Keywords: AI and FDI methods, FDI by means of structural properties, Artificial Intelligence methods
Abstract: Data-driven fault diagnosis requires training data that is representative of the different operating conditions of the system to capture its behavior. If training data is limited, one solution is to incorporate physical insights into machine learning models to improve their effectiveness. However, while previous works show the usefulness of hybrid approaches for isolation of faults, the impact of training data must be taken into consideration when drawing conclusions from data-driven residuals in a consistency-based diagnosis framework. By giving an understanding of the physical interaction between the signals, a hybrid fault diagnosis approach, can enforce model properties of residual generators to isolate faults that are not represented in training data. The objective of this work is to analyze the impact of limited training data when training neural network-based residual generators. It is also investigated how the use of structural information when selecting the network structure is a solution to limited training data and how to ameliorate the performance of hybrid approaches in face of this challenge.

Keywords: AI and FDI methods, FDI for hybrid systems
Abstract: Health monitoring approaches are usually either model-based or data-based. This article aims at using available data to learn a hybrid model in order to profit from both the data-based and model-based advantages. The hybrid model is represented under the Heterogeneous Petri Net formalism. The learning method is composed of two steps: the learning of the Discrete Event System (DES) structure using a clustering algorithm (DyClee) and the learning of the continuous system dynamics using two regression algorithms (Support Vector Regression or Random Forest Regression). The method is illustrated with an academic example.

Keywords: AI and FDI methods, Computational intelligence methods
Abstract: Machine Learning (ML) has been implemented with great successes in recent decades, both in research and in practice. In the field of Cyber-Physical Systems (CPS), ML methods are already widely used, e.g. in anomaly detection, predictive maintenance or diagnosis use cases. However, Representation Learning (RepL), which learns general concepts from data and has been a major driver of improvements, is hardly utilized so far. To be useful for CPS, RepL methods would have to produce interpretable results, work without requiring much prior knowledge, and be applicable to typical CPS datasets, including e.g. noisy sensor signals and discrete system state changes. In this paper, we provide an overview of the current state of research regarding methods for learning physical concepts in time series data, which is the primary form of sensor data of CPS. We also analyze the most important methods from the current state of the art using the example of a three-tank system. Based on concrete implementations, we discuss the advantages and disadvantages of the methods and show for which purpose and under which conditions they can be used for CPS.

Keywords: AI and FDI methods, Dependability
Abstract: The stable and efficient operation of the cold source system can significantly reduce building energy consumption and improve indoor comfort. Health monitoring of the cold source system is a necessary means to ensure such a requirement. Due to the variations of ambient temperature and cold load, the cold source system often running in multiple modes, which limits the application of conventional health monitoring methods with poor performance, e.g. high false alarm rate. Therefore, this paper uses our previously proposed method, just-time-learning aided canonical correlation analysis (JITL-CCA), to monitoring the health condition of the real cold source system. The application steps are detailed, including how to select monitoring variables and how to use reconstruction-based contribution (RBC) to assist JITL-CCA in fault diagnosis. The on-site data experiments show the applicability of the JITL-CCA method and its superior performance when comparing with the conventional CCA method.

Keywords: FDI for linear systems, Interval methods, hydraulic systems
Abstract: This paper addresses the problem of robust fault detection for Linear Parameter Varying (LPV) systems using set-based approaches. Two approaches are proposed, based respectively on set-based state and parameter estimation methods, for implementing direct and inverse test for robust fault detection (FD). The uncertainties are assumed to be unknown but bounded and their effect is propagated using zonotopic sets. These robust FD test methods aim at checking the consistency between the measured and estimated behaviour obtained from estimator in the parameter or output space considering the effect of the uncertainty. When an inconsistency is detected, a fault can be indicated. A case study based on an autonomous vehicle is employed to compare the performance of proposed FD tests.

Keywords: FDI for linear systems, Mechanical and electro-mechanical applications, Filtering and change detection
Abstract: For mechanical system structural health monitoring, a new residual generation method is proposed in this paper, inspired by a recent result on subspace system identification. It improves statistical properties of the existing subspace residual, which has been naturally derived from the standard subspace system identification method. Replacing the monitored system state-space model by the Kalman filter one-step ahead predictor is the key element of the improvement in statistical properties, as originally proposed by Verhaegen and Hansson in the design of a new subspace system identification method.

