were collected at different temperatures from -20 to +60 degreesC and SOCs ranging from 100% to 0%. The EIS data were processed to determine the time constants. We identified four dominant peaks in the medium to low-frequency range, assigned to contact resistance, Solid Electrolyte Interphase (SEI), charge transfer (CT), and diffusion. Additionally, two dominant peaks were observed in the high-frequency range. Next, we studied the SOC-dependency of the polarizations and representative time constants of the processes. The representative time constants were defined as the local maxima. The charge transfer kinetics (CT) and diffusion processes showed strong SOC dependencies. For example, the time constant of CT dropped from 1s at 0% SOC to 7ms at 50% SOC and increased to about 65ms at 100% SOC. During this cycle, its polarization changed from 27 mOhm to 0.3 mOhm and 4.9 mOhm, respectively. In contrast, the time constant and polarization of the high-frequency processes showed very small variations with SOC levels.
|
|
14:30-14:45, Paper WeCT3.5 | |
Performance Prediction of a Range of Diverse Solid Oxide Fuel Cells Using Deep Learning and Principal Component Analysis |
|
Salehi, Zeynab | University of Alberta |
Mohamadali, Tofigh | University of Alberta |
Vafaeenezhad, Sajad | University of Alberta |
Ali, Kharazmi | Cummins Inc |
Smith, Daniel J. | Cummins |
Koch, Charles Robert | University of Alberta |
Shahbakhti, Mahdi | University of Alberta |
Keywords: Machine Learning in modeling, estimation, and control, Modeling and Validation, Power and Energy Systems
Abstract: Solid oxide fuel cell (SOFC) is a common fuel cell type that has high efficiency. SOFC is a complex non-linear system, subject to aging and manufacturing variation, which makes performance prediction difficult. To estimate the SOFC performance data-driven methods allow a trade-off between computation cost and accuracy. Eight different SOFC tubular cells with different properties are fabricated and experimentally tested in 18 different operating conditions. A deep neural network (DNN) is used to predict the output voltage of the cells. The input features of this network are the cell physical properties determined by scanning electron microscope (SEM) analysis and the operating parameters. As a first step, all measurable features are provided to principal component analysis (PCA) for feature selection resulting in a 50% reduction in features, resulting in a corresponding 50% reduction in the training time of the DNN. This trained DNN is able to capture the non-linear voltage drop of concentration polarization in the current density-voltage (J-V) curves. The prediction performance of the network is evaluated using three performance metrics of coefficient of determination (R2), root mean square error (RMSE), and mean absolute percentage error (MAPE) with satisfactory accuracy for both the training and test datasets. For all predictions, R2 is 0.99, MAPE is less than 1%, and RMSE is 0.0001 on the test dataset. During the training process, both the validation loss and training loss approach zero indicating that the trained model is not over- fitted. The DNN model can be useful for design and operation optimization purposes.
|
|
14:45-15:00, Paper WeCT3.6 | |
Lithium-Ion Battery State of Charge Estimation with a Multi-Objective L1/H2 Robust Observer |
|
Shen, Heran | The University of Texas at Austin |
Zhou, Xingyu | University of Texas at Austin |
Ahn, Hyunjin | The University of Texas at Austin |
Kung, Yung-Chi | The University of Texas at Austin |
Wang, Junmin | University of Texas at Austin |
Keywords: Estimation, Power and Energy Systems, Control Design
Abstract: The state of charge is a vital signal for the battery management system in lithium-ion battery-powered electric vehicles. This paper presents a novel method to estimate this signal by a multi-objective robust observer. Our approach prioritizes two key objectives: first, it attenuates the L_1 induced norm from modeling errors and disturbances to the estimation error; secondly, it alleviates the impact of sensor noise through H_2 optimization. The synthesis of this robust observer is framed as a bilinear-matrix-inequality optimization problem. Validation of the proposed method is conducted using experimental data. Further, its performance is compared against a benchmark algorithm to demonstrate its superiority.
|
|
WeCT4 |
Prime 2 |
Robotic, Mechatronic, and Autonomous Vehicle Systems |
Regular Session |
Co-Chair: Ahuja, Nitisha | The Pennsylvania State University |
|
13:30-13:45, Paper WeCT4.1 | |
Mechanics and Control of a Freely Rolling Two-Link Robot with Joint Actuation |
|
Moradi, Hamidreza | Department of Mechanical Engineering and Engineering Science, Un |
Wolek, Artur | University of North Carolina at Charlotte |
Kelly, Scott David | University of North Carolina at Charlotte |
Keywords: Nonlinear Control Systems
Abstract: We model the mechanics of a planar mobile robot comprising a pair of rigid links centered upon wheels that are free to roll and to pivot on the ground but not to slip sideways. The links are joined by a hinge to which an internal torque is applied as a control input. We show that the dynamics governing the robot's position and orientation in its environment can be reduced to the dynamics of a scalar nonholonomic momentum coupled to the dynamics of the joint angle. We also develop a feedback control law whereby the robot, given an initial nonzero rolling speed, can steer itself to a sequence of target points while resisting a natural tendency to enter a singular configuration.
