Keywords: Active diagnosis, test selection, FDI for linear systems
Abstract: This paper analyzes the injection of finite sequences as a physical watermarking method for the detection of sensor replay attacks under an unknown but zonotopically bounded uncertainty description. A temporal mismatch between the watermark sequences that are injected during the record and replay phases of the attack is achieved by splitting the input space into a pre-established number of sets. These sets are designed offline by solving an optimization problem whose solution guarantees that, in each set, there exists a finite sequence that ensures attack detection for all possible recorded watermarks belonging to any of the remaining sets. Additionally, several approaches for reducing the complexity of the optimization problem are discussed. The results are validated in simulation using a four-tanks system.

Keywords: Active diagnosis, test selection, FDI for linear systems, Filtering and change detection
Abstract: A double residual generator for fault detection in closed-loop systems is presented in this paper. The residual generator gives the standard residual vector and a dual residual vector as outputs. The standard residual vector can be applied directly for fault detection. The dual residual vector depends on known inputs to the system e.g. reference inputs and auxiliary inputs but decoupled from the disturbance in the system. It is also an input vector that can be applied for active fault detection (AFD). The dual residual vector can be considered as a generalized input vector in connection with AFD in closed-loop systems. This gives the possibility to use all known inputs for detection. The dual residual vector gives a link between passive and AFD in closed-loop systems. The double residual generator is given directly as the left Bezout matrix from the Bezout equation.

Keywords: Active diagnosis, test selection, FDI for hybrid systems, Mechanical and electro-mechanical applications
Abstract: The condition of systems, such as production equipment, typically deteriorates over time, increasing the risk of failure and associated unscheduled downtime. Predictive maintenance is a strategy to prevent failure, while maximizing the life cycle of equipment within a system. This paper contributes in this context to the theory of real-time fault diagnosis with an approach that can jointly estimate additive and parametric faults. The proposed fault diagnosis system consists of detection filters which are complemented with a residual evaluator, enabling effective fault isolation and fault estimation for open-loop and closed-loop controlled systems. The effectiveness of this unified approach is illustrated on a mass-spring-damper system.

Keywords: Active diagnosis, test selection, Reconfigurable control, sensor and actuator faults, Filtering and change detection
Abstract: Unmanned underwater vehicles operate in unstructured and harsh environments, and fault diagnosis together with fault tolerant control play a crucial role to prevent the mission from failing. When a fault occurs in presence of unknown disturbances, such as unknown marine current or hitting an obstacle, actuator fault diagnosis becomes more challenging, because the unexpected forces that arise from disturbances may be mistaken for actuator faults. In this paper, we propose an active fault diagnosis method to discern between an actuator fault and any other disturbance: the former is modelled as a multiplicative input fault, and the latter is an unknown, additive, time varying force. Once the actuator faults have been isolated and identified, their estimations are fed to the control allocation algorithm to perform fault tolerant control allocation. Simulation results are performed with a realistic non-linear model to show the effectiveness of the proposed strategy.

Keywords: Active diagnosis, test selection, FDI for linear systems, FDI theory for networked systems
Abstract: The problem of detecting replay attacks in linear time-invariant discrete-time systems is considered in this paper. In the same spirit of watermarking techniques that apply a distinctive signature to the plant's signals, we propose an event-triggered control scheme, that is purposely designed to generate a unique sequence of switching intervals, by computing an appropriate input value to be held constant while the communication is not triggered. We provide a detailed undetectability characterization in the time domain and design a controller that achieves the desired behavior. Our proposed method results in a control scheme that makes it hard for an attacker to satisfy undetectability conditions, and, as a result, a standard observer-based residual generator can be employed to reveal replay attacks. We finally validate the method using a numerical example.

Keywords: FDI for nonlinear Systems, Applications and fault tolerant control, reconfigurable control, Reconfigurable control, sensor and actuator faults
Abstract: In this paper seminal approaches for FDI introduced by Frank (Frank, 1990) are revisited in a bilinear setting. The proposed developments, including Fault Tolerant Control (Noura et al, 2000) are applied to Building Heating Ventilation and Air Conditioning (HVAC) systems. That is, the control system parameters and objective functions are adapted/reconfigured in the presence of a fault or performance deviation by means of an intermediate reconfigurable control layer. It allows maintaining building and HVAC operation within its specified energy and comfort performance requirements when a mechanical or operational fault takes place, until the fault is corrected. An integrated design, composed of two levels, respectively fault diagnosis and reconfiguration mechanism is proposed to recover performances after fault occurrence. This approach is applied to a two-zones building and simulation results are given to show its effectiveness.

