Keywords: Modeling and Validation, Control Applications, Power and Energy Systems
Abstract: This paper presents the acausal modeling, validation, and control of floating offshore wind turbines (FOWTs) utilizing the Control-oriented, Reconfigurable, and Acausal Floating Turbine Simulator (CRAFTS), a wind turbine simulation tool under development by the authors to facilitate the control co-design (CCD) of FOWTs. The model simulates the IEA-15MW reference turbine and the semi-submersible VolturnUS-S platform, incorporating essential aero and hydrodynamics. Verification and validation are conducted using numerical data from the industrial-standard simulation platform OpenFAST and experimental data from the Floating Offshore-wind and Controls Advanced Laboratory (FOCAL) project, in which the authors were involved. Numerical results demonstrate CRAFTS’ ability to qualitatively capture loads and responses across various load cases, while highlighting the impact of the control system under different wind and wave conditions. This paper serves as a reference for wind turbine researchers, providing relevant information such as general characteristics, system frequencies, damping effects, and various internal reaction forces.

Keywords: Unmanned Ground and Aerial Vehicles, Modelling, Identification and Signal Processing, Control Applications
Abstract: This paper presents an approach to wind stress estimation and wind impact modeling using data collected onboard a quadrotor drone. First, wind force and wind velocity are derived based on experimental data obtained in varying wind conditions. Second, a wind disturbance is characterized by exploiting an auto-regressive moving average with an exogenous input (ARMAX) model. Lastly, validation of the methodologies presented for wind estimation and disturbance modeling is proposed using a baseline control architecture, which was designed for calm air conditions.

Keywords: Mechatronic Systems, Modeling and Validation, Motion and Vibration Control
Abstract: This paper presents a comprehensive study on the modeling of a motor-driven servotable with significant flexible modes and nonlinear disturbances (i.e. Coulomb friction and gears’ backlash). Through in-depth system identifications, system parameters are offline estimated and the complex flexible modes are captured. Fitting models in different frequency ranges are obtained via the MATLAB toolbox, revealing the high-order dynamics. In addition, a novel nonlinear two-mass model is proposed and it successfully captures the flexibility in the table’s shaft, the Coulomb friction in the bearing and the backlash in the gear set. Model validations have been successfully conducted to demonstrate the effectiveness of the proposed model.

Keywords: Estimation, Large Scale Complex Systems, Modeling and Validation
Abstract: Traditional models used in nuclear reactor power modeling face challenges related to system parameter uncertainties, loss of dynamic information due to model simplifications or linearizations, and, in some cases, lack of proper rod calibration for accurate input modeling. This article presents a predictive design for the power of the IAN-R1 nuclear reactor in Colombia, based on the Koopman operator approach using Extended Dynamic Mode Decomposition(EDMD). Since this model relies solely on system measurements, it not only addresses the aforementioned issues but also provides a system that is easily adaptable to changes in core configuration or parameter modifications. The poles of the Koopman model are compared with those of other models to validate its operation, and the results are contrasted with real data on the operation of the reactor, evidencing the good performance of the proposed model.

Keywords: Machine Learning in modeling, estimation, and control, Modeling and Validation, Multi-agent and Networked Systems
Abstract: Modeling human trajectories in crowded environments is challenging due to the complex nature of pedestrian behavior and interactions. This paper proposes a geometric graph neural network (GNN) architecture that integrates domain knowledge from psychological studies to model pedestrian interactions and predict future trajectories. Unlike prior studies using complete graphs, we define interaction neighborhoods using pedestrians' field of view, motion direction, and distance-based kernel functions to construct graph representations of crowds. Evaluations across multiple datasets demonstrate improved prediction accuracy through reduced average and final displacement error metrics. Our findings underscore the importance of integrating domain knowledge with data-driven approaches for effective modeling of human interactions in crowds.

Keywords: Biomechanical Systems, Marine Systems, Modelling, Identification and Signal Processing

Keywords: Automotive Systems, Machine Learning in modeling, estimation, and control
Abstract: This paper introduces a comprehensive two-stage methodology designed to enhance the efficiency and cost-effectiveness of second-life batteries (SLBs) retired from electric vehicles (EVs) for stationary applications. Initially, advanced machine learning techniques, including convolutional neural network (CNN) and long short-term memory (LSTM), are employed to forecast the starting threshold for utilizing SLBs, termed the second-life starting (SLS) threshold, based on battery state of health (SOH) data. By leveraging data from the NASA Battery Dataset, comprising voltage, current, and temperature profiles, the proposed model achieves precise estimations, thereby enhancing decision-making processes. Subsequently, leveraging the estimated second-life starting time, capacity, and SOH of SLBs, comprehensive optimal battery sizing is conducted for SLB-based DC fast charger (DCFC) in rural areas, based on the experimental data collected from a 2019 Nissan Leaf EV. The results of the SLS threshold based on SOH estimation demonstrate high accuracy, with minimal estimation errors of 0.0257 and 0.0216 for CNN and LSTM, respectively. This approach demonstrates practical applicability, offering a means to maximize the potential of SLBs within the stationary energy storage domain, thus advancing sustainable energy solutions with significant impact.

Keywords: Assistive and Rehabilitation Robotics, Optimal Control, Machine Learning in modeling, estimation, and control
Abstract: Hip exoskeletons offer significant potential for enhancing human movement, especially for those with mobility impairments. However, optimizing their performance typically involves lengthy discrete and continuous optimization methods. To address this, we propose a novel approach using machine learning to predict controller parameter classes, aiming to improve the tuning process. Our method relies on subject-specific anthropometric data to predict optimal controller parameters for hip exoskeletons. Through a machine learning framework, we develop predictive models to determine the most effective parameter settings tailored to individual users. By employing feature engineering, data synthesis techniques, and model training, we enhance the initialization of Bayesian Human-in-the-loop (HIL) optimization. Results indicate that our machine learning models can predict control parameter classes with 75% accuracy, leading to a 9.98% improvement in optimized exoskeleton performance for users.