|
|
13:45-14:00, Paper WeCT4.2 | |
On Multi-Fidelity Impedance Tuning for Human-Robot Cooperative Manipulation |
|
Lau, Ethan | Michigan State University |
Srivastava, Vaibhav | Michigan State University |
Bopardikar, Shaunak D. | Michigan State University |
Keywords: Human-Machine and Human-Robot Systems, Machine Learning in modeling, estimation, and control, Robotics
Abstract: We examine how a human-robot interaction (HRI) system may be designed when input-output data from previous experiments are available. Our objective is to learn an optimal impedance in the assistance design for a cooperative manipulation task with a new operator. Due to the variability between individuals, the design parameters that best suit one operator of the robot may not be the best parameters for another one. However, by incorporating historical data using a linear auto-regressive (AR-1) Gaussian process, the search for a new operator's optimal parameters can be accelerated. We lay out a framework for optimizing the human-robot cooperative manipulation that only requires input-output data. We characterize the learning performance using a notion called regret, establish how the AR-1 model improves the bound on the regret, and numerically illustrate this improvement in the context of a human-robot cooperative manipulation task. Further, we show how our approach's input-output nature provides robustness against modeling error through an additional numerical study.
|
|
14:00-14:15, Paper WeCT4.3 | |
Bayesian Statistical Method for Real-Time Tool Yield Calibration of Mud Motor |
|
Liu, Yang | Halliburton |
Xu, Shichao | Halliburton |
Pho, Vy | Halliburton |
Demirer, Nazli | Halliburton |
Bhaidasna, Ketan | Halliburton |
Keywords: Automotive Systems, Modelling, Identification and Signal Processing, Estimation
Abstract: The steerability of a mud motor in directional drilling is achieved by alternating between the ”rotate” and ”slide” modes. Consequently, the real-time directional data of mud motor exhibits a segmented structure, reflecting the distinct modes of operation during sliding and rotating phases. In this paper, a novel calibration method of steerability (tool yield) for mud motor, leveraging Bayesian statistical methods, specifically Reversible Jump Markov Chain Monte Carlo (RJMCMC) is proposed. The proposed method can precisely identify segment boundaries and execute segment-specific regression to capture the dynamic behavior within each mode, contributing to a precise and effective calibration process for enhanced steerability in mud motor applications.
|
|
14:15-14:30, Paper WeCT4.4 | |
A Comparison of Multi-Objective Servo Controller Synthesis Methodologies in Dual-Stage Hard Disk Drives |
|
Lehner, Erik | University of Minnesota |
Caverly, Ryan James | University of Minnesota |
Huang, Bin | Seagate Technology |
Sosseh, Raye | Seagate Technology |
Keywords: Control Design, Mechatronic Systems, Optimal Control
Abstract: This work provides a comprehensive comparison of four mixed H2/H-infinity-optimal control synthesis methodologies for dual-stage hard disk drives (HDD). Combinations of the commonly-used H2 and H-infinity objective functions and constraining metrics on the voice-coil motor (VCM), piezoelectric (PZT) actuator, and the position error signal (PES) drive the comparative analysis to better understand the control tradeoffs involved in their choice. The controllers are given identical design inputs and are synthesized using linear-matrix-inequality-based optimization. A simulated disturbance environment is used to analyze the closed-loop frequency response, and demonstrate the performance and benefits of each control method. This work demonstrates where each of the constraint metrics excel, which serves as a basis for tailored design based on the need of the HDD control engineer.
|
|
14:30-14:45, Paper WeCT4.5 | |
Constrained Turret Defense with Fixed Final Time |
|
Von Moll, Alexander | Air Force Research Laboratory |
Gerlach, Adam | United States Air Force |
Bakker, Craig | Pacific Northwest National Laboratory |
Rupe, Adam | Pacific Northwest National Laboratory |
Pachter, Meir | AFIT/ENG |
Keywords: Optimal Control, Aerospace, Intelligent Autonomous Vehicles
Abstract: In this paper, we extend existing turret defense differential game formulations involving a turn-constrained turret and mobile agent to include specified final time and a constraint. For the purposes of this analysis, the specified final time may represent some exogenous input, perhaps representing the time at which some other event will take place. As for the constraint, it represents a keep-out zone for the mobile agent. The scenario is formulated as a two-player, zero-sum differential game and solved via the method of characteristics (i.e., back-propagation of equilibrium trajectories). Three different trajectory types make up the solution: trajectories that end with the turret aligned with the mobile agent, trajectories that end with the mobile agent on the constraint boundary, and regular trajectories.
|
|
14:45-15:00, Paper WeCT4.6 | |
Real-Time Implementation of Differentiable Predictive Control on Embedded Microcontroller Hardware: A Case Study |
|
Boldocky, Jan | Slovak University of Technology in Bratislava |
Gulan, Martin | Slovak University of Technology in Bratislava |
Vrabie, Draguna | Pacific Northwest National Laboratory |
Drgona, Jan | Pacific Northwest National Laboratory |
Keywords: Machine Learning in modeling, estimation, and control, Control Design, Control Applications
Abstract: This paper presents the embedded implementation of differentiable predictive control (DPC) in a real-time control application with fast dynamics. DPC is a model-based policy optimization method that utilizes automatic differentiation to learn explicit control policies. The laboratory device features an aeropendulum mechatronic system equipped with a low-cost 32-bit microcontroller development board. We aimed to explore the possibilities of utilizing the DPC with linear and nonlinear dynamics models on lean hardware in a setpoint tracking application while satisfying the input and state constraints. The resulting control performance, memory footprint, and execution time are evaluated for different lengths of the prediction horizon.