Keywords: FDI for nonlinear Systems, Computational methods for FDI, FDI for robust nonlinear systems
Abstract: Li-ion batteries suffer from reliability issues due to thermal instability. For this reason, it is important to design suitable Battery Management Systems (BMSs) able to enhance safety and guarantee high battery performance. In this paper we address the problem of detection of thermal faults within a Li-ion cell using a set-based fault detection scheme where unknown but bounded uncertainties are considered. In particular, an equivalent circuit model (ECM) is used for state estimation taking into account bounded parametric uncertainties and measurement noise. The proposed method relies on constrained zonotopes (CZ), a set representation useful for performing standard operations with a lower computational effort with respect to polytopes. Numerical simulations highlight the effectiveness of the proposed approach in detecting thermal faults at an early stage and show the benefits when compared to an interval based method relying on inclusion functions and constraint propagation.

Keywords: FDI for nonlinear Systems, Mechanical and electro-mechanical applications, Structural analysis and residual evaluation methods
Abstract: In this paper, a universal residual generator is proposed for nonlinear Euler-Lagrange systems to detect sensor faults. The proposed residual generator is characterized by a simple structure, easy implementation, low requirement on model knowledge of the nonlinear system and strong robustness to unknown disturbances and unmodelled dynamics. The key idea is to apply the time-delayed estimation technique, which was originally developed in the framework of controller design for robot manipulators. By the time-delayed estimation, the measurement information at the last time instant is used to estimate all the unknown factors in the system at the current time instant. A simulation example of the well-known industrial robot PUMA560 is used to illustrate the proposed approach and show the effectiveness of the proposed universal residual generator.

Keywords: FDI for nonlinear Systems, Signal and identification-based methods, Computational methods for FDI
Abstract: Heating, Ventilation and Air-Conditioning (HVAC) systems are critical components that consume a significant percentage of the energy in the building sector. Faults in the sensing and actuation equipment of HVAC systems can create uncomfortable indoor conditions, as well as cause significant waste of energy. State-of-the-art fault detection techniques typically require control input data that maybe inaccessible in practice, since HVAC systems in large-scale buildings are controlled and monitored by proprietary Building Management Systems (BMS). This paper proposes a learning-based fault detection approach for closed-loop HVAC systems with inaccessible control information. The learning-based adaptive estimation scheme aims to estimate the temperatures within each zone-air handling unit pair and the unknown control gains that have been selected by the manufacturer. The proposed scheme enables the design of less conservative detection thresholds, which enhances the performance of the fault detection procedure for the closed-loop HVAC system. Simulation results illustrate the effectiveness of the proposed method in a large-scale HVAC system using the EnergyPlus software.

Keywords: FDI for nonlinear Systems
Abstract: This paper proposes a distributed model-based methodology for the detection and isolation of sensor faults in marine fuel engines. The proposed method considers a Mean Value First Principle model and a wide selection of heterogeneous sensors for monitoring the engine components. The detection of faults is realised based on residuals generated using nonlinear Differential Algebraic estimators combined with adaptive thresholds. The isolation of faults is, then, realised in two levels; local sensor fault detection and isolation agents are designed to monitor specific sensor sets and aim to detect faults in these sets; and a global decision logic is designed to isolate multiple sensor faults that may be propagated between the local monitoring agents. Finally, simulation results are used to illustrate the application of this method and its efficiency.

Keywords: Filtering and estimation
Abstract: The basal metabolic rate characterizes the energy consumption of the human body at rest. It can be estimated by using respiratory gas exchange analyzers and indirect calorimetry. During cycling this value could vary due to several adaptation processes such as an increase in breathing and blood circulation, controlling body temperature, among others. In this paper we propose to reconstruct the instantaneous value of this basal metabolic rate, referred here as basal power (measured in Watts), by using a robust discrete-time PI observer. The observer design is based on a solution of linear matrix inequalities and uses an uncertain linear parameter varying model of the gas exchange dynamics. The proposed methodology allows the reconstruction of the basal power while making the respiratory gas exchange estimation robust to bounded uncertainties and disturbances. The PI observer has been validated in simulation.

Keywords: Filtering and change detection, Signal and identification-based methods, Filtering and estimation
Abstract: Many practical problems of on-line fault detection in complex technical and cyber-physical systems can be reduced to the sequential reliable detection of transient changes. The goal of this paper is to continue the study of the transient change detection under less restrictive hypotheses than in previous publications. Specifically, it is assumed now that the fault dynamic profile can be arbitrary. Moreover, the new upper bounds for the probabilities of false alarm and missed detection are obtained under less restrictive hypotheses for the centralized and decentralized Finite Moving Average (FMA) test.

Keywords: Signal and identification-based methods, Active diagnosis, test selection
Abstract: For decades, manufacturers have been collecting and storing high amounts of data with the aim of better controlling and managing their processes. With the vast amount of information and hidden knowledge in all of these data, the challenge for these manufacturers to monitor their equipment units, is the extraction of an appropriate health indicator from these data that illustrates the actual state of their equipment units. In this paper, we are interested in extracting the health indicator of semiconductor equipment where manufacturing is performed by batch. For that, a novel automatic approach named Significant Points combined to the Least Absolute Shrinkage and Selection Operator (SP-LASSO) is proposed. This approach is mainly based on LASSO regression model. Its accuracy is illustrated by numerical application on simulated data.