Keywords: FDI for linear systems, Structural analysis and residual evaluation methods, Signal and identification-based methods
Abstract: In this work, a probabilistic projection approach is proposed to data-driven fault detection (FD) for stochastic dynamic systems. To this end, a stable kernel representation based residual generator is first constructed with main attention to the design of a projection matrix based on system input and output data in kernel space. Concerning the practically inaccessible probability distribution for stochastic disturbance and limited priori knowledge of fault, a distributionally robust optimal FD problem is formulated in the sense of maximizing the fault detectability for an acceptable upper bound of false alarm rate, wherein the distributional profile of stochastic disturbance is characterized by a mean-covariance based ambiguity set. By means of worst-case conditional value-at-risk and singular value decomposition, an analytical solution to the targeting FD problem is derived in the probabilistic context, that achieves the best fault detectability in worst-case setting. Simultaneously, the robustness of the designed FD system against distributional uncertainties can be guaranteed. A simulation study is finally illustrated to show the applicability of the proposed method.

Keywords: Signal and identification-based methods, FDI for linear systems, Mechanical and electro-mechanical applications
Abstract: In this paper, we revisit subspace identification for wind turbines and more generally rotating periodic systems. Previous works have stressed the difficulty of modeling such systems as Linear Time Invariant and thus to apply classical Stochastic Subspace Identification. Such work plead for periodic or augmented theories. In this paper, we prove and show that a classical SSI can be applied and recover an eigenstructure information that is related to the eigenstructure of the instrumented system despite the system excitation being modeled as non-stationary.

Keywords: Active diagnosis, test selection, FDI for linear systems, FDI for nonlinear Systems
Abstract: This paper presents a new method of fault detection based on residual signal generation. Most of the existing diagnostic methods that use the residual to detect a failure are often based on the knowledge of the system model. The developed method does not require a precise knowledge or deep information about the system model. It is based on the reconstruction of the system output via an ultra-local model and a model-free controller. The reconstructed/estimated output is used to build the residual signal which is the fault indicator. Several simulation tests have been performed to evaluate the potential of the proposed approach for fault diagnosis. A fault on an actuator of the system is simulated in linear, non-linear and multi-input non-linear case studies. The simulation results reveal that the fault is successfully detected for all these systems under a noisy environment.

Keywords: Applications and fault tolerant control, reconfigurable control, Active diagnosis, test selection
Abstract: This paper proposes an active fault tolerant control (AFTC) scheme for detecting and correcting faulty measurements from the ambient temperature sensor that is utilized to calculate the optimal setpoints for a rooftop gas cooler unit operation. The faulty measurements are caused by improper placement of the ambient temperature sensor so that it is exposed to heat generated directly or indirectly by the sun. The AFTC is designed in a plug & play manner with no prior knowledge of the underlying system. The underlying system is identified by applying an online identification algorithm. To ensure the validity of the model several criteria is established. The resulting model is utilized in an observer based fault tolerant controller scheme including a bump-less implementation of the fault correction when a fault has been detected. The validity of the proposed AFTC scheme is investigated using both experimental data as well as simulation.

Keywords: Applications and fault tolerant control, reconfigurable control
Abstract: In this paper, we investigate the fault-tolerant path following control problem for an overactuated underwater vehicle subject to input saturation and rate of change constraints, actuator faults and parametric (mass and inertia) uncertainty. A robust adaptive control allocation strategy is proposed capable to reconfigure the distribution of the control effort among the remaining healthy actuators in the case some of them failed. The online estimation of actuators' effectiveness and bias fault parameters is integrated with the control allocation and reconfiguration unit. The robustness and performance of the scheme are shown by evaluating the results of simulations and pool experiments of different path-following control problems using overactuated underwater vehicles.

Keywords: Applications and fault tolerant control, reconfigurable control, FDI for robust nonlinear systems
Abstract: Robust adaptive control (RAC) approaches have many state-of-the-art capabilities; however, they cannot provide near-real-time performance, especially for novel fault conditions. To address such issues, we introduce a framework called Intelligent Robust Adaptive Control (IRAC) for switched systems, which is based on randomized blending of controllers' actions. We show how our approach provides guarantees for stability, robustness, adaptivity, and near-real-time performance. We empirically compare worst-case performance of our approach with other methods, using a ``hypothetical baseline" that takes some time for control inference but provides exact control outputs. This shows the impact on near-real-time control of the time for inference, and illustrates the benefits of IRAC.

Keywords: Applications and fault tolerant control, reconfigurable control, Fault accommodation, Reconfiguration strategy, Reconfigurable control, sensor and actuator faults
Abstract: In this paper, the design of an event-triggered fault-tolerant control for leader-following consensus in multi-agent systems subject to actuator faults is presented. The problem under consideration is to reduce the control update rate and to guarantee that all the agents follow the trajectories of a leader when one or all the agents are subject to actuators faults. The proposed fault-tolerant strategy is based on distributed virtual actuators without re-tuning the nominal leader-following control. Linear matrix inequalities-based conditions are synthesized to guarantee the stability of the synchronization error and estimation error. The efectiveness of the event-triggered fault-tolerant strategy is illustrated through numerical examples.