|
|
WeCT5 |
Streeterville E/W |
Advanced Optimization, Control, and Planning for Dynamic Systems |
Regular Session |
Chair: Bevly, David M. | Auburn University |
Co-Chair: Diagne, Mamadou | University of California San Diego |
|
13:30-13:33, Paper WeCT5.1 | |
Adversarially Robust Pursuit-Evasion Games with Asymmetric and Incomplete Information |
|
Athalye, Surabhi | Georgia Institute of Technology |
Vamvoudakis, Kyriakos G. | Georgia Tech |
Keywords: Optimal Control, Linear Control Systems
Abstract: In this paper, we consider the problem of a two-player linear pursuit-evasion game with incomplete and asymmetric information. Particularly, we dispense with the assumption that either player is aware of their rival's decision-making scheme. We also introduce asymmetry to the game, whereby one player has more knowledge of the operating environment than the other and can utilize its opponent's full state to design its control action. We develop two separate customized receding horizon control-based differential games for the pursuer and evader such that the derived policies are optimal and robust against the asymmetric game scenario. This formulation obviates the need for an equal extent of knowledge among the agents, thereby enabling the disadvantaged player to achieve its objective despite the information discrepancy. The proposed method is implemented on simulated vehicles.
|
|
13:33-13:36, Paper WeCT5.2 | |
Output Feedback Control of Suspended Sediment Load Entrainment in Water Canals and Reservoirs |
|
Somathilake, Eranda | Department of Mechanical and Aerospace Engineering, University O |
Diagne, Mamadou | University of California San Diego |
Keywords: Modelling and Control of Environmental Systems, Distributed Parameter Systems, Estimation
Abstract: This paper addresses the management of water flow in a rectangular open channel, considering the dynamic nature of both the channel's bed-load sediment and the suspended sediment particles caused by entrainment and deposition effects. The control-oriented model under study is a set of coupled nonlinear partial differential equations (PDEs) describing conservation of mass and momentum while accounting for constitutive relations that govern sediment erosion and deposition phenomena. The proposed boundary control problem presents a fresh perspective in water canal management and expands Saint-Venant Exner (SVE) control frameworks by integrating dynamics related to the transport of fine particles. After linearization, PDE backstepping design is employed to stabilize both the bed-load sediment, the water dynamics together with the concentration of suspended sediment particles. Two underflow sluice gates are used for flow control at the upstream and downstream boundaries with only the downstream component being actuated. An observer-based backstepping control design is carried out for the downstream gate using state measurement at the upstream gate to globally exponentially stabilize the linearized system to a desired equilibrium point in mathscr{L}^2 sense. The stability analysis is performed on the linearized model which is a system of four coupled PDEs, three of which are rightward convecting and one leftward. The proposed control design has the potential to facilitate efficient reservoir flushing operations. Consistent simulation results are presented to illustrate the feasibility of the designed control law.
|
|
13:36-13:39, Paper WeCT5.3 | |
A Hierarchical Asymmetric Nash Bargaining Game for Load Management through Optimized EV Charging and Discharging |
|
Hu, Miaomiao | University of Florida |
Kushwaha, Dhruv | University of Florida |
Abdollahi Biron, Zoleikha | University of Florida |
Keywords: Control of Smart Buildings and Microgrids, Power and Energy Systems, Transportation Systems
Abstract: The large capacity of Electric Vehicles (EVs) battery energy storage and their flexible scheduling capabilities offer a dynamic virtual load that can be strategically managed to benefit both the power system and EV owners. By incentivizing EVs to charge during off-peak hours and discharge during peak hours, effective load management and enhanced grid stability in the microgrid can be achieved. This paper aims to develop a control strategy for managing EV charging and discharging at Charging Station (CS) to optimize the load demand profile and enhance grid stability. This paper proposes an Asymmetric Nash Bargaining (ANB) game with overload penalties aimed at facilitating energy trading between CS and EVs. Through the implementation of an adaptable bargaining power method, an optimal load profile is maintained for the CS while ensuring a fair distribution of benefits between all participants. The proposed ANB game achieves a significant 15.97% reduction in Mean Square Error (MSE) and a 3.29% increase in benefits compared to the classic Nash Bargaining (NB) game.
|
|
13:39-13:42, Paper WeCT5.4 | |
Instabilities and Pattern Formation in Epidemic Spread Induced by Nonlinear Saturation Effects and Ornstein-Uhlenbeck Noise |
|
Singh, Aman Kumar | Vellore Institute of Technology |
Buschmeyer, Cole | University of Dayton |
Ramakrishnan, Subramanian | University of Dayton |
Kumar, Manish | University of Cincinnati |
Keywords: Stochastic Systems, Modeling and Validation, Nonlinear Control Systems
Abstract: We analytically study the emergence of instabilities and the consequent steady state pattern formation in a stochastic partial differential equation (PDE) based, compartmental modelof spatiotemporal epidemic spread. The model is characterized by: (1) strongly nonlinear forces representing the infection transmission mechanism, and (2) random environmental forces represented by the Ornstein-Uhlenbeck (O-U) stochastic process which better approximates real-world uncertainties. Employing second-order perturbation analysis and computing the local Lyapunov exponent, we find the emergence of diffusion-induced instabilities and analyze the effects of O-U noise on these instabilities. We obtain a range of values of the diffusion coefficient and correlation time in parameter space that support the onset of instabilities. Notably, the stability and pattern formation results depend critically on the correlation time of the O-U stochastic process; specifically, we obtain lower values of steady-state infection density for higher correlation times. Also, for lower correlation times the results approach those obtained in the white noise case. The analytical results are valid for lower-order correlation times. In summary, the results provide insights into the onset of noise-induced, and Turing-type instabilities in a stochastic PDE epidemic model in the presence of strongly nonlinear deterministic infection forces and stochastic environmental forces represented by Ornstein-Uhlenbeck noise.