Keywords: Graph-theoretical methods, process engineering examples, Signal and identification-based methods
Abstract: Quality-relevant root cause diagnosis is essential for the quality improvement and maintenance of dynamic processes. However, the traditional dynamic latent variable (DLV) modeling methods are mainly unsupervised ones that extract dynamic relations from one dataset (process data only). In this paper, in order to extract latent dynamics between two datasets (process data and quality data), a multi-layer DLV based quality anomaly online monitoring and root cause diagnosis method is proposed. A solution of dynamic inner CCA for modeling two datasets is provided, then quality-relevant dynamic variations, process residuals, and quality residuals are isolated. The dynamic variations are subsequently decomposed to dynamic and static ones to form a clear decomposition. Based on these decompositions, a multi-layer DLV-based quality-relevant fault monitoring method is proposed. Then, a contribution plot in the MLDLV framework is defined to diagnose the possible quality-relevant faulty candidates that are used in the subsequent transfer entropy-based root cause diagnosis. Finally, the experimental results on the Tennessee Eastman benchmark demonstrate the effectiveness of the proposed method.

Keywords: Filtering and change detection, Filtering and estimation, Signal and identification-based methods
Abstract: Abstract: This paper is devoted to the problem of increasing the reliability of monitoring and localization of violations in inertial satellite navigation system. Proposed solutions to the problem are based on the decomposition of diagnostic models of such systems and the use of combined statistical criteria. The results of natural and seminatural experiments with the integrated inertial satellite navigation system are presented and analyzed.

Keywords: Applications and fault tolerant control, reconfigurable control, FDI by means of structural properties, Artificial Intelligence methods, Controller reconfiguration, networked systems
Abstract: Offshore wind turbine (WT) rotors are subjected to asymmetrical loads, resulting in enhanced fatigue of the blade structures, which requires reliable load mitigation techniques (e.g. individual pitch control, IPC). The IPC-enhanced pitch motion increases the occurrence of pitch actuator faults. This phenomenon has accelerated the emergence of new research areas combining IPC with the fault tolerant control (FTC)-based fault compensation. The research considering the IPC and pitch system-related FTC is referred to as a fault-tolerant IPC “co-design” scheme. However, it is important to enhance the robustness of this scheme by careful design of the IPC system. The robustness is tuned using a Bayesian optimization-based Proportional-Integral PI control. The Bayesian optimization algorithm is adopted to search for optimal controller coefficients by evaluating the Gaussian process model between the designed objective function and the coefficients. The pitch actuator faults are estimated and compensated by the robust unknown input observer UIO-based FTC strategy. The robustness and effectiveness of this “co-design” scheme are verified using the 5MW NREL FAST WT with Monte Carlo simulations.

Keywords: Computational intelligence methods, Fault accommodation, Reconfiguration strategy, Applications and fault tolerant control, reconfigurable control
Abstract: To enhance both the safety and the efficiency of offshore wind park systems, faults must be accommodated in their earlier occurrence, in order to avoid costly unplanned maintenance. Therefore, this paper aims at implementing a fault tolerant control strategy by means of a data--driven approach relying on fuzzy logic. In particular, fuzzy modelling is considered here as it enables to approximate unknown nonlinear relations, while managing uncertain measurements and disturbance. On the other hand, the model of the fuzzy controller is directly estimated from the input--output signals acquired from the wind farm system, with fault tolerant capabilities. In general, the use of purely nonlinear relations and analytic methods would require more complex design tools. The design is therefore enhanced by the use of fuzzy model prototypes obtained via a data--driven approach, thus representing the key point if real--time solutions have to implement the proposed fault tolerant control strategy. Finally, a high--fidelity simulator relying on a hardware--in--the-loop tool is exploited to verify and validate the reliability and robustness characteristics of the developed methodology also for on--line and more realistic implementations.

Keywords: Mechanical and electro-mechanical applications, Filtering and estimation, FDI for nonlinear Systems
Abstract: This work analyzes the effect of estimating the excitation force in a model-based fault diagnosis (FD) of a wave energy converter (WEC). This force is a non-measurable variable related to the elevation of the sea waves at the location of the floater through a non-causal kernel function. Hence, two strategies are studied here: real-time estimates from predicted wave elevation and smoothed estimates from available measurements. To compare the performance of FD schemes based on each estimate, one considers faults in the damping subsystems of a WEC based on the Archimedes wave swing prototype. The FD scheme is also composed of an unknown input observer for fault detection and an estimator of the magnitude of the faults. Numerical simulations with irregular sea waves show improved diagnosis with smoothed estimates of the excitation force, at the expense of introducing some time delay.