Keywords: Applications and fault tolerant control, reconfigurable control
Abstract: In this paper, the active fault-tolerant tracking control issue is studied for a class of discrete-time switched linear parameter varying system by using the virtual actuator approach. First, a switched gain-scheduled virtual actuator is proposed to improve the fault tolerance performance. Second, in order to obtain better tracking performance, a sub-optimal trajectory recovery target is determined by minimizing the H-infinity-norm of the virtual actuator's relevant transmission channel. By using the multiple parameter-dependent Lyapunov function approach, sufficient conditions for the tracking closed-loop system and virtual actuator system to be globally uniformly exponentially stable are derived, which also guarantee the desired H-infinity performance indices requirement. Finally, effectiveness and applicability of the developed approaches are validated via a GE-90 turbofan-engine model system.

Keywords: Mechanical and electro-mechanical applications, Signal processing for FDI, AI and FDI methods
Abstract: The fault diagnosis of wind turbines includes extremely challenging aspects that motivate the research issues considered in this paper. In particular, this work studies fault diagnosis solutions that are considered in a viable way and used as advanced techniques for condition monitoring of dynamic processes. To this end, the work proposes the design of fault diagnosis techniques that exploits the estimation of the fault by means of data--driven approaches. These fuzzy and neural network structures are integrated with auto--regressive with exogenous input regressors, thus making them able to approximate unknown nonlinear dynamic functions with arbitrary degree of accuracy. The capabilities of fault diagnosis schemes are validated by using a real--time simulator of a wind turbine system. Moreover, at this stage the benchmark is also useful to analyse the robustness and the reliability characteristics of the developed tools in the presence of model--reality mismatch and modelling error effects featured by the wind turbine simulator. This realistic simulator relies on a hardware--in--the--loop tool that is finally implemented for verifying and validating the performance of the developed fault diagnosis strategies in an actual environment.

Keywords: Reconfigurable control, sensor and actuator faults, Filtering and estimation, Applications and fault tolerant control, reconfigurable control
Abstract: Reduction of spaceflight costs calls for development of new technologies that render rockets reusable. This new requirement and the continuous improvement of rocket engines require pro-active approach towards the possibility of integrating health monitoring systems on-board. These health monitoring strategies should also take into consideration the state of degradation and the remaining useful life prediction. In this paper, an Extended Kalman Filter is used to estimate the state of health and the dynamics of the degradation, and the remaining useful life is predicted with respect to failure thresholds pre-set by the user. The first-order inverse reliability method is employed to assess the quality of the remaining useful life prediction by quantifying the associated uncertainty. The overall method is validated using simulation study involving degradation data provided by Centre national d'études spatiales (CNES) applied to liquid propulsion rocket engine (LPRE) combustion chamber.

Keywords: Power plants and power systems, FDI for linear systems
Abstract: This work deals with the problem of the diagnosis of two shunt faults in a high-voltage direct current (HVDC) transmission line modeled by a set of linear partial differential equations. Because the fault scenario of a transmission line subject to two simultaneous shunts is indistinguishable from measurements at the boundaries in steady state, a sequential fault scenario is assumed in this work. This consideration allows estimating on-line the magnitude of the shunt current and the position for a weighted average fault by using the magnitude position fault estimator (MPFE), and also isolating the faults by recording the data history of the line. To identify the sequence of faults from the error of the MPFE, a new weighted average position relation (WAPR) is derived, which is part of the main contribution. Thus, the solution consists of a combination of detection and isolation stages; the first one determines the presence of weighted average shunt faults through an MPFE. Furthermore, the second one uses historical data to calculate the physical fault parameters. Simulations for HVDC transmission line data by ATPDraw show the performance of the approach.

Keywords: AI and FDI methods, Signal processing for FDI
Abstract: In this paper, we propose a two-stage online data-based diagnostic method that detects issues in the In Circuit Test (ICT) equipment from a Surface-Mount Technology (SMT) production line. The first stage performs anomaly detection in a univariate stream of test values. The second stage achieves fault detection and isolation based on the process capability and a Gaussian mixture clustering method. The combination of the two stages allows improving the online cost of the second stage, and also improving the confidence and the interpretability of the first stage, the anomaly detection. Two solutions are compared for the first stage, an anomaly detection by Extreme Value Theory (EVT) and a sliding window method. The comparison is done with an automotive industrial database and shows that using EVT delivers almost the same performance in detection with less computation time.

Keywords: Applications and fault tolerant control, reconfigurable control, FDI for linear systems, Active diagnosis, test selection
Abstract: The paper presents further work on combining Iterative Learning Control (ILC) with Fault-Tolerant Control. Main goals here are to drive the considered plant to the prescribed reference with simultaneous elimination of the effect of possible process faults. The first result presented is about designing the fault estimator and linking it with the original plant. Next, for a such an extended system the ILC scheme is provided. Since the system is occupied by the process and measurement uncertainties, the fault observer along with the ILC controller are designed under the H infinity regime. The proposed control strategy is verified with application to the DC servo motor.