|
|
13:42-13:45, Paper WeCT5.5 | |
Determining Critical Vehicle Connectivity in Connected Autonomous Vehicles Using Information Theory |
|
Ramlall, Poorendra | Embry-Riddle Aeronautical University |
Roy, Subhradeep | Embry-Riddle Aeronautical University |
Keywords: Intelligent Autonomous Vehicles, Modeling and Validation, Stochastic Systems
Abstract: The idea of connected autonomous vehicles, which can share information among themselves, offers the potential to enhance traffic efficiency. However, putting this technology into practice comes with challenges. Real-world challenges such as data throughput limitations can make it hard for vehicles to share information smoothly. Consequently, it becomes crucial to identify critical vehicle connectivity, which specifies the minimum number of connected vehicles required to maintain stable traffic flow. This paper proposes an information-theoretic metric that uses information flow among connected vehicles to identify critical vehicle connectivity. The model-free nature of information-theoretic tools eliminates the need for closed-form expressions of the model, which are necessary for stability analysis methods to identify critical connectivity. We demonstrate our proposed approach using a recent connected vehicles model. To the best of our knowledge, this paper presents the first application of information theory for analyzing critical vehicle connectivity in the context of connected autonomous vehicles.
|
|
13:45-13:48, Paper WeCT5.6 | |
Nash Bargaining Games for Load Management with Different Levels of Information Sharing |
|
Hu, Miaomiao | University of Florida |
Kushwaha, Dhruv | University of Florida |
Abdollahi Biron, Zoleikha | University of Florida |
Keywords: Control of Smart Buildings and Microgrids, Power and Energy Systems, Transportation Systems
Abstract: Encouraging parked EVs to charge during off-peak hours and discharge during peak hours improves load management in the microgrid. Sharing information and forming coalitions can offer additional benefits to both the Charging Station (CS) and the EVs. However, it is crucial to ensure the privacy and confidentiality of the information generated by EV users. Considering different levels of information sharing, three Nash Bargaining (NB) games with overload penalty are proposed to manage the energy trading among participants in the microgrid. Experimental findings demonstrate that the proposed NB games achieve substantial reductions in Mean Squared Error (MSE) while simultaneously enhancing social welfare compared to noncooperative Stackelberg game. Furthermore, the experiments confirm that increased information sharing leads to improved MSE outcomes.
|
|
13:48-13:51, Paper WeCT5.7 | |
Energy-Aware Collaborative Teaming and Route Planning for Unmanned Ground Vehicles |
|
Miller, Noah | Clemson University |
Goulet, Nathan | Clemson University |
Ayalew, Beshah | Cemson University |
Keywords: Unmanned Ground and Aerial Vehicles, Path Planning and Motion Control
Abstract: Given a list of unordered spatial tasks to be completed by a team of unmanned ground vehicles (UGVs), this paper outlines a mission planning formulation that solves for task assignments, charging rendezvous between worker UGVs and a mobile charging host UGV, and plans routes that minimize overall energy usage and task completion time for the team. When energy capacity constraints are imposed for the charging host, which is allowed to only re-charge at the depot, we find that mixed-integer solvers for the mission planner fail to converge to exact solutions even for small team sizes and number of tasks. To reduce the computational burden of the mission planner, we incorporate a k-means clustering algorithm for the pre-assignment of tasks to specific UGV sub-teams. Demonstrative results are included showing the benefits of this approach against a less formal rule-based approach.
|
|
13:51-13:54, Paper WeCT5.8 | |
Comparison of Real Time Hybrid Simulation with Magnetorheological Damper to Virtual and Physical System Reproductions |
|
Nguyen, Nicholas | University of Connecticut |
Christenson, Richard | University of Connecticut |
Tang, Jiong | University of Connecticut |
Keywords: Motion and Vibration Control, Modeling and Validation, Linear Control Systems
Abstract: Real-time hybrid simulation (RTHS) is a seismic response simulation method utilized for assessing systems with nonlinear and challenging-to-model elements. This approach integrates numerical and physical components, allowing for time efficient and cost-effective experimentation. In this instance, a magnetorheological (MR) damper is used as the physical substructure due to its inherent nonlinear, hysteretic behavior, while a simple four-story structure serves as the numerical substructure. With RTHS comes challenges with time delays emerging from hardware components and computational time. These may cause system instability and are therefore compensated for accordingly. The effectiveness of the proposed real-time hybrid testing framework will be validated by comparing its results with both analytical and physical recreations of the experimental system. This comparison aims to confirm the RTHS systems reliability and its potential as a testing method for systems with more complex numerical and physical substructures.
|
|
13:54-13:57, Paper WeCT5.9 | |
Guided Policy Search for Stabilizing Contact-Rich Motion Plans |
|
Dagher, Christopher | Boise State University |
Silva, Chandika | Boise State University |
Satici, Aykut C. | Boise State University |
Poonawala, Hasan A. | University of Kentucky |
Keywords: Machine Learning in modeling, estimation, and control, Path Planning and Motion Control, Robotics
Abstract: Learning policies for contact-rich manipulation is a challenging problem due to the presence of multiple contact modes with different dynamics, which complicates state and action exploration. Contact-rich motion planning uses simplified dynamics to reduce the search space dimension, but the found plans are then difficult to execute under the true object-manipulator dynamics. This paper presents an algorithm for learning controllers based on guided policy search, where motion plans based on simplified dynamics define rewards and sampling distributions for policy gradient-based learning. We demonstrate that our guided policy search method improves the ability to learn manipulation controllers, through a task involving pushing a box over a step.