Keywords: Machine Learning in modeling, estimation, and control, Distributed Parameter Systems, Sensors and Actuators
Abstract: Estimating the 2-dimensional (2-D) temperature distribution of lithium-ion pouch cells is essential to ensure their safety and performance. In large-format pouch cells, mapping sparse temperature measurements to the non-uniform temperature distribution is a challenging problem. The standard approach requires a higher dimensional spatio-temporal model, which is often difficult to identify accurately. In this work, we propose a data-driven optimal temperature sensor placement scheme and 2-D temperature distribution estimation using Principal Component Analysis (PCA). Without any physics-based model, the thermal distribution is captured by training PCA using full-grid thermal data. The reconstructability metric derived from the principal components (PCs) identifies the optimal locations of the temperature sensors. The thermal characteristics learned by the PCs are used to reconstruct the remaining grid point temperatures from the available measurements estimating the full 2-D temperature distribution. The proposed approach is validated using 5-point experimental temperature measurement data for a commercial pouch cell and proved to accurately estimate the temperature distribution with an average error less than 0.29^oC for realistic current profiles with sparse and multiple temperature sensing scenarios.

Keywords: Intelligent Autonomous Vehicles, Machine Learning in modeling, estimation, and control, Path Planning and Motion Control
Abstract: This work investigates an alternative approach to current control systems for the Automated Driving (AD) of shuttle vehicles on dedicated roads. The proposed solution decouples the problem into two levels: a Deep Deterministic Policy Gradient (DDPG) Reinforcement Learning (RL) agent and a dedicated vehicle logic generating Virtual Lane (VL) data to eliminate redundancy and allow for smooth lane changes on curved roads. The training uses an environment defined through a model-based simulation, exploiting MATLAB Inc. (2020) and Simulink tools, and has been conducted following a Curriculum Learning strategy. The performance of the introduced approach have been evaluated by testing the agent capabilities and exploring its behavior in the presence of external disturbances in the controlled states.

Keywords: Optimal Control, Unmanned Ground and Aerial Vehicles, Adaptive and Learning Systems
Abstract: The control of quadcopters poses significant challenges due to their complex dynamics characterized by highly nonlinear couplings, high system order, and under-actuation. This paper presents a novel control solution aimed at achieving near-optimal trajectory tracking for quadcopters. A near-optimal solution based on approximate dynamic programming is proposed to address the curse of dimensionality inherent in traditional dynamic programming, employing a single network adaptive critic. Extensive simulations validate the effectiveness and robustness of the proposed solution.

Keywords: Education for modeling, estimation and control, Modeling and Validation
Abstract: Most educational laboratory equipment utilized in vibrations and control labs is heavy and bulky, limiting its use within lab spaces. Further, mechanical engineering students typically gain hands-on experience only in labs that are offered separately from fundamental courses, unlike other engineering programs. Thus, the extent of student learning from labs, given their limited time, remains a concern considering students’ learning opportunities. This study presents the design and development of two portable and low-cost lab equipment devices: a compliant mechanism with unbalanced rotating masses and a DC motor with beam control under disturbances. While the former is utilized to demonstrate fundamentals of vibrations, derivation of the mathematical model, simulation, and comparison of theory and experimental data, the latter can be used for demonstration of disturbance control. Virtual labs are developed for both mechanisms in MATLAB Simscape.

Keywords: Security and Privacy, Cyber physical systems, Control Design
Abstract: Cyber-physical systems (CPSs) require reliable, safe, and secure control of critical infrastructure, combining computational and networking capabilities, which heighten the risk of cyber attacks. These attacks can disrupt the physical process, causing unforeseen consequences. One solution is the use of fully homomorphic encryption (FHE) to protect the control loop, allowing for secure computations and communications without compromising signal and control system privacy. One challenge with FHE, however, is its requirement for inputs to be integers. This paper introduces a modified FHE scheme based on the Learning With Errors (LWE) problem. Our proposed scheme leverages a generalized LWE encoding function and modifies the Gentry-Sahai-Waters (GSW) gadget decomposition tool to encrypt the control system. Using the modified LWE scheme, we formalize a fully encrypted control system, supported by simulated results.

Keywords: Intelligent Autonomous Vehicles, Security and Privacy, Machine Learning in modeling, estimation, and control
Abstract: Connected and Autonomous Vehicles (CAVs) are vulnerable to security risks due to their large dependence shared communication networks. Motivated by this limitation, existing literature has largely focused on the CAV security problems from cyber-attack detection and accommodation point of view. On the other hand, understanding adversarial models and adversarial examples are critical in verification and testing of attack detection and accommodation strategies. However, existing research lack such adversarial example studies in the context of CAVs. In this work, we address this gap and propose a framework for model-free reinforcement learning-based adversarial attack generation examples, which can be used for verification and testing attack diagnostic strategies. We show a simulation study to demonstrate the effectiveness of the proposed framework.

Keywords: Machine Learning in modeling, estimation, and control, Adaptive and Learning Systems, Control Design
Abstract: Reinforcement learning (RL) has seen considerable development in terms of safety guarantees over last few years for nonlinear dynamical systems. Unlike control theoretic approaches, RL still has scope for improvement in terms of system stability for a computed policy and state constraint satisfaction. In this work we address the problem of system stability by proposing use of Lyapunov theory to provide stability guarantees for the system for the policy computed via RL. A soft actor-critic framework is considered where the actor network is updated based on the proposed Lyapunov violation condition. We propose use of Koopman operator theory to approximate the nonlinear continuous time dynamical system as a linear system. Performance of the proposed framework is tested on a parking environment by comparing results between a baseline vanilla soft actor-critic and proposed Lyapunov-based soft actor-critic (Lb-SAC) algorithm . The numerical results show the proposed framework minimizes Lyapunov condition violation, produce smoother actions and achieves desired parking environment goals.

