Navigation algorithms are responsible for decision-making in autonomous vehicles, generating safe trajectories based on post-processed sensor data to prevent collisions. However, failures in sensors or software modules can cause deviations from nominal behavior, potentially leading to unsafe situations. To comply with the ISO 26262 functional safety standard, road vehicles must perform a transition to a safe state when a failure is detected, until the corresponding safety goal is reached. This paper presents a novel control framework for navigation in autonomous vehicles, integrating data-driven reachability analysis and zonotopic predictive control to ensure functional safety (FuSa) under nominal conditions and potential system-level anomalies. The proposed framework explicitly accounts for interactions between vehicles, surrounding traffic participants, and infrastructure by incorporating traffic and infrastructure constraints into the control formulation. The proposed approach uses signal temporal logic specifications to formally define safe states and safety goals, guiding the zonotopic predictive control in executing safe maneuvers when they are required. The controller also enforces constraints on obstacle avoidance and infrastructure conditions, all represented as intervals or zonotopes for a unified formulation. The complete control framework is validated using data extracted from a real-world autonomous Renault Mégane (SAE Level 3), demonstrating its effectiveness in executing safe recovery maneuvers while ensuring compliance with the functional safety standard in real-world scenarios.

This paper proposes a novel non-linear lane selection algorithm for autonomous vehicles based on the Optimality Condition Decomposition algorithm. An initial algorithm based on mixed integer quadratic programming was studied, which proved untreatable due to its computational complexity, leading to similar formulations that do not require such a nature in their variables. As a result, a cascade strategy combining both distributed local planning and decision-making layers is implemented, leading to promising results. The proposed approach can determine paths that consider individual goals and the environment so that agents can collaborate to establish a formation that re-configures online, allowing vehicles to reach their individual goals.

This paper proposes a novel fault-tolerant control (FTC) reconfiguration strategy for autonomous vehicles using Model Predictive Control (MPC) based on a Linear Parameter Varying (LPV) model to address steering faults. The proposed approach compensates for the lack of redundant steering actuators by using force differences between the left and right sides of the vehicle to generate corrective yaw moments. By integrating both lateral and longitudinal dynamics, the MPC optimally allocates actuator efforts based on fault severity and desired speed. Simulation results validate the effectiveness of the proposed strategy in maintaining vehicle stability and performance under various fault scenarios, including complete steering failure, thereby ensuring safe autonomous operation.

Human Action Recognition (HAR) is integral to various domains, particularly in intelligent transportation systems where real-time understanding of driver behavior is critical for road safety. Accurately recognizing driver actions is challenging due to inter-class similarity, viewpoint variation, and real-world occlusions. This paper proposes a multi-view HAR approach utilizing a Siamese neural network architecture, fine-tuned with DenseNet201, to learn discriminative feature embeddings across front and side perspectives. Evaluations on the challenging 3MDAD dataset demonstrate the robustness of our method against viewpoint and modality shifts, achieving 58.22

The rapid advancements in autonomous vehicle software present both opportunities and challenges, especially in enhancing road safety. The primary objective of autonomous vehicles is to reduce accident rates through improved safety measures. However, the integration of new algorithms into the autonomous vehicle, such as artificial intelligence methods, raises concerns about the compliance with established safety regulations. This paper introduces a novel supervisor architecture based on behavior trees, aligned with established standards and designed to supervise vehicle functional safety in real time. It specifically addresses the integration of algorithms into industrial road vehicles, adhering to the ISO 26262. The proposed supervisor architecture involves the detection of hazards and compliance with functional and technical safety requirements when a hazard arises. This methodology, implemented in this study in a Renault Mégane (currently with SAE level 3 of automation capabilities), not only guarantees compliance with safety standards, but also paves the way for safer and more reliable autonomous driving technologies.