|
|
13:57-14:00, Paper WeCT5.10 | |
Tractor-Trailer Vehicle Rollover Avoidance Using Chance-Constrained Reference Governor and Data-Driven Ultra-Local Model |
|
Ward, Jacob | Auburn University |
Li, Nan | Tongji University |
Bevly, David M. | Auburn University |
Brown, Lowell S. | Daimler Truck North America |
Keywords: Uncertain Systems and Robust Control, Stochastic Systems, Path Planning and Motion Control
Abstract: This work presents a novel method for preventing tractor-trailer rollovers through the use of a chance-constrained reference governor and a data-driven ultra-local model. Typical commercial rollover prevention systems function by adding a system that interfaces with the trailer brakes and is capable of measuring the lateral acceleration of the trailer. If certain lateral acceleration thresholds are exceeded the trailer brakes will activate to reduce the speed of the tractor-trailer system. The system proposed in this work assumes that the truck is semi-autonomous and following a path generated by some high-level path planner. Rollover avoidance is then achieved by minimally modifying the reference path to ensure that a prescribed lateral acceleration threshold is not exceeded through the use of a reference governor. Because, in a realistic system, stochastic disturbances such as estimation errors or wind disturbances exist and since an analytical model of the truck is likely unknown, a data-driven model is leveraged along with a chance-constrained formulation of the reference governor. This chance-constrained reference governor is demonstrated to successfully constrain the tractor-trailer lateral acceleration below a prescribed threshold with a probability β.
|
|
14:00-14:03, Paper WeCT5.11 | |
A Mixed Integer Linear Programming Approach to Minimum-Time Trajectory Generation Considering Traffic Lights for Class 8 Vehicles |
|
Ward, Jacob | Auburn University |
Ellison, Evan | Auburn University |
Bevly, David M. | Auburn University |
Brown, Lowell S. | Daimler Truck North America |
Keywords: Intelligent Autonomous Vehicles, Optimal Control, Transportation Systems
Abstract: This work presents a new methodology for planning and tracking a time-optimal trajectory for class 8 vehicles on roads in which traffic lights exist. This approach takes the non-convex nature of the problem and simplifies it by introducing a set of integer decision variables. Due to the planned trajectory taking place over distances greater than a kilometer, an inner loop controller is developed to refine the high-level trajectory and guarantee that the ego vehicle does not cross into intersections during a red light phase. In order to account for model differences between the inner loop and outer loop systems an event triggered replan methodology is introduced. This event triggered replan allows for the high-level plan to adjust the trajectory to the current states of the ego vehicle without creating an unnecessary computational burden. This full system is then evaluated on three test scenarios in which the ego vehicle must navigate through two intersections with varying red light phases. The tests successfully demonstrate the feasibility of this method for generating safe, time-optimal trajectories for class 8 vehicles.
|
|
14:03-14:06, Paper WeCT5.12 | |
Graph Attention Inference of Network Topology in Multi-Agent Systems |
|
Kolli, Akshay | University of Massachusetts, Lowell |
Azadeh, Reza | University of Massachusetts Lowell |
Jerath, Kshitij | University of Massachusetts Lowell |
Keywords: Machine Learning in modeling, estimation, and control, Multi-agent and Networked Systems, Linear Control Systems
Abstract: Accurately identifying the underlying graph structures of multi-agent systems remains a difficult challenge. Our work introduces a novel machine learning-based solution that leverages the attention mechanism to predict future states of multi-agent systems by learning node representations. The graph structure is then inferred from the strength of the attention values. This approach is applied to both linear consensus dynamics and the non-linear dynamics of Kuramoto oscillators, resulting in implicit learning the graph by learning good agent representations. Our results demonstrate that the presented data-driven graph attention machine learning model can identify the network topology in multi-agent systems, even when the underlying dynamic model is not known, as evidenced by the F1 scores achieved in the link prediction.
|
|
14:06-14:09, Paper WeCT5.13 | |
Input and Output Variables Selection within Non-Parametric System Identification |
|
Chernyshov, Kirill | V.A. Trapeznikov Institute of Control Sciences |
Keywords: Modelling, Identification and Signal Processing
Abstract: For the first time in the literature, a unified single approach to the selection of both input and output variables of a stochastic model, i.e. to the problem of identifiability of a system with structural identification, is presented. This approach is based on a Rényi-consistent dependence measure of random variables and random vectors and does not assume any other a priori assumptions about the system itself or restrictive assumptions about the characteristics of the probability distributions of the components included in the system. The heterogeneity measure plays a fundamental role in the selection of variables, and in its construction, the fundamental role belongs to Rényi-consistent dependence measures. In turn, constructing a Rényi consistent dependence measure is not straightforward, meanwhile availability of such a procedure has enabled one to extend properly the Rényi axioms for dependence measures to the case of multivariate dependence.