Keywords: Modelling, Identification and Signal Processing, Nonlinear Control Systems, Linear Control Systems
Abstract: The paper presents a framework for online learning of the Koopman operator using streaming data. Many complex systems for which data-driven modeling and control are sought provide streaming sensor data, the abundance of which can present computational challenges but cannot be ignored. Streaming data can intermittently sample dynamically different regimes or rare events which could be critical to model and control. Using ideas from subspace identification, we present a method where the Grassmannian distance between the subspace of an extended observability matrix and the streaming segment of data is used to assess the `novelty' of the data. If this distance is above a threshold, it is added to an archive and the Koopman operator is updated if not it is discarded. Therefore, our method identifies data from segments of trajectories of a dynamical system that are from different dynamical regimes, prioritizes minimizing the amount of data needed in updating the Koopman model and furthermore reduces the number of basis functions by learning them adaptively. Therefore, by dynamically adjusting the amount of data used and learning basis functions, our method optimizes the model's accuracy and the system order.

Keywords: Optimal Control, Linear Control Systems, Machine Learning in modeling, estimation, and control
Abstract: This paper demonstrates that the value iteration (VI) algorithm of reinforcement learning of discrete-time (DT) linear-quadratic (LQ) optimal control problem converges very slowly mostly linearly, compared to the quadratic rate of convergence of the corresponding policy iteration (PI) algorithm. The VI algorithm produces non-monotonically decreasing or increasing sequences that converge to the optimal value either from below or from above depending on the choice of initial conditions. It is remarkable that the VI algorithm converges even in the case when the initial condition is very far from the optimal value by several orders of magnitude, and when the initial condition is not stabilizing. The PI algorithm generates a non-increasing sequence that monotonically converges from above to the optimal value assuming the initial condition (feedback gain) is stabilizing. The convergence rate for the PI algorithm is quadratic, which assures its fast convergence. It is shown in this paper that the convergence of the VI algorithm can be made quadratic by using the doubling algorithm. We precisely state a condition needed for convergence of the VI algorithm, which is milder than the corresponding convergence condition for the PI algorithm. We have also shown that the newly proposed VI algorithm requires less computational effort than the PI algorithm. Several numerical examples are solved to document the presented results.

Keywords: Machine Learning in modeling, estimation, and control, Nonlinear Control Systems, Electromechanical systems
Abstract: Mechanical system models derived from the Euler-Lagrange equations exhibit nonlinear characteristics, especially in electromechanical systems, where factors like coupling terms and dissipation effects play a significant role. Using measured signals and appropriate variable adjustments, the associated nonlinear physics-based model can be transformed into equivalent state-space descriptions in the quasi Linear Parameter Varying (qLPV) framework. In this paper, we leverage an extended Lagrangian-informed deep neural network to learn the dynamics of nonlinear systems in the qLPV framework. Our approach integrates underlying physical system knowledge into the neural network architecture, hence enhancing its ability to predict scheduling parameters and learn the qLPV model. Furthermore, we address the reference tracking model predictive control (MPC) design problem using the learned qLPV model and show that our designed MPC can accurately track references.

Keywords: Machine Learning in modeling, estimation, and control, Multi-agent and Networked Systems, Linear Control Systems
Abstract: We propose a novel deep learning-based distributed control approach to solve the problem of reference tracking using distributed agents with a continuous-time model. The problem is considered as an optimization problem with a defined loss function consisting of tracking errors and consensus errors. To obtain control policies for the agents to cooperatively track a reference signal, the proposed method incorporates deep neural networks (DNNs) into a distributed control scheme and builds a computation cell consisting of a DNN and the dynamic model for each agent. The DNN in each computation cell receives agent’s output and the reference signal as its input, producing a control policy as output to act on the agent model. Next, the weights of the DNN are updated using a modified gradient algorithm tailored for continuous-time dynamics, minimizing the defined loss function over a finite time horizon. Simulation results show that the approach is able to drive distributed agents with general, possibly exponentially unstable, continuous-time dynamics to cooperatively track a reference signal. Compared to the relevant DNN based works, we utilize only the output measurement of agents and design a new structure of output feedback in which the input is linear feedback of the output and the reference signal with appropriate gains. Simulation results provide a detailed comparison of the proposed method with a reinforcement learning-based design.

Keywords: Machine Learning in modeling, estimation, and control, Aerospace, Stochastic Systems
Abstract: A coupled multidisciplinary system is a complex system involving the integration and interaction of multiple disciplines. In such systems, the disciplines are interconnected and influence each other, and each contains some sources of uncertainty, leading to a complicated and uncertain behavior of the overall system. Accurate modeling of the individual disciplines is crucial in the design, control, and analysis of coupled multidisciplinary systems. However, acquiring data for these disciplines through experiments or computational simulations is costly. To mitigate this challenge, we present an efficient framework aimed at acquiring the least number of informative data from each discipline, ensuring accurate estimation of the joint distribution of coupling variables. Our approach constructs a surrogate model for each discipline and focuses on acquiring a selected subset of data by prioritizing uncertainty reduction in regions critical for estimating the stationary behavior of the coupled multidisciplinary system. The efficacy of the proposed framework is demonstrated in numerical experiments using a coupled aerodynamics-structures system.