Keywords: AI and FDI methods, Computational intelligence methods, FDI for nonlinear Systems
Abstract: Clustering is an important tool in data-driven fault diagnosis to make use of unlabeled data. Collecting representative data for fault diagnosis is a difficult task since faults are rare events. In addition, using data collected from the field, e.g., logged operational data and data from different workshops about replaced components, can result in labelling uncertainties. A common approach for fault diagnosis of dynamic systems is to use residual-based features that filter out system dynamics while being sensitive to faults. The use of conventional clustering algorithms is complicated by that the distribution of residual data from one fault class varies for different realizations and system operating conditions. In this work, a clustering algorithm is proposed for residual data that clusters data by estimating fault signatures in residual space. The proposed clustering algorithm can be used on time-series data by clustering batches of data from the same fault scenario instead of clustering data sample-by-sample. The usefulness of the proposed clustering algorithm is illustrated using residual data from different fault scenarios collected from an internal combustion engine test bench.

Keywords: AI and FDI methods, FDI for nonlinear Systems, Power plants and power systems
Abstract: This work proposes a new fault diagnosis approach for a wind energy conversion (WEC) system. The proposed technique merges the benefits of feature extraction based on Gaussian Process Regression (GPR) and Multi-Class Random Forest (MCRF)-based fault classification where instances are classified into one or more classes. In the developed GPR-MCRF approach, the nonlinear statistical features including the mean vector M_GPR and the variance matrix C_GPR are computed using the GPR model with aim of extracting the most relevant features from the WEC system. Then, these features are introduced to the RF classifier for classification and diagnosis purposes. Therefore, the application of the GPR-MCRF technique for WEC systems aims to enhance the use of the classical raw data-based MCRF and diagnosis accuracy. Three kinds of faults (wear-out, open-circuit, and short-circuit faults) are considered in this work. Different case studies are investigated in order to illustrate the effectiveness and robustness of the developed technique compared to the state-of-the-art methods. The obtained results show the high diagnosis accuracy of the proposed approach (an average accuracy of 99.9%).

Keywords: AI and FDI methods, Diagnostic methods based on qualitative models, Computational intelligence methods
Abstract: The paper is aimed to discuss the problem of fault isolation robustness. According to the authors, this problem has been so far underestimated and poorly discussed in the community of process diagnostics. It was pointed out that the lack of robustness of fault isolation may be one of the main reasons limiting the broader application of advanced diagnostic systems. This was a strong motivation to undertake the research and evaluation of the robustness of fault isolation approaches. Therefore, a significant part of the paper was devoted to identifying and discussing the main reasons for the lack of fault isolation robustness. Also, for research and comparative purposes, the term robustness of fault isolation was defined together with a preliminary proposal for its estimation which supposedly may find industrial acceptance. Finally, a heuristic qualitative metric of fault isolation robustness has been proposed. It allows for a quick and straightforward assessment of the fault isolation robustness, which promises its practical usability.

Keywords: AI and FDI methods, FDI by means of structural properties, Artificial Intelligence methods, Power plants and power systems
Abstract: The self-attention mechanism comes from the human visual function, which imitates the internal process of living beings when observing, and is widely used in the field of deep learning, such as natural language processing and image recognition. In the dynamic industrial process, it not only needs to mine the information of massive data, but also needs to analyze the correlation between samples. The self-attention mechanism can analyze the internal characteristics of data well and focus on global and local important information, so it is suitable for process monitoring problems. Currently, in the field of process monitoring, the self-attention mechanism has not been widely used. This paper innovatively proposes a dynamic process monitoring framework based on self-attention mechanism named self-attention principal component analysis (self-attention PCA). Experiments have verified that self-attention PCA has a great monitoring effect on incipient faults in the dynamic process.

Keywords: AI and FDI methods, Computational intelligence methods
Abstract: As a kind of the forest “fault”, fire is highly destructive and difficult to rescue. Fire segmentation is helpful for firefighters to understand the fire scale and formulate a reasonable fire-fighting plan. Therefore, this paper proposes a real-time fire segmentation method based on a deep learning method. This method is an improved version of deeplbav3+, which is an encoder-decoder structure network. Encoder network is composed of deep convolutional neural network and atrous spatial pyramid pooling. Different from deeplabv3+, in order to improve the segmentation speed, this paper uses the lightweight network mobilenetv3 to build a new deep convolutional neural network and does not use atrous convolution, but it will affect the segmentation accuracy. Therefore, in order to compensate for the loss of segmentation accuracy, on the basis of the original decoder network, this paper adds two different shallow features to make the network contain rich fire feature information. Experimental results show that the comprehensive performance of this method is better than the original deeplabv3+, especially the segmentation speed of the network is greatly improved, which is about 59 FPS.