|
|
14:09-14:12, Paper WeCT5.14 | |
Optimized Excitation Signal Tailored to Pertinent Dynamic Process Characteristics |
|
Herkersdorf, Max Heinz | University of Siegen |
Nelles, Oliver | University of Siegen |
Kösters, Tarek | University of Siegen |
Keywords: Machine Learning in modeling, estimation, and control, Estimation
Abstract: The effectiveness of data-driven techniques significantly relies on the input signal used to generate the training data. Nevertheless, there is a notable gap in research when it comes to designing excitation signals for identifying nonlinear dynamic systems, likely because of the challenges involved. Based on current knowledge, it is crucial for excitation signals to effectively capture the nonlinearity across the entire operational area and to gather insights into the area-specific dynamic process characteristics. The Incremental Dynamic Space-Filling Design (IDS-FID) strategy designs excitation signals to achieve a space-filling distribution across the input space of a nonlinear approximator used in external dynamics modeling, gathering information throughout its operational area. Simultaneously, the approach enables for a heightened focus on either the system’s steady- state or transient responses during information acquisition by altering the excitation signal’s dynamics, facilitating targeted insights into dynamic process characteristics.
|
|
14:12-14:15, Paper WeCT5.15 | |
Integrated Sensor Anomaly Detection and Structural Health Monitoring in Pressure Vessel Using a Unified Hierarchical Model |
|
Zhou, Qianyu | University of Connecticut |
Zhang, Yang | University of Connecticut |
Tang, Jiong | University of Connecticut |
Keywords: Machine Learning in modeling, estimation, and control, Cyber physical systems, Sensors and Actuators
Abstract: In the domain of structural health monitoring (SHM) for pressure vessels used in space habitats, accurate identification of sensor anomalies and structural damage is paramount. Sensor anomalies, which manifest as deviations from expected measurement patterns, pose significant challenges for reliable SHM. To address this, we propose a single-model hierarchical framework that initially detects sensor anomalies to filter out aberrant data. This foundational step leverages synthetic sensor anomalies, generated using physical equations to simulate realistic faults such as drift, bias, noise, and freeze. This method ensures that the deep learning model is trained on data that not only represents statistically significant anomalies but also adheres to physical plausibility. The integrated approach enhances the robustness of the anomaly detection and systematically ensures the extraction of dependable data for structural damage identification. This framework facilitates efficient data management across multiple sensors and ensures that physically implausible anomalies are comprehensively identified, thereby providing a more effective and reliable approach to structural health monitoring of pressure vessels, enhancing safety in critical space habitat applications.
|
|
14:15-14:18, Paper WeCT5.16 | |
Data-Driven Decentralized Control Design for Discrete-Time Large-Scale Systems |
|
Liao, Jiaping | Southwest Jiaotong University |
Lu, Shuaizheng | Augusta University |
Wang, Tao | Southwest Jiaotong University |
Xiang, Weiming | Augusta University |
Keywords: Control Design, Large Scale Complex Systems, Adaptive and Learning Systems
Abstract: In this paper, a data-driven approach is developed for controller design for a class of discrete-time large-scale systems, where a large-scale system can be expressed in an equivalent data-driven form and the decentralized controllers can be parameterized by the data collected from its subsystems, i.e., system state, control input, and interconnection input. Based on the developed data-driven method and the Lyapunov approach, a data-driven semi-definite programming problem is constructed to obtain decentralized stabilizing controllers. The proposed approach has been validated on a mass-spring chain model, with the significant advantage of avoiding extensive modeling processes.
|
|
14:18-14:21, Paper WeCT5.17 | |
Temporal Logic Guided Safe Navigation for Autonomous Vehicles |
|
Parameshwaran, Aditya | Clemson University |
Wang, Yue | Clemson Univeristy |
Keywords: Unmanned Ground and Aerial Vehicles, Path Planning and Motion Control, Optimal Control
Abstract: Safety verification for autonomous vehicles (AVs) and ground robots is crucial for ensuring reliable operation given their uncertain environments. Formal language tools provide a robust and sound method to verify safety rules for such complex cyber-physical systems. In this paper, we propose a hybrid approach that combines the strengths of formal verification languages like Linear Temporal Logic (LTL) and Signal Temporal Logic (STL) to generate safe trajectories and optimal control inputs for autonomous vehicle navigation. We implement a symbolic path planning approach using LTL to generate a formally safe reference trajectory. A mixed integer linear programming (MILP) solver is then used on this reference trajectory to solve for the control inputs while satisfying the state, control and safety constraints described by STL. We test our proposed solution on two environments and compare the results with popular path planning algorithms. In contrast to conventional path planning algorithms, our formally safe solution excels in handling complex specification scenarios while ensuring both safety and comparable computation times.
|
|
14:21-14:24, Paper WeCT5.18 | |
Model-Based Control of Water Treatment with Pumped Water Storage |
|
Mauery, Ryan | The Pennsylvania State University |
Busse, Margaret | Pennsylvania State University |
Kovalenko, Ilya | Pennsylvania State University |
Keywords: Modelling and Control of Environmental Systems, Control Applications, Chemical Process Control
Abstract: Water treatment facilities are critical infrastructure they must accommodate dynamic demand patterns without system disruption. These patterns can be scheduled, such as daily residential irrigation, or unexpected, such as demand spikes from withdrawals for fire management. The critical necessity of clean, safe, and reliable water requires water treatment control strategies that are insensitive to disturbances to guarantee that demand will be met. One essential problem in achieving this is the minimization of energy costs in the process of meeting water demand, especially as the need for decarbonization persists. This work develops a control-oriented hydraulic model of a water treatment facility with integrated pumped storage and introduces a model predictive control strategy for scheduling treatment plant system operations to minimize greenhouse gas emissions and safely meet water demand.