Keywords: Interval methods, hydraulic systems, Power plants and power systems
Abstract: This paper presents a model-based prognostics procedure using a zonotopic Kalman filter (ZKF) in tandem with a zonotopic set based propagation of degradation, aiding in the quantification of uncertainties associated with prognostics. The prognostic procedure is then applied to the degradation of a wind turbine blade material subjected to a forecasted bounded set description of wind profile. To facilitate an online condition based implementation, an otherwise nonlinear based Kalman filter from the nonlinear wind turbine model is presented in a pseudo-linear form, a polytopic linear parameter varying (LPV) representation, decreasing computational cost and easing in the propagation of the positive invariant zonotopic uncertainty sets to a reachable set that triggers an end of life (EOL). Using this information of health, the remaining useful life (RUL) with its associated uncertainties can be predicted.

Keywords: AI and FDI methods, Signal processing for FDI
Abstract: A normal behavior model based on a boosted stacked regressor, trained in a k-fold cross-validation with optimal power curve data, is proposed for wind turbine fault detection. In order to obtain the optimal power curve data, a signal processing scheme based on density-based spatial clustering of applications with noise, along with a robust estimation algorithm are employed, from which an upper-lower bound envelop is established. The experimental results based on real wind turbine data from supervisory control and data acquisition system indicate the effectiveness and impact of the proposed method in practical applications.

Keywords: Computational methods for FDI, Active diagnosis, test selection
Abstract: In this paper we present a novel switching function for multiplicative watermarking systems. The switching function is based on the algebraic structure of elliptic curves over finite fields. The resulting function allows for both watermarking generator and remover to define appropriate system parameters, sharing only limited information, namely a private key. We prove that the resulting watermarking parameters lead to a stable watermarking scheme.

Keywords: Active diagnosis, test selection, FDI for linear systems, Filtering and estimation
Abstract: The interconnectedness of the CPS allows sharing data with relevant systems and taking more informed process control decisions; however, it also broadens the attack surface. One pain point of IoT, technology usually applied in CPS, is that embedded devices are usually low energy and do not have enough processing power to operate while keeping up with security measures, such as encryption and data validation. As a result, attackers may use such devices to launch attacks on the network, compromising the whole CPS infrastructure. One common type of attack is the False Data Injection (FDI), in which the attacker has access to a communication channel and can change the value read by a sensor or sent to an actuator. One way of coping with such attacks is to develop a bank of observer by using Functional Observers, which are better suited for high-dimensional, sparse systems. It does not suffer from sparsity's numerical problems and uses a reduced-order system to observe only the desired states, significantly reducing the computational burden of the observer. We propose an LMI-based approach to design a bank of residual generators for functional observers to detect such attacks. This approach has the advantage of using a reduced order arbitrary dynamic system, making it suitable for large-scale smart grids, and the use of LMI, allowing the easy insertion of restrictions.

Keywords: Active diagnosis, test selection, Filtering and estimation, Diagnostic methods based on qualitative models
Abstract: The paper puts forward a method for the detection of hydrothermal aging of selective catalytic reduction (SCR) units. First, a dynamic model of SCR performance including heat transfer, ammonia adsorption, NOx reduction, and ammonia oxidation is developed. The model then is parameterized and tuned with respect to published experimental data of two SCR units before and after hydrothermal aging. Such tuning also determines the target parameters of the SCR dynamic model most relevant to hydrothermal aging. The constant terms in SCR reaction and ammonia oxidation rates are the identified target parameters. Next, a model-based detection algorithm is proposed which is based on non-linear parameter estimation techniques. The performance of the proposed detection algorithm is tested via numerical simulation scenarios generated by high fidelity dynamic model of a cogeneration plant running on fast pyrolysis bio-oil. Satisfactory performance and robustness of fault detection are illustrated when it is subject to measurement noise and varying operating conditions

Keywords: Active diagnosis, test selection
Abstract: This paper proposes a sensor additive switching watermark methodology for detecting physical faults and replay attacks, and then identifying the occurring threat type: either the physical fault or the replay attack. The sensor watermark methodology includes a switching watermark generator in the plant side and a switching watermark remover in the control and monitoring side. The switching protocols of the generator and the remover and their common watermark seeds are specifically designed, guaranteeing the switch synchronization in both the nominal and fault cases, and allowing the switch asynchronization in the replay attack case. The latter is used for discriminating the two considered threat cases. The detectability and discrimination ability of the proposed watermark approach, characterizing the class of attacks and faults that can be detected and discriminated, is rigorously investigated. A simulation example is presented to illustrate the effectiveness of the proposed approach.