Keywords: Estimation
Abstract: The estimation of loads in the sucker-rod pump system is critical, as it significantly influences operational conditions and system performance. However, this task is inherently non-trivial due to the complex dynamics and time-dependent incompletely defined boundary conditions. In this work, the system is modeled as a damped wave equation, incorporating uncertainties present at the down-hole boundaries. By employing the PDE backstepping design method, an observer is formulated to deduce the load at the plunger end. Notably, in the absence of other measurements, the load data serve as only accessible input to drive the observer. In addition, an upper limit bound for the estimation error under the uncertain system conditions is determined. The paper is concluded with numerical simulations that suggest the applicability of the proposed estimation approach.

Keywords: Unmanned Ground and Aerial Vehicles, Robotics

Keywords: Modeling and Validation
Abstract: Thick electrodes suffer from underutilization of active materials due to ion transport limitations in thick porous structures. Flow-though electrolyte can improve cell electrochemical performance by faster ion transport. A reduced-order, electrolyte-enhanced, single particle model of lithium (Li) cells with flow convection (ESPM-C) is developed and simulated to show that flow-through electrolyte can facilitate full utilization of active materials in thick electrodes (≥ 100 μm) during intercalation at high C-rates (≥ 1C) by maintaining uniform ion concentration distribution. The model exhibits less than 1% maximum error relative to a full-order COMSOL model in concentration and voltage profiles in graphite-NCM cells with different C-rates at no flow condition. In Li metal batteries, the ESPM-C model predicts uniform ion concentration and reduced impedance in Li-NCM cells.

Keywords: Manufacturing Systems, Modelling, Identification and Signal Processing, Modeling and Validation
Abstract: Electrohydrodynamic jet (e-jet) printing has emerged as a promising technique for precise and high-resolution additive manufacturing. However, the dynamic behavior of the jet during printing, particularly the occurrence of transitional behaviors between Continuous Jet and Pulsating Jet modes, known as Natural Pulsation Initiation (NPI), poses challenges in controlling printing speed and quality. In this study, we investigate the NPI behavior of the cone-jet mode in e-jet printing, focusing on its modeling and characterization. Through experimental investigation involving 15 different liquids with varying material properties and applied potentials, we capture high-speed recorded videos to analyze the jetting dynamics. Our methodology involves calculating and modeling the flow rate of each experiment as an exponentially decaying function, enabling the extraction of time constants for these decaying functions. We develop two fitted models for the time constants of NPI and Continuous Jet (CJ) modes as a function of dimensionless voltage. Notably, our findings reveal an approximate boundary limit between NPI and CJ modes, offering valuable insights for predesign considerations of e-jet printing processes.

Keywords: Manufacturing Systems, Modeling and Validation, Control Design
Abstract: Additive manufacturing (AM) processes have experienced a surge in demand, largely driven by the growing need for customization across various industries. This paradigm shift underscores the need for more robust AM processes, where precise customization is vital to meet individual requirements. However, there does not exist a structured framework for deriving models that characterize the relationship between process parameters and pattern characteristics. Additionally, determining critical inputs and outputs, as well as control strategies for in situ AM control, further compounds this challenge.. This work addresses these challenges by proposing a structured approach to closing the loop for an extrusion-based AM printing process, demonstrating its transferability across multiple printers within the same AM family. The framework is validated through experiments conducted on two AM systems, offering insights into the underlying dynamics of the AM process, and paving the way for enhanced performance and reliability in AM.

Keywords: Machine Learning in modeling, estimation, and control, Manufacturing Systems, Uncertain Systems and Robust Control
Abstract: The complexity of Atmospheric Pressure Plasma Jet (APPJ) dynamics presents significant challenges for control design. This study introduces a learning- and Scenario-based Model Predictive Control (ScMPC) approach within a Linear Parameter-Varying (LPV) framework to address these challenges. The proposed control strategy aims to achieve robust and precise regulation of APPJs amidst modeling uncertainties and external disturbances.

The first phase of the work involves data-driven LPV-state-space system identification of APPJs using Artificial Neural Network (ANN). To account for the inevitable mismatch, Bayesian Neural Networks (BNNs) are employed. This approach not only provides the mismatch estimation but also quantifies the associated uncertainty. The statistical results derived from this estimation are used to define scenarios affecting APPJ performance. A scenario tree is constructed, and the moment-matching technique is applied to estimate the probability of these scenarios by matching the first four central moments of the samples through an optimization problem. To ensure recursive feasibility, soft constraints are incorporated into the control design formulation.

Extensive simulations were conducted to evaluate the effectiveness of the learning-based identification method and the performance and robustness of the control strategy. The accuracy of the LPV representation was assessed through the Best Fit Rate (BFR) metric, demonstrating an acceptable level of precision for the control task. The closed-loop control performance of the proposed approach was analyzed in two case studies: reference temperature tracking and thermal dose delivery. In the first scenario, the LPV-based ScMPC exhibited superior tracking performance and robustness over an LTI-based ScMPC in the presence of measurement noise and external disturbances. In the second scenario, measured by the Cumulative Equivalent Minutes (CEM) metric, the LPV-based ScMPC achieved the target CEM more quickly with less control effort.

In conclusion, the LPV-based ScMPC strategy demonstrates significant improvements in control performance and robustness for APPJ systems, while also reducing the conservativeness typical of many robust control approaches.

Keywords: Automotive Systems, Modeling and Validation
Abstract: The development of control-oriented models for the external components of an active engine cooling system in a gas engine vehicle is discussed. The external components which include the transmission oil heat exchanger, engine oil heat exchanger, and radiator play a vital role in the overall engine cooling system and are associated with specific types of losses. To achieve accurate temperature estimation, we delve into the modeling and prediction of these losses within the system. By quantifying and understanding these losses, we gain a comprehensive understanding of the thermal behavior of the system. To derive the dynamics of these temperatures, a lumped parameter concept with a mean-value approach is applied. This approach enables the development of a control-oriented model that captures the thermal dynamics of the system effectively. Simulation results with test data demonstrate the accuracy and effectiveness of our control-oriented model. These results show the model's ability to accurately predict temperatures and validate its suitability for control system design.