|
|
14:24-14:27, Paper WeCT5.19 | |
Singularity-Free Approximate Waypoint Tracking Controller for Underactuated Magnetic Robots |
|
Raval, Suraj | University of Maryland |
Bhattacharjee, Anuruddha | Johns Hopkins University |
Chen, Xinhao | Johns Hopkins University |
Mair, Lamar | Weinberg Medical Physics, Inc |
Krieger, Axel | Johns Hopkins University |
Diaz-Mercado, Yancy | University of Maryland |
Keywords: Nonlinear Control Systems, Path Planning and Motion Control, Robotics
Abstract: Magnetic robots use external magnetic fields to draw energy, generating steering capabilities crucial for minimally invasive surgeries and enabling next generation untethered surgical tool miniaturization. However, accurate control is challenging due to configuration-dependent singularities in the manipulation Jacobian, which can cause unsafe behavior with standard controls. We analyze the nonlinear nature of magnetic fields to understand singularity-free control limits without adding more magnetic actuators, which increases bulk and cost. Using Chow's Theorem, we study the motion feasibility of a single magnetic robot moving in a plane, powered by stationary electromagnets. We determine the degree of nonholonomy for an underactuated case and show that any desired motion in the state-space can be approximated with more complex controls. We deploy an approximate-tracking controller to steer a magnetic robot between any two points in the state-space, avoiding singularities. Simulations show a 0.82 mm RMS positional tracking error for an 8 mm long cylindrical magnetic tool using our method.
|
|
14:27-14:30, Paper WeCT5.20 | |
Design and Cost Analysis for a Second-Life Battery-Integrated Photovoltaic Solar Container for Rural Electric Vehicle Charging |
|
Abdullah Eissa, Magdy | Tennessee Technological University |
Chen, Pingen | Tennessee Technological University |
Keywords: Automotive Systems, Power and Energy Systems
Abstract: Mobile charging stations (MCSs) play a pivotal role in mitigating charging deserts prevalent in rural areas by offering the flexibility to be transported to desired locations for electric vehicle (EV) charging. MCSs address concerns related to power infrastructure limitations and locational constraints, thereby alleviating range anxiety. Despite this significance, current research exhibits a notable dearth of investigations focusing on off-grid energy storage systems that integrate renewable energy sources and repurpose second-life batteries (SLBs) retired from EVs for EV charging stations (EVCS). While conventional power grid sources are conventionally relied upon for EV charging, further inquiry is imperative to explore the potential of off-grid systems leveraging renewable energy and SLBs. Addressing this research gap holds substantial promise in advancing sustainable EV charging infrastructure. This study endeavors to fill this void by presenting the sizing design and cost analysis of a standalone photovoltaic (PV) system integrated with an SLB bank for EVCS in public parks. The methodology commences by utilizing real-world power demand data collected from Tennessee state park as input and subsequently determining capacity loss based on the selected aging model to decide appropriate battery sizes. Finally, the Life Cycle Cost (LCC) estimation of proposed charging stations inputs for the cost analysis. The results indicate that the proposed SLB-based EVCS can reduce LCC by 32.16%, when compared to the baseline.
|
|
WeCT7 |
Prime 3 |
Learning and Adaptive Control |
Regular Session |
Chair: Chen, Xu | University of Washington |
Co-Chair: Anubi, Olugbenga | Florida State University |
|
13:30-13:45, Paper WeCT7.1 | |
Efficient Neural Hybrid System Learning and Transition System Abstraction for Dynamical Systems |
|
Yang, Yejiang | Southwest Jiaotong University |
Mo, Zihao | Augusta University |
Xiang, Weiming | Augusta University |
Keywords: Modeling and Validation, Machine Learning in modeling, estimation, and control
Abstract: This paper proposes a neural network hybrid modeling framework for dynamics learning to promote an interpretable, computationally efficient method of dynamics learning and system identification. First, a low-level model will be trained to learn the system dynamics, which utilizes multiple simple neural networks to approximate the local dynamics generated from data-driven partitions. Then, based on the low-level model, a high-level model will be trained to abstract the low-level neural hybrid system model into a transition system that allows Computational Tree Logic (CTL) verification to promote the model's ability to handle human interaction and verification efficiency.
|
|
13:45-14:00, Paper WeCT7.2 | |
Learning to Detect Slip through Tactile Estimation of the Contact Force Field and Its Entropy Properties |
|
Hu, Xiaohai | University of Washington |
Venkatesh, Aparajit | University of Washington |
Wan, Yusen | University of Washington |
Zheng, Guiliang | University of Washington |
Jawale, Neel Anand | University of Washington, Seattle |
Kaur, Navneet | University of Washington |
Chen, Xu | University of Washington |
Birkmeyer, Paul | Amazon |
Keywords: Robotics, Mechatronic Systems
Abstract: Slip detection during object grasping and manipulation plays a vital role in object handling. Visual feedback can help devise a strategy for grasping. However, for robotic systems to attain a proficiency comparable to humans, integrating artificial tactile sensing is increasingly essential, especially in consistently handling unfamiliar objects. We introduce a novel physics-informed, data-driven approach to detect slip continuously for control-oriented tasks. Our work leverages the inhomogeneity of tactile sensor readings during slip events to develop distinct features and formulates slip detection as a classification problem. We test multiple data-driven models on 10 common objects under different loading conditions, textures, and materials to evaluate our approach. The resulting best classification algorithm achieves a high average accuracy of 95.61%. Practical application in dynamic robotic manipulation demonstrates the effectiveness of the proposed real-time slip detection and prevention.