Keywords: Active diagnosis, test selection, Diagnostic methods based on qualitative models
Abstract: The performance improvement of alarm systems helps enhance control, operator effectiveness, facility up-times, safety, and eventually reduce losses. During the facility's operation, a large amount of alarm system data are collected and historized. Valuable information can be extracted from the alarms history and used to enrich the facility's control and performance, which prompted the development of alarm management methods and resulted in a need to transform such information into useful visual forms. These visual forms need to help facility operators and engineers understand alarm system behavior and facilitate decision making. This paper presents a framework to monitor and evaluate the alarm system's performance in a semiconductor manufacturing facility. We provide a mathematical formulation of elaborated calculations in this context. We also demonstrate the proposed framework's effectiveness using a real dataset issued from the ST-Rousset semiconductor manufacturing facility

Keywords: FDI for robust nonlinear systems, AI and FDI methods, Applications and fault tolerant control, reconfigurable control
Abstract: The interest towards renewable energy resources is increasing, and in particular it concerns wind and hydro powers, where the key point regards their efficient conversion into electric energy. To this end, control techniques can be used to meet this purpose, especially the ones relying on fuzzy models, due to their capabilities to manage nonlinear dynamic processes working in different conditions, and affected by faults, measurement errors, uncertainty and disturbances. The design methods addressed in this paper were already developed and validated for wind turbine plants, and important results can be achieved from their appropriate design and application to hydroelectric plants. This is the key issue of the paper, which recalls some considerations on the proposed solutions, as well as their validation to these energy conversion systems. Note that works available in the related literature that consider both wind and hydraulic energy conversion systems investigate a limited number of common issues, thus leading to little exchange opportunities and reduced common research aspects. Another important point addressed in the paper is that the proposed control design solutions are able to take into account the different working conditions of these power plants. Moreover, faults, uncertainty, disturbance and model--reality mismatch effects are also considered when analyzing the reliability and robustness features of the derived control schemes. To this end, proper hardware--in--the--loop tools are considered to verify and validate the developed control schemes in more realistic environments.

Keywords: Graph-theoretical methods, process engineering examples, FDI for nonlinear Systems
Abstract: In this paper we extend the work presented in cite{rathore2021leakage} to leakage localization in water distribution networks with multiple inlets. A self-adaptive reduced order model is used to estimate network pressures under nominal condition. These estimated pressures are compared to the measured network pressure to generate pressure residuals. Further, the pressure residuals are compared to expected residual signatures for leakage localization. For the reduced order model to be valid for water distribution network with multiple inlets, a control requirement is to be placed on the network. In this paper we also present the test results from a laboratory setup, which is present at Aalborg University, Denmark, to demonstrate the localization framework.

Keywords: FDI for nonlinear Systems, Graph-theoretical methods, process engineering examples, FDI for robust nonlinear systems
Abstract: We develop a novel fault diagnosis method that combines information entropy with transfer entropy for chemical processes. On the basis of the correlation information entropy, an online process monitoring model for the abnormal conditions of chemical processes is first established. Then, the combination of information entropy and transfer entropy is used for fault diagnosis. We apply the generalized correlation coefficient of mutual information in information entropy in extracting the condition feature of process variables. Subsequently, transfer entropy analysis is performed to obtain the fault chain. The database between the fault feature variables and the root cause is established. Finally, the Tennessee-Eastman (TE) chemical process is taken as a case study.

Keywords: Signal and identification-based methods, Fault accommodation, Reconfiguration strategy, Advanced condition monitoring systems for railway vehicles
Abstract: Unsupervised and autonomous operations require new, more efficient, and smarter technical solutions. Digitalization is one of the enablers of autonomous operation, with installing new sensors and using the data for AI and machine learning. The trend is to move the handling the vast amount of data from sensors, installed in machines, and devices, close to the sensors. This way, data communication can be heavily reduced, while the latency of decisions based on data can be minimal. This is achieved by moving the advanced data analysis capabilities from the cloud to the edge. In this paper, the driver of technological development is the concept of an unmanned machine room, where the normal inspections performed by personnel are automated using sensors and algorithms. We present an analysis of vibration data, recorded at a cruise ferry’s engines, using a signal processing method called cepstrums. A cepstrum is a version of frequency analysis, intending to identify and visualize periodic structures in data using cepstrograms. Compared to spectrogram analysis, usage of cepstrogram enables better visualization of engine behavior into the shape of audio signals. This helped us discover the run-time vibration routine of the engines from where we could observe unusual vibration sensing generated from the engine. We started studying the unusual behavior and came to the conclusion that the engine is sometimes misfiring the cylinders, generating anomalies.

Keywords: Signal and identification-based methods, FDI for nonlinear Systems
Abstract: The paper proposes a new method for anomaly detection based on multilinear low-rank models. No a priori knowledge about the investigated system is needed for data-driven parameter identification of these models. Multilinear parameter identification is able to cover more dynamic phenomena than linear black box identification. A minimal model of rank 1 has a tiny number of parameters which is equal to the dynamic order plus the number of inputs. These multilinear parameters are moreover directly interpretable as each parameter indicates the influence of one corresponding state or input to the next state of the MTI model. As example, the method is demonstrated by an anomaly detection with real data from the HVAC system of a test room.

Keywords: Signal processing for FDI, AI and FDI methods
Abstract: This paper proposes an efficient multivariate time-series fault detection and classification approach aiming to detect faulty wafers( i.e. pieces of silicon) during semiconductor manufacturing process. This approach is based on using Independent Component Analysis (ICA) and several Machine Learning Ensemble Techniques. The main objective is to extract the most useful information from each time-series and combine them to build a set of fully concatenated features. Thereafter, Extra Trees, Random Forest, Gradient Boosting and Extreme Gradient Boosting, one of the prevalent evolutions of tree-based algorithms, are fitted to the extracted features subset to design and implement an efficient anomaly detection strategy. The obtained results show that the proposed technique is very efficient and very promising.