Keywords: Machine Learning in modeling, estimation, and control, Automotive Systems, Estimation
Abstract: DC fast charging (DCFC) of a lithium-ion battery for EV applications can cause deterioration to the battery cell health due to lithium-plating of the anode. While anode potential is a good indicator for lithium plating reaction occurrence, it cannot be monitored in commercial cells due to lack of a reference electrode. A DCFC controller informed by anode potential has the potential to achieve fast charging while simultaneously mitigating the lithium plating risk. We propose a data-driven anode voltage estimation method based on novel physics inspired features, a time delay neural network, and experimental 3-electrode data. The estimation method is shown to be accurate and fast, and its use in a closed-loop DCFC controller is illustrated through a simulation study.

Keywords: Machine Learning in modeling, estimation, and control, Automotive Systems, Modeling and Validation
Abstract: This paper presents a hybrid modeling methodology that integrates a Neural Network with a physics-based vehicle model, for estimating the range of battery electric buses. The neural network component is implemented to model the regenerative braking in the vehicle, as an accurate analytical approach is challenging due to limited knowledge of the electric drive system control, the Battery Management System and scarcity of experimental data. The proposed method leverages the strengths of hybrid modeling and allows for the Neural Network component being trained on a very small dataset and focusing on improving the regenerative breaking prediction. Comparing the predicted energy regeneration against experimental data, results show that the hybrid model outperforms the Physics Based Model in predicting the battery negative current. Specifically, the negative current errors were reduced significantly in 8 out of 9 simulated velocity profiles.

Keywords: Power and Energy Systems, Machine Learning in modeling, estimation, and control, Automotive Systems
Abstract: In urban traffic environments, driver behaviors exhibit considerable diversity in vehicle operation, encompassing a range of acceleration and braking maneuvers as well as adherence to traffic regulations, such as speed limits. It is well-established that these intrinsic driving behaviors significantly influence vehicle energy consumption. Therefore, establishing a quantitative relationship between driver behavior and energy usage is essential for identifying energy-efficient driving practices and optimizing routes within urban traffic. This study introduces a data-driven approach to predict the equivalent fuel consumption of a plug-in hybrid electric vehicle (PHEV) based on an integrated model of driver behavior and vehicle energy consumption. Unlike traditional models that provide point predictions of fuel consumption, this approach uses Conformalized Quantile Regression (CQR) to offer prediction intervals that capture the variability and uncertainty in fuel consumption. These intervals reflect changes in fuel consumption, as well as variations in driver behavior, and vehicle and route conditions. To develop this model, driver-specific data were collected through a driver-in-the-loop simulator, which tested different human drivers’ responses. The CQR model was then trained and validated using the experimental data from the driver-in-the-loop simulator, augmented by the synthetic data generated from Monte Carlo simulations conducted using a calibrated microscopic driver behavior and vehicle energy model. The CQR model was evaluated by comparing its predictions of equivalent fuel consumption intervals with those of baseline prediction interval methods that rely solely on conformal prediction.

Keywords: Automotive Systems, Modelling, Identification and Signal Processing, Chemical Process Control
Abstract: This paper presents the development and identification of a reduced-ordered linearized model for Selective Catalytic Reduction (SCR) systems with Ammonia Slip Catalysis (ASC) for aging diagnostics. The model, derived from first principles, incorporates several assumptions that are reflective of the operating conditions. It is then symbolically simplified into a transfer function matrix whose parameters are estimated using constrained least-squares. The validity of the model is subsequently confirmed through experimental data from test cell.

Keywords: Transportation Systems, Mechatronic Systems, Power and Energy Systems
Abstract: This paper presents an optimal co-design method for managing energy flow and sizing energy storage systems in heavy-duty series electric-hydraulic hybrid vehicles. Integrating hydraulic components into electrified powertrains allows efficient regenerative braking, making these vehicles attractive for heavy-duty applications. Optimizing energy management and energy storage systems concurrently is crucial for overall efficiency. Toward this end, a bi-level optimal co-design approach is proposed to determine optimal values for hydraulic accumulator volume, number of lithium-ion battery modules, and high-level energy control. Numerical validation confirms its effectiveness. The hydraulic accumulator supports the pump during high power demand, reducing battery usage and electric stress. This can potentially lead to smaller batteries, extended lifespan, or increased driving range. The Pareto front plot highlights the trade-off between driving range and cargo-carrying capacity.

Keywords: Power and Energy Systems, Machine Learning in modeling, estimation, and control, Estimation
Abstract: Efficient parameter identification of electrochemical models is crucial for accurate monitoring and control of lithium-ion cells. This process becomes challenging when applied to complex models that rely on a considerable number of interdependent parameters that affect the output response. Gradient-based and metaheuristic optimization techniques, although previously employed for this task, are limited by their lack of robustness, high computational costs, and susceptibility to local minima. In this study, Bayesian Optimization is used for tuning the dynamic parameters of an electrochemical equivalent circuit battery model (E-ECM) for a nickel-manganese-cobalt (NMC)-graphite cell. The performance of the Bayesian Optimization is compared with baseline methods based on gradient-based and metaheuristic approaches. The robustness of the parameter optimization method is tested by performing verification using an experimental drive cycle. The results indicate that Bayesian Optimization outperforms Gradient Descent and PSO optimization techniques, achieving reductions on average testing loss by 28.8% and 5.8%, respectively. Moreover, Bayesian optimization significantly reduces the variance in testing loss by 95.8% and 72.7%, respectively.