|
|
14:00-14:15, Paper WeCT7.3 | |
Nullspace Adaptive Identification of Plant and Actuator Model Parameters for Underactuated Ground Vehicles: Theory and Experimental Evaluation |
|
Elsberry, Allan | Johns Hopkins University |
Dawkins, Jeremy | United States Naval Academy |
Mao, Annie | Johns Hopkins University |
Whitcomb, Louis | Johns Hopkins Univ., |
Keywords: Modeling and Validation, Estimation, Adaptive and Learning Systems
Abstract: This paper reports a novel Nullspace Adaptive Identification (NSAID) algorithm to estimate plant and actuator model parameters for an underactuated ground vehicle with dynamics represented by a 3 degree-of-freedom second-order dynamic model and reports an evaluation of its performance in simulation and experiments. Precise identification of ground vehicle plant and actuator model parameters is critical for physically realistic model-based simulation, control, navigation, and fault detection. However, ground vehicle plant model parameters, such as vehicle mass, moments of inertia, tire cornering stiffness and actuator model parameters such as motor torque constants are generally not possible to estimate a priori from first principles, and can change with varying payloads, configurations, and driving environments, and thus must be determined experimentally. Some parameter identification methods such as least squares regression depend on acceleration measurements, and most identification methods assume a priori knowledge of actuator parameters. In contrast, NSAID estimates plant and actuator parameters simultaneously, does not require acceleration measurements, can be utilized offline or online during vehicle operation, and can be applied with open or closed-loop control.
|
|
14:15-14:30, Paper WeCT7.4 | |
Vision-Based Learning of Emergent Behavior of Magnetic Microrobots |
|
Alaviani, Seyyed Shaho | Florida State University |
Katuri, Jaideep | FAMU-FSU College of Engineering |
Ali, Jamel | FAMU-FSU College of Engineering |
Anubi, Olugbenga | Florida State University |
Keywords: Robotics, Machine Learning in modeling, estimation, and control, Multi-agent and Networked Systems
Abstract: In this paper, the problem of forecasting/predicting emergent behavior of untethered magnetic microrobots (MMs) from raw grayscale time-series image data is considered. An interpretable machine learning (ML) method is proposed to forecast/predict such behavior without measuring position, velocity, and/or orientation of each untethered MM. The approach is reduced-order and totally data-driven without knowing any knowledge about the physics and/or interacting dynamics of MMs. The proposed method may provide information/hypothesis about the unknown underlying dynamics of the phenomenon. The method is a promising approach for crossing the reality gap in microrobotics. Experimental results are given in order to validate the proposed method and its effectiveness.
|
|
14:30-14:45, Paper WeCT7.5 | |
Iterative Stochastic Model Predictive Control with Multi-Step Uncertainty Learning Model: Vehicle Lane Change Case Study |
|
Zakeri, Hasan | Illinois Institute of Technology |
HomChaudhuri, Baisravan | Illinois Institute of Technology |
Keywords: Optimal Control, Control Applications, Intelligent Autonomous Vehicles
Abstract: This paper focuses on developing an iterative stochastic model predictive control (MPC) method that uses a multi-step Gaussian Process (GP) regression-based state distribution error prediction model for systems with state- and control-dependent uncertainties. A mismatch between the actual system and its control-oriented model is ubiquitous, and these molding errors are generally state- and control-dependent. These errors can disrupt the operation of safety-critical systems. Stochastic MPC methods can ensure that the system stays within the safe region with a given probability, but they require prediction of the future state distributions of the system over the horizon. Predicting the future state distribution of systems with state- and control-dependent uncertainty is difficult as we generally do not have their closed form expressions. We thus first present a multi-step Gaussian Process (GP) regression method to learn the uncertainty propagation model for systems with state- and control-dependent uncertainties. Since the predicted state distributions are outputs of GP regression models, it is difficult to obtain the exact deterministic equivalent of the stochastic problem for an optimization solver. Hence, we develop a new computationally tractable iterative stochastic MPC approach that can utilize the multi-step GP regression model and numerically solve the optimization problem while enforcing the probabilistic constraints. We also perform a case study on vehicle lateral control problems, where we learn the vehicle's error propagation model during lane changes. Simulation results show the efficacy of our proposed method.
|
|
14:45-15:00, Paper WeCT7.6 | |
Learning Dissipative Hamiltonian Dynamics with Reproducing Kernel Hilbert Spaces and Random Fourier Features |
|
Smith, Torbjørn | Norwegian University of Science and Technology |
Egeland, Olav | Norwegian Univ. of Sci. & Tech |
Keywords: Machine Learning in modeling, estimation, and control
Abstract: This paper presents a new method for learning dissipative Hamiltonian dynamics from a limited and noisy dataset. The method uses the Helmholtz decomposition to learn a vector field as the sum of a symplectic and a dissipative vector field. The two vector fields are learned using two reproducing kernel Hilbert spaces, defined by a symplectic and a curl-free kernel, where the kernels are specialized to enforce odd symmetry. Random Fourier features are used to approximate the kernels to reduce the dimension of the optimization problem. The performance of the method is validated in simulations for two dissipative Hamiltonian systems, and it is shown that the method improves predictive accuracy significantly compared to a method where a Gaussian separable kernel is used.
|
| |