Keywords: Computational methods for FDI, FDI for linear systems, Structural analysis and residual evaluation methods
Abstract: Efficient estimation of the origin-destination (OD) matrix is a crucial requirement for traffic monitoring and control. The OD matrix estimation problem has received significant attention over the past decades and various approaches using traffic counts from fixed location sensors have been developed and tested. A significant challenge when using information obtained from traffic sensors, is that such sensors are subject to considerable disruptions due to system errors, that affect the quality and reliability of the information provided. This paper presents a novel approach for OD matrix estimation in the presence of faulty sensors. For the purposes of this work we assume that the network under study operates under free-flow conditions and utilise a variation of the cell transmission model to capture the traffic flow dynamics in a pre-specified time window. The problem is formulated in an optimisation framework and solved by taking into account the presence of faulty measurements. We test and validate the proposed methodology on a sample network and investigate the advantage of OD matrix estimation in the presence of faulty measurements compared to OD matrix estimation when faulty sensors are not considered.

Keywords: FDI for linear systems, Interval methods, hydraulic systems
Abstract: This paper addresses the system identification problem, as well as its application to robust fault detection, considering parametric uncertainty and using zonotopes. As a result, a Zonotopic Recursive Least Squares (ZRLS) estimator is proposed and compared with the Set-membership (SM) approach when applied to fault detection, taking as a reference the minimum detectable fault generated in the worst-case. To illustrate the effectiveness of the proposed robust parameter estimation and fault detection methodologies, a quadruple tank process is employed.

Keywords: FDI for linear systems, Advanced actuator technologies, FDI for nonlinear Systems
Abstract: Dampers in semi-active suspension systems may be subject to various types of damper faults, e.g., oil leakage and electrical issues, that need to be estimated for diagnosis and isolation purposes. A new method to design the so-called multi-objective grid-based Lipschitz Nonlinear Parameter Varying (NLPV) Proportional Integral (PI) observer is here developed for damper fault estimation, where the fault is the loss of efficiency of the damper modeled as a slow-varying input. While the damper nonlinearity is bounded by the Lipschitz condition, the H-infinity and generalized H2 conditions are used to minimize the effects of the input disturbance and the measurement noise, respectively, on the estimation error. Moreover, the observer is designed with Linear Matrix Inequalities (LMIs) formed and solved in a grid-based manner (considering a parameter-dependent Lyapunov function) to reduce the level of conservatism. Analyses in the frequency domain using Bode plots as well as in the time domain using realistic simulations illustrate the effectiveness of the proposed observer.

Keywords: FDI for linear systems, AI and FDI methods, Signal and identification-based methods
Abstract: In this paper, a fault detection and isolation scheme for multiplicative faults in dynamic systems based on data-driven K-gap metric and k-nearest neighbour (kNN) classification is proposed. To detect multiplicative faults, the standard classification task of kNN is studied from the viewpoint of system analysis. To this end, the data-driven stable kernel representation based on input/output data is presented for feature extraction capturing the dynamic of linear time-invariant (LTI) systems. Data-driven K-gap metric is used as an alternative tool for distance measure between two kernel subspaces in the kNN algorithm. A simulation example on the three-tank system (DTS200) demonstrates the successful detection and isolation of various multiplicative faults.

Keywords: FDI for linear systems, Signal processing for FDI, Computational methods for FDI
Abstract: Fault tolerance and constant, remote control of the process becomes important because of a continuous increase in complexity and high-performance demand for industrial systems. Therefore, there are developed advanced control systems that can detect and identify faults without being present at the site. This article presents the methodology of designing a PLC based fault isolation and remote monitoring system using the Industrial Internet of Things technologies. An algorithm of fault detection and isolation based on Unknown Input Observer and its implementation on a PLC controller are described and analyzed based on the simulated level control process in a three-tank system. The article contains a description of multiple technologies related to the concept of Industry 4.0 that can increase the performance of modern industrial systems.

Keywords: Applications and fault tolerant control, reconfigurable control, MPC methods, Filtering and estimation
Abstract: This paper develops a novel model predictive control (MPC) formulation for aircraft with simultaneous state and fault estimation. An observer based method for state and fault estimation is combined with a standard integral action MPC scheme to accurately detect randomly occurring faults within the system inputs (actuators). Actuator faults are modeled as loss of effectiveness (LOE). As this MPC is based on a delta-input formulation, the closed loop system in both fault-free and fault conditions is stable, thus eliminating the need for the time consuming reconfiguration of the internal MPC model. Numerical simulations demonstrate the effectiveness of the proposed approach.