Keywords: Modelling, Identification and Signal Processing, Power and Energy Systems, Automotive Systems
Abstract: Battery-management systems require mathematical models of the battery cells that they monitor and control. Commonly, equivalent-circuit models are used. We would like to be able to determine the parameter values of their equations using simple tests and straightforward optimizations. Historically, it has appeared that nonlinear optimization is required to find the state-equation time constants. However, this paper shows that the relaxation interval following a current or power pulse provides data which can be used to find these time constants using linear methods. After finding the time constants, the remaining parameter values can also be found via linear regression. Overall, only linear algebra is used to find all of the parameter values of the equivalent circuit model. This yields fast, robust, and simple implementations, and even enables application in an embedded system, such as a battery management system, desiring to retune its model parameter values as its cell ages.

Keywords: Modelling, Identification and Signal Processing, Power and Energy Systems, Transportation Systems
Abstract: We introduce a partial-differential-equation model for rechargeable lithium-metal battery (LMB) cells whose parameter values are fully identifiable from cell-level experiments. From this model, we formulate a computationally tractable transfer-function (TF) model for use within optimization loops. A strategy is proposed for regressing the TF model to cell electrochemical impedance spectroscopy (EIS) measurements to estimate parameter values. We validate the regression using a synthetic dataset before application to a single-layer LMB pouch cell. The voltage RMSE between the fully identified model’s predictions and laboratory measurements is about 4mV for a GITT profile. We provide MATLAB code to simulate the model in COMSOL, compute cell impedance from the TF model, and perform model regression.

Keywords: Optimal Control, Nonlinear Control Systems, Estimation
Abstract: Lithium-ion (Li-ion) batteries are ubiquitous in modern energy storage systems, highlighting the critical need to comprehend and optimize their performance. Yet, battery models often exhibit poor parameter identifiability which hinders the development of effective battery management strategies and impacts their overall performance, longevity, and safety. This manuscript explores the integration of Fisher Information (FI) theory with Model Predictive Control (MPC) for battery charging. The study addresses the inherent hurdles in accurately estimating battery model parameters due to nonlinear dynamics and uncertainty. Our proposed method aims to ensure safe battery charging and enhance real-time parameter estimation capabilities by leveraging adaptive control strategies guided by FI metrics. Simulation results underscore the effectiveness of our approach in mitigating parameter identifiability issues, offering promising solutions for improving the control of batteries during safe charging process.

Keywords: Modelling, Identification and Signal Processing, Power and Energy Systems, Transportation Systems
Abstract: Battery-management systems for lithium-metal batteries, like their lithium-ion counterparts, require computationally efficient reduced-order models (ROMs) for estimating internal cell properties. In this work we use a finite-difference method to derive an enhanced single-particle model (SPMe) that characterizes both the concentration of lithium in the positive electrode and the electrolyte. Using these results, we develop approximations to the electrode overpotentials and electrolyte potentials. The ROM equations are derived for implementation as a continuous-time state-space system. We then compare the performance of the SPMe with full-order model results generated from COMSOL.

Keywords: Power and Energy Systems, Estimation, Modeling and Validation
Abstract: This paper introduces a test trajectory optimization approach for the parameterization of a machine learning-enhanced lithium-ion battery model. The model embeds a dense neural network within an equivalent circuit model (ECM) of battery dynamics. This hybridization strikes a balance between the interpretability of the ECM versus the ability of the dense neural network to provide a detailed nonlinear representation of open-circuit voltage (OCV) versus state of charge (SoC). The study's main contribution is optimizing battery cycling to reduce the extensive data requirements for hybrid model parameterization. Towards this goal, the paper performs Fisher analysis on all the model parameters simultaneously. Poor parameter identifiability is observed, particularly due to the inherent redundancy of neural network parameters. A novel optimization method is then employed to minimize the additional information needed for a well-conditioned Fisher information matrix, as opposed to directly maximizing Fisher information. The effectiveness of the proposed approach is validated through simulation, demonstrating its potential to enhance model parameterization significantly.

Keywords: Automotive Systems, Modeling and Validation, Estimation
Abstract: Models from the literature that have previously been utilized for target vehicle tracking typically contain vehicle motion equations described in inertial coordinate systems. Using these inertial models to track the trajectories of target vehicles in a rotating or accelerating sensor coordinate system, such as a coordinate frame attached to a turning ego vehicle, requires knowledge of the absolute location and accurate orientation of the ego vehicle itself in the inertial coordinate system. However, accurately obtaining these ego variables is not only challenging, but also needs expensive sensors and extra computational resources. To bypass this process in certain applications such as collision warning detection, this paper develops a novel trajectory tracking model that describes the relative motion of target vehicles in the frame attached to the ego vehicle. Unlike traditional models, this relative trajectory model is more accurate and only requires knowledge of the velocity and yaw rate of the ego vehicle, which are easier to obtain compared to position and orientation. Simulation and experimental findings demonstrate that the relative model can effectively replace traditional models in vehicle tracking, proving to be more accurate when the ego-vehicle has non-zero steering and advantageous in scenarios where vehicle position and orientation data are unavailable.

Keywords: Modelling, Identification and Signal Processing
Abstract: A model for the longitudinal and pitch dynamics of an electric unicycle (EUC) ridden by a human is derived while considering the independent pitch movements of the EUC body and that of the rider. Experimental data is collected for multiple maneuvers through a high-precision motion capture system and the processed data is used for parameter identification. It is demonstrated that the longitudinal and pitch dynamics are well captured by the model even for maneuvers which involve lateral motion.