This paper considers the presence of movable bridges in inland waterway transport, and presents a control framework for the joint dynamic coordination of bridge operations and autonomous vessel navigation to minimize waiting times of vessels at bridges. Simultaneous evolution of bridge occupancy and vessel position is captured by a control-oriented model that incorporates qualitative behavior in the form of propositional logic expressions. A model predictive control (MPC) strategy is designed considering adaptable bridge opening regimes to exploit vessel passage demand, and operational preferences of both vessel skippers and bridge operators are taken into account to reach fair trade-off decisions. A realistic case study pertaining to the Rhine-Alpine corridor is used to demonstrate the effectiveness of the approach. Appropriate key performance indicators (KPIs) are defined and employed for a quantitative comparison with a mixed-integer programming (MIP) strategy with fixed opening regimes. Furthermore, sensitivity to the main MPC parameters is examined by carrying out extensive testing to assess the effect of each design parameter on the solution.

In this paper, a novel distributed set-based localisation algorithm for autonomous vehicles based on zono-topes is proposed. The proposed approach structurally allows embedding data fusion locally and in a vehicle-to-vehicle or infrastructure-to-vehicle manner under treatable computational times. Moreover, the localisation is only based on zonotopic intersections, allowing each individual agent to freely acquire their local estimation by any means, being the single requirement the access of a piece of hardware providing a coupled measurement. The proposed approach is implemented in a distributed manner in a ROS network, mimicking the physical layout of a platoon of autonomous vehicles.

The safe and reliable operation of autonomous vehicles heavily relies on the precise identification and diagnosis of faults within their systems. Multiscale Kernel Principal Component Analysis (MSKPCA) model is well established for fault diagnosis. This study demonstrates the effectiveness of MSKPCA for detection in autonomous vehicles. It utilizes wavelet analysis and KPCA to extract pertinent feature from process data for fault detection. KPCA model transforms data into a feature space for extraction and selection pertinent data. However, the KPCA is often struggle with noisy data. To address this issue, we integrate a multiscale filtering technique with KPCA, to reduce the impact of the noise while preserving crucial features from the raw data. The proposed methodology uses raw data for training and testing across three sensor fault scenarios: sensor position faults, sensor velocity faults, and angle orientation faults. Square Predicted Error (SPE) and Hotelling T-squared (T2) statistics are computed in the feature space and compared to their respective thresholds to quantify the detection accuracies. The results show that the proposed approach MSKPCA performs well in fault detection in autonomous vehicles.

This paper presents an MPC-based control scheme for unmanned aerial vehicles (UAVs) of quadrotor type that can operate in real-time. The proposed approach is based on a multi-layer model predictive control architecture that uses linear parameter varying (LPV) and feedback linearization techniques in the inner and outer layer, respectively. In the outer layer, the non-linear dynamics of the transitional control are linearized using feedback linearization, followed by the implementation of a combination of linear model predictive control (MPC) and moving horizon estimator (MHE) schemes. In the inner layer, the attitude control utilizes the LPV approach to formulate an LPV MPC controller and MHE estimator. Through simulation using a high-fidelity simulator of a real UAV, the effectiveness of the proposed control scheme is demonstrated in terms of tracking performance and computation time. Furthermore, a comparison with other similar approaches in the literature is included to showcase the advantages of the proposed scheme.

Autonomous vehicles require sophisticated planning algorithms to navigate safely and efficiently in complex environments. Traditional centralised planning approaches face scalability challenges, especially as the number of agents increases. Distributed planning strategies have emerged as a promising solution to overcome these limitations. This paper presents a novel approach to distributed set-based planning for autonomous vehicles. Each vehicle can autonomously plan its path by lever-aging set-based representations of possible trajectories while considering uncertainty and dynamic environment changes. The proposed method enables vehicles to collaborate efficiently while maintaining decentralised decision-making capabilities, thus enhancing scalability and robustness. Furthermore, safe trajectories are derived by contemplating system uncertainties through set theory.

This paper proposes a robust fault diagnosis method for Autonomous Ground Vehicles (AGVs) modeled as a Linear Parameter Varying (LPV) system with bounded uncertainties. The proposed approach combines the zonotope-based Set-Membership Approach (SMA) and Set Invariance Approach (SIA). Firstly, an online fault detection strategy based on zonotopic set-membership state estimation is introduced, where the optimal observer gain is calculated offline by solving LMI optimization problems. To characterize the Minimum Detectable Fault (MDF) and Minimum Isolable Fault (MIF), the invariant residual sets are first obtained for the system operated in healthy and faulty modes. The proposed method relies on the propagation of the zonotopic state estimation error in steady state based on SIA. Then, MDF and MIF are characterized for several types of faults by solving optimization problems subject to set separation conditions. Finally, experiment validations using a prototype vehicle are performed to illustrate the effectiveness of the proposed approach.