Keywords: Fault accommodation, Reconfiguration strategy, MPC methods, Applications and fault tolerant control, reconfigurable control
Abstract: Predictive maintenance techniques are getting increasing interest in industry because instrumental to determine suitable reliability conditions of in-service equipment aimed at smartly establishing when a maintenance operation should take place. This approach presents economic benefits when compared to more traditional routines or time-based preventive maintenance because tasks are performed only when warranted. Within this context, this work presents a control allocation scheme based on Model Predictive Control ideas to deal with actuator loss of effectiveness related to their usage. These undesired events occur when the associated degradation effects, whose evolution is assumed to be measurable and/or predictable, overcome a given threshold. The proposed scheme consists of two modules: a prognostic unit in charge of monitoring the actuators' health and a reconfigurable control allocation module whose action is based on the current actuators' degradation level. Its main goal is to establish a suitable time interval during which a maintenance service can occur without affecting the stability of the controlled system and the feasibility properties of the control allocation scheme.

Keywords: Applications and fault tolerant control, reconfigurable control, Fault accommodation, Reconfiguration strategy, Active diagnosis, test selection
Abstract: This paper is focused on Wireless Sensor Network (WSN) leveraging on Bluetooth Low Energy (BLE) connectivity for low energy applications which is fault tolerant versus communication path failures. The topic is important to create a robust sensorized environment to be applied in industrial context or smart infrastructure to enable scheduled monitoring with low power consumption applications. Currently BLE applications are mainly thought for smart home solutions, health care and positioning systems. In those applications the BLE nodes are continuously supplied by external power suppliers. Our goal is to design a self-configuring network with a synchronized deep sleep behavior, aimed to optimize the energy consumption, with an overall active time interval constraint optimized with a data-driven method. The aim is to find a tradeoff between the on time and the ability to collect all the nodes data, pursuing a low power consumption. Our research is based on BLE protocols, interaction between edge systems for data collection and cloud system for data analysis and software agent optimization system. The paper analyses different configurations and describes the possible optimization algorithm to be used for the software agent design, in order to reach a fine-tuned control to improve the fault tolerance and fault diagnosis of the system. Finally experimental results are compared with the estimates obtained via a software simulation tool implemented for this architectural pattern.

Keywords: Applications and fault tolerant control, reconfigurable control, Controller reconfiguration, networked systems, FDI theory for networked systems
Abstract: This paper deals with the problem of distributed passive fault tolerant formation tracking control for cooperative second order multi-agent systems (MASs) subject to disturbances and sensor/actuator faults. The proposed scheme is based on decentralized observers used to robustly estimate the actuator and sensor faults in spite of disturbances. These estimates are then injected into a dynamic control law in order to mitigate their effects on the control objective. Using the mathcal{H}_infty method, graph theory properties and the projection lemma, sufficient conditions in the form of a set of linear matrix inequalities (LMIs) are derived to guarantee the stabilization of the tracking errors while reducing the effects of sensor and actuator faults and disturbances. A numerical simulation illustrates the effectiveness of the proposed passive fault-tolerant control scheme.

Keywords: AI and FDI methods, Flight control, fault detection
Abstract: Anomaly detection is a crucial aspect of embedded applications. However, limited computational power, evolving environments, and lack of training data are difficulties that can limit anomaly detection algorithms. One class classification algorithms are often used for this task to circumvent the need of anomalous data in the training set. This paper presents a new machine learning algorithm for anomaly detection called Dynamic Double anomaly Detection DyD² that is suited to evolving environments and on-board requirements. The contributions made by DyD² are thoroughly presented and an experimental evaluation is set up to compare DyD² to state-of-the-art algorithms.

Keywords: AI and FDI methods, FDI by means of structural properties, Artificial Intelligence methods
Abstract: This paper presents a fault diagnosis approach that combines structural and data-driven techniques. The proposed method involves two phases. Firstly, the residuals structure is obtained from the structural model of the system using structural analysis without considering mathematical models (only the component description of the system). Secondly, the analytical expressions of residuals are obtained from available historical data using a robust identification approach. The diagnosis part consists in checking the evolution of residuals during the process, any inconsistency of residuals can be considered as a fault, so that the thresholds for each residual are introduced. The residuals are obtained using the identified interval model that takes into account the uncertainty and noises affecting the system. Once a faulty scenario is detected, it is also possible to determine which is the most probable fault that occurred in the system through a Bayesian reasoning approach that uses the FSM (Fault Signature Matrix) obtained from the structural analysis of the system and residual activation signals. The proposed approach is applied to a brushless DC motor (BLDC) used as a case study. Simulation experiments illustrate the performance of the approach.