Keywords: Signal and identification-based methods, Active diagnosis, test selection, Controller reconfiguration, networked systems
Abstract: To detect contaminant intrusion and, in general, to assess quality problems in their water distribution systems, water utilities need quality sensors that continuously measure, directly from the network, conductivity, PH, concentration of different substances, and other related parameters. Due to the nature of the objectives involved, the decision about where to place sensors in the network and the amount of them to be installed is a very challenging problem. In this investigation, we present a multi-objective approach to cast light on those decisions. Instead of a crisp solution, the multi-objective approach will provide a wide spectrum of solutions representing the best trade-off among all the decision criteria of the problem. This approach aims to integrate the practical experience of engineers into the decision-making process since, eventually, the solution will be selected among the Pareto front of solutions using the engineers’ experience and the specific characteristics of their utility. To this end, the used algorithm adds agents based on both technical and user-preference rules on top of evolutionary search techniques to explore the decision space. The algorithm runs as a part of the Agent Swarm Optimization framework, a consolidated multi-objective software. Another novelty of this contribution is computational: the evaluation of the objective functions is executed directly in the MS SQL server and simulation data is never required to be loaded in their entirety. Without this important implementation detail, the solution for “large” water network models would not be affordable with the hardware typically used in desktop computers. To illustrate the solution process, a use case focused on a mid-size water supply network is addressed.

Keywords: Signal and identification-based methods, Dependability
Abstract: Predicting the remaining useful life (RUL) of in-service commercial plants is closely linked with safety production and maintenance cost optimization. It is noteworthy that real-life degradation processes could be affected by their past states coupled with unknown disturbances. To make a comprehensive consideration of multisource dependencies and uncertainties, we develop a generalized non-stationary, nonlinear, and non-Markovian degradation model consisting of a Gaussian disturbed drift term and a sub-fractional Brownian motion (sub-FBM) based diffusion term. Both parts are state-dependent, and reflect two different types of memory effects. The main parameters and the approximate probability density function (PDF) of RUL can be solved on foundation of the Markovian transformation theories. A case study finally illustrates the effectiveness of the proposed scheme.

Keywords: Applications and fault tolerant control, reconfigurable control, Advanced actuator technologies, Dependability
Abstract: Model-Free Control (MFC) approach is a novel technique to handle nonlinearities and uncertainties in control systems in order to provide enhanced performance level. The design of MFC is based on an ultra-local model, which is an approximation of the dynamics of the controlled system for a short period of time. This structure also involves a model-based robust control, which guarantees the accurate tracking performance of the closed-loop system. In this paper, a novel technique is proposed for designing the robust control for taking into account the varying characteristics of the ultra-local model. A novel modeling method is presented, which combines the original plant of the controlled system and the ultra-local model. Then, a robust control design is proposed, by which the tracking performance and the robustness of the closed-loop system can be guaranteed. The design steps and effectiveness of the proposed control strategy are demonstrated through a vehicle-oriented control problem using the high-fidelity simulation software, CarMaker.

Keywords: Applications and fault tolerant control, reconfigurable control
Abstract: This paper proposes a collision-free trajectory design for dance choreography of virtual drones using a hierarchical design structure. The trajectory on two levels is designed, i.e., on the level of individual virtual drones and on the level of their centralized coordination. The design on the drone level through reinforcement learning has been carried out, with which dance steps of virtual drones in their motion can be achieved. The coordination level contains a prediction algorithm, which guarantees the avoidance of collision, even at the faultiness of some virtual drones. The effectiveness of the hierarchical design on the example of a Viennese waltz dance choreography in augmented reality has been illustrated.

Keywords: Applications and fault tolerant control, reconfigurable control, Mechanical and electro-mechanical applications, Power plants and power systems
Abstract: This paper presents a gain-scheduling wind-turbine control strategy to mitigate the effects of adverse weather conditions on the degradation of the drive-train. By choosing a suitable control gain according to the wind class, it is possible to establish a trade-off between generated energy and drive-train degradation. The dissipated energy in the mechanical transmission is used as an indicator of degradation. The drive-train is modeled as a flexible shaft using non-linear dynamics of a mass-spring-damper system. Simulations consider a variable speed-fixed pitch turbine of 2 MW (100 m rotor diameter), with a horizontal axis and fixed gearbox. The results show that the proposed gain-scheduling control strategy maintains the desired turbine efficiency by adequately managing the drive-train degradation according to the wind turbulence conditions.