In this article, the problem of generating multimodel state space descriptions in a data-driven context to embed the dynamic behavior of nonlinear systems is addressed. The proposed methodology takes advantage of three ingredients: 1) linear time-invariant system behavior; 2) data-driven modeling; and 3) reinforcement learning (RL) technicalities. These elements are properly combined to develop a data-driven algorithm capable to derive an accurate outer convex approximation of the nonlinear evolution. In particular, an actor-critic RL scheme is designed to efficiently comply with the exhaustive research on the whole parameter space. At each iteration, the effectiveness of the obtained uncertain polytopic model is tested by a probabilistic approach based on a confidence level metrics. As the main merits of the proposed approach are concerned, the following aspect clearly stands up: the development of an interdisciplinary methodology that takes advantage of system theory, probabilistic arguments and RL capabilities giving rise to an harmonized architecture in charge to deal with a vast class of nonlinear systems. Finally, the validity of the proposed approach is tested by resorting to benchmark examples that allow to quantify the level of accuracy of the computed convex hull.

The aim of this paper is to present and discuss an analysis approach that can be deployed independently of the modelling tool capable of evaluating the samples of an assembled machine. By including an analysis of the results given the context of the input variables and the training performance, the method could be applied independently to the changes in modelling or the modelling approach. A real case study based on a water-cooling device has been used to illustrate the proposed approach.

As processing power becomes more affordable, computer vision tends to push towards controlling complex processes in the industry. Surface inspection, with changing environmental conditions and the usual lack of homogeneity of the inspected parts, makes it a real challenge to overcome even for a skilled specialist. In addition, the scarcity of positive samples and the extremely small size of the defects, makes it even harder to cluster them in different classes. In this work, we propose a novel training strategy tailored to handle these challenges for the problem of image defect segmentation and classification. First, we propose a Context Aggregation Network with different dilation factors, in order to keep as much information as possible from every feature map, especially for the smallest defects. By splitting the loss in classification and segmentation and positively weighing both terms, we accomplish an optimal learning process counteracting possible imbalances in the dataset. Additionally, we introduce a novel guided-crop image augmentation method, which generates new images by cropping real defects from existing images, pasting them in real non-defective ones and finally tweaking their configuration. This augmentation strategically performed, guided by the evolution of each class loss, allows the model to identify better the least common and complicated to identify defects. We validate our solution with the Magnetic Tile and the Severstal Steel Defect Detection dataset, demonstrating that our approach consistently outperforms models such as ResNet-50, DenseNet-121, HRNet or UPerNet.

A Nonlinear Model Predictive Control (NMPC) strategy aimed at controlling a small-scale car model for autonomous racing competitions is presented in this paper. The proposed control strategy is concerned with minimizing the lap time while keeping the vehicle within track boundaries. The optimization problem considers both the vehicle’s actuation limits and the lateral and longitudinal forces acting on the car modeled through the Pacejka’s magic formula and a simple drivetrain model. Furthermore, the approach allows to safely race on a track populated by static obstacles generating collision-free trajectories and tracking them while enhancing the lap timing performance. Gazebo simulations using the F1/10 simulator showcase the feasibility and validity of the proposed control strategy. The code is released as open-source making it possible to replicate the obtained results.

This article presents a novel secure-control framework against sensor deception attacks. The vulnerability of cyber-physical systems with respect to sensor deceptive attacks makes that all sensor measurements are not reliable until the system security is assured by an attack detection module. Most of the active attack detection strategies require some time to assess the system security, while injecting a watermark signal to ease the detection. However, the injection of watermark signals deteriorates the performance and stability of the plant. The proposed control framework consists of a dual-rate control (DRC) that is able to stabilize the plant using: 1) a model predictive controller that operates at a slower sampling time; 2) a state-feedback predictor-based controller that operates in the nominal sampling time disregarding the use of the untrustworthy measurements until the attack detector is able to certify the security; and 3) a reconfiguration block (RB) for palliating the effect of the watermarking. Simulation results indicate the efficacy of the proposed DRC framework to defend the system from cyber-attacks and the ability of the RB to improve the closed-loop performance during the watermark injection.