Keywords: Human-Machine and Human-Robot Systems, Intelligent Autonomous Vehicles, Automotive Systems
Abstract: We derive and validate a generalization of the two-point visual control model, an accepted cognitive science model for human steering behavior. The generalized model is needed as current steering models are either insufficiently accurate or too complex for online state estimation. We demonstrate that the generalized model replicates specific human steering behavior with high precision (85% reduction in modeling error) and integrate this model into a human-as-advisor framework where human steering inputs are used for state estimation. As a benchmark study, we use this framework to decipher ambiguous lane markings represented by biased lateral position measurements. We demonstrate that, with the generalized model, the state estimator can accurately estimate the true vehicle state, providing lateral state estimates with under 0.15 m error across participants. However, without the generalized model, the estimator cannot accurately estimate the vehicle's lateral state.

Keywords: Intelligent Autonomous Vehicles, Machine Learning in modeling, estimation, and control, Path Planning and Motion Control
Abstract: We develop a robust reinforcement learning-based planning approach that handles uncertain observations and occlusions in driving scenes while meeting safety-critical constraints with a specified confidence level. We use noise models to simulate observation uncertainty and ray-tracing for occlusion modeling. The resulting uncertainty in the state impacts the derived stochastic approximation of the responsibility-sensitive safety constraints, which ensures desirable driving behaviors. The performance of two uncertainty-aware approaches are compared with an uncertainty-unaware baseline. The results demonstrate the effectiveness of the uncertainty-aware approach, which safely completes the driving task in 99-100% of cases while the uncertainty-unaware policy fails to do so in 42-57% cases.

Keywords: Automotive Systems, Human-Machine and Human-Robot Systems
Abstract: One-pedal driving (OPD) for electric vehicles (EVs) has undergone numerous studies in recent years for its longitudinal benefits in eco-driving and smooth vehicle control. In this study, the authors delve into how these longitudinal benefits may also improve lateral vehicle stability in certain cases (e.g., lane changing or hairpin turn). An experimental analysis was conducted over different test scenarios in a controlled environment in order to compare lateral stability of a battery electric vehicle (BEV) in OPD and two-pedal driving (TPD). Comparisons of the two driving methods via inertial measurement unit data showed a significant (21.5%) improvement in peak lateral acceleration in certain scenarios for OPD over TPD, potentially due to the decreased peak longitudinal acceleration (39.1%) caused by smoother braking in OPD cases directly before cornering.

Keywords: Automotive Systems, Control Applications
Abstract: This paper presents the design, modeling, and control of a hardware-in-the-loop (HIL) testbed for off-road vehicles. The proposed HIL testbed employs a transient hydrostatic dynamometer to load a diesel engine to emulate any driving cycles of a wheel loader, which is a representative off-road vehicle. A fully validated wheel loader model is used to calculate the engine load including both the drive and work functions. The developed HIL testbed is used to demonstrate more than 26% energy benefits of automated wheel loaders through systematic optimization compared with human-operated wheel loaders.

Keywords: Motion and Vibration Control, Modelling, Identification and Signal Processing
Abstract: The development of novel elastoacoustic devices, such as diodes and logic gates, relies heavily on the capability of mechanical metamaterials to achieve irreversible modal energy transfer between frequency bands and within the excited band itself. However, the energy transfer mechanism, attributed to nonlinear energy scattering within the medium, is typically limited to transferring energy from low to high frequencies and requires excitation within a narrow frequency band. In this work, we aim to overcome these limitations by investigating a 1D multiple local resonator metamaterials with local vibro-impact nonlinearity. These multiple resonators give rise to multiple optical bands, each corresponding to a resonator being out-of-phase with the holding cell. Hence, when excited within a specific optical band, the local resonator can trigger the vibro-impact nonlinearity, enabling a broader range of excitation for energy scattering. Numerical simulations have revealed considerable energy scattering at moderate input wave amplitudes. The scattered energy is subsequently postprocessed to analyze its frequency-wavenumber characteristics. This analysis highlights the system's capability to transfer energy not only from low to high frequency bands but also from high to low frequency bands, in addition to scattering within the excited band itself. The scattered energy is then decomposed into different frequency bands to further demonstrate irreversible energy transfer and quantify the energy scattered in each mode.

Keywords: Control Applications, Linear Control Systems, Estimation
Abstract: This paper presents the development of a new collaborative road profile estimation and active suspension control framework in connected vehicles, where participating vehicles iteratively refine the road profile estimation and enhance suspension control performance through an iterative learning scheme. Specifically, we develop a robust iterative learning approach to tackle the heterogeneity and model uncertainties in participating vehicles, which are important for practical implementations. In addition, the framework can be adopted as an add-on system to augment existing suspension control schemes. Comprehensive numerical studies are performed to evaluate and validate the proposed framework.

Keywords: Modeling and Validation, Motion and Vibration Control
Abstract: We investigate the nonlinear vibration behavior of a shape-morphing cantilever plate excited by base acceleration and shape-morphing deformation, imposed by a periodic moment on the sides of the plate. The interplay of shape-morphing and base excitation causes the system to demonstrate distinctive and tunable nonlinear behavior. We present frequency responses based on a finite element parametric study of the actuation parameters, and propose a minimal modeling of the system based on the Duffing oscillator. This modeling is related to the physical actuation for analysis of nonlinear curvature-based tunable systems and could be used in future design scenarios.

Keywords: Adaptive and Learning Systems, Control Design, Machine Learning in modeling, estimation, and control
Abstract: This paper presents a novel suspension control framework for enhancing performance and safety of ground vehicles operating on rough terrain. To address the adverse effects of road disturbances on ride comfort, road handling, and safety, the proposed control scheme combines learning-based model predictive control (MPC) with road preview information. By generating a predictive window of states and proactively adapting to changing vehicle dynamics and upcoming road conditions, this approach seeks to optimally control the vehicle’s suspension system in real-time. Additionally, Gaussian process regression is employed to account for unmodeled suspension system dynamics, reducing modeling uncertainties in the MPC controller. Finally, Simulation results show the effectiveness of the proposed control scheme.