Keywords: AI and FDI methods, Mechanical and electro-mechanical applications
Abstract: With the availability of sensor data and increased processing power, data-driven approaches have been widely investigated to improve various production processes. It is easier to keep track of faults that are hard to perceive during an operation through data-driven condition monitoring systems. In addition, the operator or the supervisor of the processes can recognize the need for service in time and provide maintenance. A defect in agricultural machinery can reduce the quality of fieldwork noticeably and cause damages to the crops as well as to the machine itself. Moreover, such damages and low-quality work in agriculture are expensive to fix and sometimes even irreversible. Thus, it is crucial to develop intelligent condition monitoring systems for agricultural machinery. Specifically, machines with rotating components, such as disc mowers, are prone to damage if they are frequently deployed in places where they might hit solid objects. These anomalies cannot always be easily recognized by the operator and may cause suboptimal results. This paper proposes concepts of condition monitoring for a disc mower to optimize the mowing operation. In this study, the first concepts of deep learning-based systems were developed to provide notifications to the operator when a failure occurs.

Keywords: AI and FDI methods, Signal and identification-based methods
Abstract: Today, due to the growing complexity of engineered systems, it is crucial to develop technologies able to deal with the systems' behavior to maintain a high degree of safety, reliability, and efficiency while reducing operating expenses such as maintenance costs. The idea is to develop a Prognostics and Health Management (PHM) framework to monitor these complex systems' behavior using sensory data and then apply machine learning models to infer the current health state. One important goal is to estimate the Remaining Useful Life (RUL) essential for optimizing maintenance processes and sustainable practices in industrial settings. This work presents an empirical analysis for RUL estimation via a model degradation built using condition monitoring data. A support vector machine regression (SVR) model and a similarity measure algorithm are employed to extract degradation trends and compute the RUL. We evaluate the prognostics performance and compare the results with reported benchmarks from publicized works.

Keywords: AI and FDI methods, Neural approximations for optimal control and estimation, FDI for nonlinear Systems
Abstract: Different from the traditional model-based fault diagnosis paradigm which is established upon the well-known observer design and analysis, a novel data-driven framework is proposed by combing systems modeling with fault detection for a class of 1-D unknown distributed parameter systems. The key idea is to transfer the on-line modeling error into the residual signal for fault detection. The proposed methodology only utilizes the I/O data and does not require extra knowledge of the system model, which increases its usability at large. Numerical simulations on a commonly used benchmark are presented for method validation.

Keywords: Interval methods, hydraulic systems, Filtering and estimation, FDI for nonlinear Systems
Abstract: This paper proposes a novel fault detection method for distributed uncertain systems based on moving horizon estimation (MHE). In the proposed design, the system is distributed into multiple subsystems and local MHEs are designed for each subsystem to estimate the local states. Interval arithmetic is applied to describe the unknown-but bounded uncertainties that are present in practical systems. Based on this, a new MHE problem is formulated which incorporates the uncertainties. Data loss in the communication is also handled. The solution of the MHE problem provides an optimal estimate of the interval-valued states. Faults in the subsystems are detected by evaluating the result of the intersection of the optimal interval-valued states with the local uncertain measurements. Finally, the proposed fault detection method is demonstrated using measured data of a nonlinear three-tank system. Furthermore, the new method is compared with the classical MHE-based global approach.

Keywords: FDI for linear systems, Computational methods for FDI, Reconfigurable control, sensor and actuator faults
Abstract: This paper extends the general technique for additive system faults estimation for linear discrete-time systems. Based on the discrete-time observer form, the systematic design procedure for the proportional-derivative observer structure is addressed, assuming that the considered faults are additive. Conditions for fault estimation design are found using Lyapunov function reflecting the descriptor form of the estimation error, which states the results robust against the effects from the disturbances and fault boundaries. Under used descriptor formulation, the conservatism of the fault estimator can be reduced reflecting more design degrees of freedom in design. The design procedure gives an efficient method to reappear the fault timely and accurately, reflecting the given design constraints.

Keywords: FDI for linear systems
Abstract: This work proposes a set-valued extension of the classical unknown input observers for linear continuous-time systems without using set-by-step set-membership computations or imposing the positivity property of the dynamics of the estimation error. In fact, based on an explicit set-integration method of the estimation error, the framing and convergence properties of the proposed set-valued unknown input observer are demonstrated. Moreover, thanks to the proposed design approach, robust threshold on the residual can be established a priori. Simulation results are reported to highlight the effectiveness of the proposed state estimator in the presence of unknown inputs.

Keywords: Filtering and estimation, FDI for linear systems
Abstract: Adaptive residual generator is a powerful technique that delivers a simultaneous co-estimation of the state variables and unknown parameters with stability and convergence guarantee. This paper reports our recent study on a projection-aided adaptive residual generator design for disturbance-decoupled process monitoring. The core of the study is the implementation of a projection on the orthogonal complement of disturbance subspace, which results in a robust state and parameter estimation against unknown disturbance, furthermore, accurate process monitoring can be achieved.