Keywords: Applications and fault tolerant control, reconfigurable control, FDI for linear systems
Abstract: In this paper, we propose a novel fault tolerant multi-sensor switching control framework for linear parameter-varying systems considering multiplicative and additive sensor faults simultaneously. We show that the closed-loop stability is preserved under sensor fault situation if established guaranteed fault detection conditions based on invariant sets are fulfilled. At each time instant, the switching control strategy selects the sensor group that provides the optimal closed-loop performance, which is characterized by a quadratic performance objective function. Furthermore, faults on each channel of all group of sensors can be also pre-guaranteed to be isolated under the fulfillment of established fault isolation conditions, which ensures a more precise determination of the location of fault occurrence. At the end, a practical vehicle model is provided to illustrate the effectiveness of the proposed method.

Keywords: Controller reconfiguration, networked systems, Applications and fault tolerant control, reconfigurable control, FDI for linear systems
Abstract: As networks become more widely used in control systems, the risk of cyber attacks increases significantly. The risk arising from stealthy attacks is that an adversary may take control of the plant through the communication network without being detected by the usual monitoring systems, with consequences that are often malicious or even destructive. This paper considers a new attack scheme called robust stealthy covert attack. Unlike the standard covert attack in Smith (2011), the robust covert attack can keep its stealthiness, even if the adversary doesn’t have complete knowledge of plant model. It will be shown that the robust covert attack can be made stealthy by applying the time delayed control technique. An example is given to illustrate the stealthiness of the proposed robust covert attack in the presence of model uncertainties.

Keywords: FDI theory for networked systems, Structural analysis and residual evaluation methods, Computational methods for FDI
Abstract: In energy or communication networks, cables are omnipresent. Like any other systems, they are subject to faults. Faults in cables are divided into two families : soft faults and hard faults. Soft faults are faults that do not affect the data and/or energy transfer. However, over time they tend to turn into hard faults that can lead to a total system breakdown. In this paper, we focus on the detection of soft faults and the localization of the faulty branch in star-shaped networks. The proposed method is based on the comparison between a network Reference Transmission Coefficient (RTC) and a TC measured online through power line communication using Orthogonal Frequency Division Multiplexing (OFDM) scheme. Our proposal is discussed for both energy and communication networks. Simulation-based results are presented that confirm the ability of the method to locate the faulty branches.

Keywords: FDI by means of structural properties, Artificial Intelligence methods, AI and FDI methods, FDI for nonlinear Systems
Abstract: With trends as IoT and increased connectivity, the availability of data is consistently increasing and its automated processing with, e.g., machine learning becomes more important. This is certainly true for the area of fault diagnostics and prognostics. However, for rare events like faults, the availability of meaningful data will stay inherently sparse making a pure data-driven approach more difficult. In this paper, the question when to use model-based, data-driven techniques, or a combined approach for fault diagnosis is discussed using real-world data of a permanent magnet synchronous machine. Key properties of the different approaches are discussed in a diagnosis context, performance quantified, and benefits of a combined approach are demonstrated.

Keywords: Petri net-based diagnosis, FDI for discrete-event systems
Abstract: This paper presents a method for the diagnosis of time patterns in time Petri nets. This method uses a characterization of the pattern called Observable Simple Temporal Network, given in the form of a set of observable events with temporal constraints on their occurrence dates. The proposed diagnoser verifies if a part of an input timed sequence of observations is consistent with the characterization. If the pattern has not occurred, this consistency test will lead to the same conclusion. One the other hand, if the pattern has occurred, the consistency test will lead to an ambiguous diagnosis in the general case or to the conclusion that the pattern has definitely occurred if the underlying system is diagnosable.

Keywords: FDI theory for networked systems, FDI for hybrid systems
Abstract: In this paper, an adaptive urban smart lighting architecture allowing municipalities to manage and control public street lighting lamps is developed. In order to reduce energy consumption, the system is designed to autonomously adapt street lamps’ brightness on the basis of the presence of vehicles in specific segments of the streets/roads of interest. To this end, a hybrid traffic model is here proposed to estimate in each segment of the road the number of vehicles crossing the segment. The estimation task is carried out by means of a bank of switching observers and corresponding sensors. One observer at a time is selected from the bank and put in order. Such a selection is based on a recently developed multi-agent reputation mechanism for a class of networked systems. As one of its merits, this methodology is capable to reveal those sensors that have the highest information contents (Quality of Service) on the basis of their reputation (trust) shared among agents.

Keywords: FDI for discrete-event systems, Computational methods for FDI
Abstract: In this paper, a diagnosis problem of discrete event systems (DESs) is considered, including fault pattern detection and diagnosability. A fault pattern in a DES is modeled by an automaton that represents the occurrence of complex faults, i.e., the language of the automaton is the objective to be diagnosed. To solve the problem of fault pattern detection, two verifiers are provided. The former, based on state isolation, can perform state estimation in an efficient manner recursively such that at any point during recursion, the states can be isolated. The latter, inspired by the notion of synchronous product, allows us to concisely synthesize and analyze the system information. Comparing these two verifiers, it is found that the two structures are equivalent from the perspective of pattern detection and diagnosability. On the basis of aforementioned verifier structures, we establish their respective diagnosers, and develop algorithms for fault pattern diagnosability.