Keywords: Optimal Control, Modelling, Identification and Signal Processing, Control Applications
Abstract: A novel method for resonator characterization is presented, which uses frequency-modulated excitation, phase measurement, and extremum-seeking control. The controller locks to the resonant frequency of the device and calculates its quality factor based on estimates of the local phase gradient with respect to frequency. The controller is demonstrated in simulation and implemented experimentally on FPGA-based hardware to accurately characterize a 16 MHz quartz MEMS resonator with a Q on the order of 10^4, with response times on the order of seconds.

Keywords: Mechatronic Systems, Estimation, Modelling, Identification and Signal Processing
Abstract: This paper presents a closed-loop system diagnostic based on real-time estimation and monitoring of an inverse frequency response function for fault detection and isolation in an electrohydraulic actuator. Firstly, an adaptive Kalman filter is incorporated into an indirect two-stage inverse model estimation scheme to prevent covariance windup and enhance noise immunity. Secondly, rather than using estimated inverse model parameters directly, an inverse frequency response function is computed using the estimated parameters and monitored in real-time as a diagnostic residual, which is essential for the proposed diagnostic. Numerical validation with an electrohydraulic actuator demonstrates the robust fault-tracking performance, enabling robust fault detection and isolation.

Keywords: Uncertain Systems and Robust Control, Control Design, Robotics
Abstract: The paper presents a robust control technique that combines the Control Lyapunov function and Hamilton-Jacobi Reachability to compute a controller and its Region of Attraction (ROA). The Control Lyapunov function uses a linear system model with an assumed additive uncertainty to calculate a control gain and the level sets of the ROA as a function of the worst-case uncertainty. Next, Hamilton-Jacobi reachability uses the nonlinear model with the modeled uncertainty, which need not be additive, to compute the backward reachable set (BRS). Finally, by juxtaposing the level sets of the ROA with BRS, we can calculate the worst-case additive disturbance and the ROA of the nonlinear model. We illustrate our approach on a 2D quadcopter tracking a trajectory in the presence of disturbances and a 2D quadruped achieving height and velocity regulation in the presence of added mass.

Keywords: Aerospace, Control Design, Optimal Control
Abstract: This paper analyzes the suboptimal implementation of Shrinking Horizon Model Predictive Control (SHMPC) when a fixed number of solver iterations and a warm-start are utilized at each time step to solve the underlying Optimal Control Problem (OCP). We derive bounds on the loss of performance (regret) and on the difference between suboptimal SHMPC and optimal solutions. This analysis provides insights and practical guidelines for the implementation of SHMPC under computational limitations. A numerical example of axisymmetric spacecraft spin stabilization is reported. The suboptimal implementation of SHMPC is shown to be capable of steering the system from an initial state into a known terminal set while satisfying control constraints.

Keywords: Control Design, Linear Control Systems
Abstract: This paper presents sampled-data control laws for the global stabilization of linear systems subject to time-varying, arbitrary-tight, and one-sided control constraints. The feedback laws are based on the discrete-time variational system and exploit a quadratic Lyapunov function. Sufficient conditions for achieving global closed-loop stability under control constraints, which could be time-varying or one-sided, are presented, and numerical examples are reported.

Keywords: Adaptive and Learning Systems, Nonlinear Control Systems, Control Design
Abstract: In conventional immersion and invariance (I&I) adaptive control design, control parameter adaptation is typically linear with respect to the parameter-error-induced perturbation, resulting in quadratic-rate dissipation of energy associated with the off-the-manifold variable. Departing from such a convention, this article contributes a novel strategy - polynomial adaptation. As the name suggests, control parameter adaptation in this approach takes the form of a general polynomial in relation to the perturbation. Accordingly, this new design induces a polynomial-rate energy dissipation, which is faster than the quadratic one in the conventional scheme, thereby enhancing the closed-loop control performance. The theoretical underpinnings of the new approach are demonstrated through the design of an I&I adaptive tracking control law for a general nth-order, single-input-single-output, parametrically uncertain, nonlinear system in the controllable canonical form. Additionally, a numerical study of the proposed method on the second-order forced Duffing oscillator showcases its improved transient performance in comparison to a baseline controller developed with the standard I&I adaptive control technique.

Keywords: Modelling, Identification and Signal Processing, Estimation, Uncertain Systems and Robust Control
Abstract: White noise is a popular input in system identification, but it lacks the desirable property of parameterization invariance; when changing variables for the parameter and input, the transformed input distribution is generally no longer white noise. We formally define parameterization-invariance using diffeomorphism groups in the space of parameter-input pairs, and in certain cases construct invariant measures inspired by the Jeffreys prior. This view of random input connects disparate intuitions about identifiability, controllability, and the concentration of measure phenomenon.

Keywords: Machine Learning in modeling, estimation, and control, Nonlinear Control Systems, Modeling and Validation
Abstract: We propose to estimate the region of attraction (ROA) for the stability of nonlinear systems from only system measurement data and without knowledge of the system model. The key to our result is the use of Koopman operator theory to approximate the nonlinear dynamics in linear coordinates. This approximation is typically more accurate than the traditional Jacobian-based linearization method. We then employ the Extended Dynamic Mode Decomposition (EDMD) method to estimate the linear approximation of the system through data. This is then used to construct a Lyapunov function that helps estimate the ROA. However, this estimate is typically very conservative. The trajectory reversing method is then used on the set of points that form this conservative estimate, to enlarge the ROA approximation. The output of EDMD is also utilized in the trajectory reversing method, keeping the entire analysis data-driven. Finally, an example is used to show the accuracy of this data-driven method, despite not knowing the system.