The increasing use of renewable energies, in particular photovoltaics in recent years, leads researchers to study this type of energy in depth. Photovoltaic panels have a major problem that is when their temperature increases, their efficiency decreases. For this reason, many research works are interested to solve this problem in order to increase the electrical efficiency. The best solution used is the PV/T hybrid systems which consists of a photovoltaic panel and a thermal system co-combining in a single unit. In this paper, a presentation of different methods to increase the efficiency of PV/T hybrid systems and a mathematical model of a PV/T water system is developed. A simulation and an experimental validation showing the influence of the water flow variation on the efficiency of the real PV/T hybrid system, is tested.

The intrinsic redundancy of large-scale control systems allows to differentiate between the nominal system configuration, where the dynamics of the plant are controlled through a specific subset of actuators, and the possible configurations that arise from the combination of all the existing actuators (nominal plus back-up ones). After the occurrence of an unexpected event (component fault, isolation of a specific area due to security concerns, etc), some of the back-up system actuators can be used in order to guarantee that the system performance will remain within some admissible margins. Thus, given a scenario that causes a non-admissible system operation with the nominal configuration, the possibility to use back-up actuators in order to improve the system performance expands the traditional fault tolerant control scenario defining a new problem. Accordingly, this work is devoted to the introduction of the problem which is solved by formulating the actuator selection as a mixed-integer programming optimization. A water distribution network is used with the aim to illustrate the new problem implications as well as to demonstrate the validity of the proposed solution.

This paper exploits the analogy between the stochastic and zonotopic Kalman filters with the aim of formulating a metric that allows to assess the impact of an external zonotopically bounded signal in an state estimate optimal control scheme. To that end, the concept of linear quadratic zonotopic (LQZ) controller is introduced. Besides, the design of a zonotopically bounded watermarking signal such that enforces replay attack detectability is also addressed. The proposed detection scheme takes advantage of the observability loss of an a priori known exogenous signal during the replay phase of the attack. Consequently, by injecting the corresponding estimation error generated by a set-based observer, a zonotopically bounded signal such that imposes replay attack detectability is achieved. A numerical example is provided.

The solar PV power sector in Spain has been developing at a spectacular rate in recent years. The energy cost and the dependence on fossil fuels can be reduced by improving the efficiency of photovoltaic energy production. The performance of a solar radiation conversion system is affected by tilt angle with the horizontal plane. Thus, a photovoltaic array needs to be tilted at the correct angle to maximize its performance. In this paper, we investigate how to determine the optimum tilt angle. This procedure is illustrated for the city of Barcelona (Spain) located at latitude 41o 22' 56” North and longitude 2o 6'56” East. The optimal tilt angle varies according the seasons: In Winter (December, January, February) is 56.4, in Spring (March, April, May) is 29.11° in Summer (Jun, July, August) is 13.76° and in Autumn (September, October, November) is 48.14°. Finally, the annual optimum tilt angle for our latitude is 36.87°, with this optimal slope angle, maximum monthly and annual solar radiation is calculated. So, by means of the proposed approach, the energy generation is increased in 10.54% thanks to we are able to benefit from the maximum solar radiation.

This paper explores the use of deep learning for leak localization in Water Distribution Networks (WDNs) using pressure measurements. By using a training data set including enough samples of all possible leak localizations, a Convolutional Neural Network(CNN) can be used to learn the different pressure maps that carachterized each leak localization. The generalization accuracy has validated and evaluated by means of a testing data set. All of considered training, validation, and also testing data include leak size uncertainty, nodal water demand uncertainty and sensor noise. An innovative approach is proposed to convert every pressure residuals map to an image in order to apply a CNN. In addition with the purpose of filtering the effects of uncertainty and noise a time horizon Bayesian reasoning approach is used over each time instant classification output by the CNN. The Hanoi District Metered Area (DMA) is considered as a case study to illustrate the performance of the proposed leak localization method.