Keywords: Flight control, fault detection, Active diagnosis, test selection
Abstract: Process monitoring and fault diagnosis is the key to ensuring process safety and improving product quality. In traditional studies, process safety indicator related fault detection and evaluation have not been fully resolved. To address this problem, a method of maximum information coefficient and slow feature analysis is presented in this paper. The relevance between process and safety variables is calculated by using the maximum information coefficient, and the process variables are divided into strongly safety-related block, weakly safety-related block and safety-unrelated block. Then, the distributed local monitors are established in each block using slow feature analysis method which can identify the process dynamic anomalies. Meanwhile, a new decision fusion method based on logical analysis is proposed to process monitoring and fault grade assessment. Finally, the proposed method is practiced with the Tennessee Eastman Process, and the effectiveness and superiority are proved by comparing with other methods.

Keywords: Fault accommodation, Reconfiguration strategy, Computational intelligence methods, Reconfigurable control, sensor and actuator faults
Abstract: Design of fault-tolerant control for nonlinear systems is usually focused on maximizing some criteria defining the level of robustness with respect to potential faults without significant degradation of the system performance. Therefore, strong interest is produced by compromise approaches which would generate decent control quality for the broadest possible class of potential faulty scenarios. Here, an approach is proposed with the application to magnetic brake system making use of the repetitive character of the control task. The concept of iterative learning control driven by measurement data is utilized to properly update the control signal in order to adapt to possible actuator fault states. A learning controller is adopted build on a mixture of neural networks for various operating points. Thus, it is able to adapt to changing working conditions of the device. Moreover, using the procedure employing an ensemble of inverse models, actuator faults can be successfully accommodated. The paper provides the complete iterative learning procedure including the system identification, fault estimation and fault-tolerant control. The numerical example on the reference tracking problem for magnetic brake system is discussed, taking into account various faulty scenarios.

Keywords: Fault accommodation, Reconfiguration strategy, FDI for linear systems
Abstract: The development of any fault tolerant control solution is based on the strong assumption that fault situations can be accommodated. This paper provides a survey of four reachable set techniques to assess the fault recoverability property for constrained linear time invariant (LTI) systems by means of ellipsoid, zonotope, polytope and support function representations. These techniques are next applied to an angular velocity spacecraft model. A discussion is finally made to assess the computational complexity for the four algorithms.

Keywords: MPC methods, Interval methods, hydraulic systems, Fault accommodation, Reconfiguration strategy
Abstract: This paper proposes a resilient and robust model predictive control (MPC) scheme for a class of Cyber-Physical Systems (CPS) subject to state and input constraints, unknown but bounded disturbances and Denial of Service (DoS) attacks. The attacker blocks the controller to actuator communication and the attacks are assumed to be time-limited. The control is designed by extending a robust tube-based MPC, where a new type of invariant set, namely µ-step Robust Positively Invariant (µ-RPI) set, is introduced to deal with resilience. A set-based method is then developed for the control scheme to ensure both resilience to DoS attacks while preserving robustness to bounded disturbances. A computational algorithm is derived and a numerical example is provided to illustrate the potential of the proposed approach.

Keywords: Computational intelligence methods, Reconfigurable control, sensor and actuator faults, Applications and fault tolerant control, reconfigurable control
Abstract: Model-based fault-tolerant control (FTC) often consists of two distinct steps: fault detection & isolation (FDI), and fault accommodation. In this work we investigate posing fault-tolerant control as a single Bayesian inference problem. Previous work showed that precision learning allows for stochastic FTC without an explicit fault detection step. While this leads to implicit fault recovery, information on sensor faults is not provided, which may be essential for triggering other impact-mitigation actions. In this paper, we introduce a precision-learning based Bayesian FTC approach and a novel beta residual for fault detection. Simulation results are presented, supporting the use of beta residual against competing approaches.

Keywords: Reconfigurable control, sensor and actuator faults, Flight control, fault detection, Applications and fault tolerant control, reconfigurable control
Abstract: Control of multi-rotor unmanned air vehicles (UAV) with actuators failure is essential for modern civil aviation. UAV are increasingly used to provide service in a wide range of applications. However, many constraints slow down their integration particularly a poor reliability record. To solve this problem, a novel active fault tolerant control (AFTC) technique is used to design a control reconfiguration scheme for actuators failure in both under and full actuated cases for a hexacopter. A failure mitigation technique is developed to control only the reduced attitude dynamics and sacrifice the yaw by allowing the UAV to spin around an axis in case of failure of two adjacent rotors i.e. under actuation mode. The simulation results of the proposed AFTC technique show good performance in both fault-free and failure conditions.