Keywords: Power plants and power systems
Abstract: Electric vehicles are becoming more common and will soon be the norm in termsof road transportation, which justifies the interest of researches in topics related to healthmanagement and exploitation of such vehicles. So far, however, electric vehicle managementemphasizes charging strategies and routing optimization, mainly focused on energy consumptionand no investigation has been performed on the impact that routing and driving parameters,such as maximum speed and acceleration, have on the useful life of a battery and therefore onthe long term exploitation cost of a fleet of electric vehicles. This paper proposes a deterministicmethod to estimate the impact of a route and driving parameters on the state of health of abattery. This method is used to optimize routes while respecting operational constraints, suchas delay penalties, and the impacts of such optimization on the long-term cost are estimatedthrough simulations, mimicking real driving conditions and traffic randomness.

Keywords: Power plants and power systems, AI and FDI methods, FDI for nonlinear Systems
Abstract: With the advancement of new technologies, power systems are increasingly equipped with more sensors and actuators, heightening the risk of failure. This fact, together with the vulnerability of solar plants –not only to internal faults but also to the action of the sun, rain, wind, and animals, among others– gives rise to the need for detecting and identifying faults to deal with them. Methods that detect and diagnose faults play a crucial role in solar plants, allowing the systems to cope with them as soon as they occur and before they lead to large-scale problems. This work proposes using neural networks to detect and distinguish mirror and flow rate faults in a Fresnel plant. In addition, a defocusing stage is added to access hard-to-isolate faults, increasing the accuracy of 89.61% to 97.43%. These results contribute to the problem of isolability in thermal solar plants. The simulations for obtaining the neural networks and the results were conducted on a model of the Fresnel plant located at the Engineering School of Seville, Spain (ETSI).

Keywords: Power plants and power systems, Mechanical and electro-mechanical applications, Signal processing for FDI
Abstract: The paper addresses the development of a perturb and observe algorithm implemented for maximum power point tracking control of a permanent magnet synchronous generator. It is shown that this algorithm tracks the optimum operation point and provides fast response even in the presence of faults. The strategy implements the tracking algorithm by using real---time measurements, while providing maximum power to the grid without using online data training. The solution is simulated in the Matlab and Simulink to verify the effectiveness of the proposed approach when fault--free and faulty conditions are considered. The simulation results highlight efficient, intrinsic and passive fault tolerant performances of the algorithm for electric generators and converters with low inertia.

Keywords: Power plants and power systems, MPC methods, Applications and fault tolerant control, reconfigurable control
Abstract: The current energy situation and the possibility of exhausting fossil fuels in a relatively near period, have led to investing efforts in the development of techniques that use renewable energy sources for power generation. A configuration that allows renewable energy sources to be integrated into the overall power system, advocates dividing the grid into distributed systems incorporating small-scale generation and storage. Microgrids are a well-known type of these systems. Control systems help maintain the reliability of the energy supply while minimizing costs. In addition, it must be taken into account that faults can occur in the processes that make up the microgrid. In some cases, the control system can mask these faults, even allowing the fault to reach an irreparable level. In this context, fault-tolerant control is a tool that enables control objectives to be maintained even in the presence of faults. If necessary, the control objectives are adapted to the fault. Furthermore, the fault tolerant control system needs to be able to detect faults, quantify their intensity and act accordingly. In this way it is avoided that small faults, that in other circumstances would remain hidden by the control loop, cause faults of a greater magnitude. This article proposes a fault quantification method based on parity equations and structured residuals that, together with a fault accommodation tolerance mechanism, mitigates the consequences of possible faults in this type of system.

Keywords: Diagnostic methods based on qualitative models, Power plants and power systems, Dependability
Abstract: Safety analysis and risk assessment of Community Power Distribution Network (CPDN) are the key links of safe operation of distribution network system. Reliability indices based risk assessment is the standard method for power distribution network. However, reliability indices are not adequate for the CPDN’s intrinsic safety, which is represented by the states of physical components and network’s failure events. Meanwhile, the reliability indices is designed for long-term risk assessment and cannot refer to the operation scenarios of CPDN. Therefore, dynamic risk assessment by considering the intrinsic safety of CPDN is a significant and challenge issue. This study proposes a novel safety and dynamic risk assessment framework of CPDN using Bayesian Network (BN). First, based on the standard reliability indices, the failure rate of physical components in CPDN and the failure events are creatively introduced into risk assessment, then a novel risk indices hierarchy is obtained. This indices hierarchy can quantitatively present the dynamic risk of CPDN, with considering the reliability indices and the features of intrinsic safety. Second, a dynamic risk analysis method is proposed, which is based on Fault Tree (FT) and Bayesian Network (BN). A real CPDN is used as a case study to demonstrate the feasibility and effectiveness of the proposed method. The result of the case study suggests that the proposed framework could adequately assess the risk of CPDN and the derived rational safety actions can reduce the risk effectively.