This paper proposes an economic health-aware model predictive control (MPC) for Drinking Water Transport Networks (DWN) that includes an additional goal to extend the components and system reliability evaluated using a Bayesian Network. The components and system reliability are incorporated in the MPC model using the LPV approach. The MPC controller uses additionally an economic objective function that determines the optimal filling/emptying sequence of the tanks considering that electricity price varies between day and night. The proposed LPV-MPC control approach allows the controller to accommodate the parameter changes. By computing an estimation of the state variables during prediction, the MPC model can be modified considering the estimated state evolution at each time instant. Moreover, the solution of the optimization problem associated with the MPC problem is achieved by solving a series of Quadratic Programs (QP) at each sampling time. This iterative approach reduces the computational load compared to the solution of a non-linear optimization problem. A part of a real drinking water transport network of Barcelona is used as a case study for illustrating the performance of the proposed approach.

This paper presents a replay attack detection method that addresses the performance loss of watermarking-based approaches. The proposed method injects a sinusoidal signal that affects a subset, chosen at random, of the system outputs. The presence of the signal in each one of the outputs is estimated by means of independent observers and its effect is compensated in the control loop. When a system output is affected by a replay attack, the loss of feedback of the associated observer destabilizes the signal estimation, leading to an exponential increase of the estimation error up to a threshold, above which the estimated signal compensation in the control loop is disabled. This event triggers the detection of a replay attack over the output corresponding to the disrupted observer. The effectiveness of the method is demonstrated using results obtained with a quadruple-tank system simulator.

This paper proposes a robust economic model predictive control method based on a bounded description of the demand uncertainty for optimal energy dispatch in a smart micro-grid. The proposed method considers the effect of uncertainty by tightening the system constraints along the prediction horizon. The bounds of the tightened constraints are computed offline. A periodic terminal state constraint is included in order to guarantee stability. The proposed method is simpler and faster than closed-loop min-max based approaches. The proposed approach is assessed with the help of a smart-grid consisting of some photovoltaic (PV) panels, a wind generator, a hydroelectric generator, a diesel generator, and some storage devices interconnected via two DC-buses, from which load demands can draw required energy. Load demands are uncertain, but considered as bounded.

Reinforcement Learning (RL) systems are trial-and-error learners. This feature altogether with delayed reward, makes RL flexible, powerful and widely accepted. However, RL could not be suitable for control of critical systems where the learning of the control actions by trial and error is not an option. In the RL literature, the use of simulated experience generated by a model is called planning. In this paper, the planningByInstruction and planningByExploration techniques are introduced, implemented and compared to coordinate, a heterogeneous multi-agent architecture for distributed Large Scale Systems (LSS). This architecture was proposed by (Javalera 2016). The models used in this approach are part of a distributed architecture of agents. These models are used to simulate the behavior of the system when some coordinated actions are applied. This experience is learned by the so-called, LINKER agents, during an off-line training. An exploitation algorithm is used online, to coordinate and optimize the value of overlapping control variables of the agents in the distributed architecture in a cooperative way. This paper also presents a technique that offers a solution to the problem of the number of learning steps required to converge toward an optimal (or can be sub-optimal) policy for distributed control systems. An example is used to illustrate the proposed approach, showing exciting and promising results regarding the applicability to real systems.

This work shows how a Linker agent coordinates a cooperative MAS environment to seek a global optimum. The approach is applied to the Barcelona Drinking Water Network (DWN) administrated by AGBAR where the main problem was to coordinate the control of three different sectors of the network. Each part has a local controller (local agent) to solve the local water demands, but it also has to cooperate with the other agents to satisfy the water demands of the whole network. The cooperative Linker agent implemented, learns by using a Reinforcement Learning algorithm, called PlanningByExploration Behaviour with penalization (Javalera et al., 2019), to converge towards an optimal (or suboptimal) value of each of the variables that connect the local agents. For the training and simulation of the Linker agents real historical data of the Barcelona DWN provided by AGBAR were used, as well as the data to model the distributed topology of the DWN. Moreover, some results of the simulations of this approach in contrast with the results of a centralized Model Predictive Controller are depicted.

This paper proposes a robust fault detection and isolation (FDI) approach that combines active and passive robust FDI approaches. Standard active FDI approaches obtain robustness by using the unknown input observer (UIO) to decouple unknown inputs from residuals. Differently, standard passive FDI approaches achieve robustness by using the set theory to bound the effect of uncertain factors (disturbances and noises). In this paper, we combine the UIO-based and the set-based approaches to produce a mixed robust FDI, which can mitigate the disadvantages and exert the advantages of the two robust FDI approaches. In order to emphasize the role of set theory, the UIO design based on the set theory is named as the set-theoretic UIO (SUIO). A quadrotor subsystem is used to illustrate the effectiveness of the proposed FDI approach.

This paper evaluates the applicability of command governor (CG) strategies to the optimal management of drinking water supply systems (DWSSs) in both centralized and distributed ways. It will be shown that CG approaches provide an adequate framework for addressing the management of these large-scale interconnected systems in the presence of periodically time-varying disturbances (water demands) that can be anticipated by using time-series forecasting approaches. The proposed centralized and distributed CG schemes are presented, discussed, and compared when applied to the management of DWSS considering the same set of operational goals in all cases. This paper illustrates the effectiveness of all strategies using the Barcelona DWSS as a case study and highlighting the advantages of each approach.

The problem of energy dispatch in heterogeneous complex systems such as smart grids cannot be efficiently solved using classical control or ad-hoc methods. This paper proposes the application of Economic Model Predictive Control (EMPC) for the management of a smart micro-grid system connected to an electrical power grid. The system comprises several subsystems, namely some photovoltaic (PV) panels, a wind generator, a hydroelectric generator, a diesel generator, and some storage devices (batteries). The batteries are charged with the energy from the PV panels, wind and hydroelectric generators, and they are discharged whenever the generators produce less energy than needed. The subsystems are interconnected via a DC Bus, from which load demands are satisfied. Assuming the load demand and the energy prices to be known, this study shows that EMPC is economically superior to other Model Predictive Control (MPC) based strategies (a standard tracking MPC, and their cascaded version in form of hierarchical two-layer approach).

This paper presents a comparative study of three different approaches to design Model Predictive Control (MPC) strategies for a pasteurization plant using Linear Parameter Varying (LPV) models. The first two methods consider the LPV model in the design of the MPC controller in two different manners. The last approach uses a Robust MPC controller for taking parameter variations of the LPV model into account. It is assumed that the disturbances are unknown but bounded and the zonotopic set representation is used for modelling the uncertainty. In addition, a comprehensive comparison of the closed-loop performance accounting the proposed approaches is carried out through a high-fidelity simulator of a utility-scale pasteurization plant.

This paper presents a methodology and architecture for the advanced monitoring of an industrial process integrating several sources of information using a data warehouse (DW) that include as metadata datamart to cross technical ubications and equipments with the information given by the existing monitoring systems and the time dimension. The advanced monitoring includes functionalities that allow to diagnose faulty components and to prognose faulty situations when a problem occurs in the production process. A real car painting process is used for illustration purposes.

Inland navigation networks are composed of several artificial canals that are characterized by no slope. These canals are particularly subject to resonance phenomena, which can create waves such that the navigation condition might not be guaranteed. It is therefore required to ensure dealing with freefault measured data and actuators. In this work, a fault detection and isolation method based on the Integrator Delay Zero model (IDZ) is designed for flat navigation canals. The proposed method is dedicated to the detection and isolation of sensor and actuator faults. Finally, it is tested by considering the Cuinchy-Fontinettes canal located in the north of France.

The development of efficient methods for the control system performance verification has drawn a lot of attention recently. In this paper, the use of viability theory for this purpose is investigated in case of non-linear systems. In particular, verification algorithms based on the use of the computation of invariance and viability kernels and capture basin are proposed. A Lagrangian method has been used in order to approximate these sets for nonlinear systems. Because of simplicity and efficient computations, zonotopes are adopted for set representation. An application example based on a well known control benchmark is provided in order to show the effectiveness of the proposed method.

The purpose of this paper is to develop a dynamic model of a rigid-flexible manipulator robot with a load on its endpoint using Euler-Lagrange formulation. In order to test the performance of the studied system, several mathematical functions are used as motion profile. It choice is very important because it affects the robot's performance. Different factors intervene in this choice. However, the most important is the torque's continuity and the movement's smoothness. Numerical simulations show the robustness of the dynamic model of the studied system for several motions profiles.

Fault estimation plays an important role in the fault diagnosis system since provides information about the fault magnitude and temporal evolution. In this paper, we present an approach that allows to obtain a simultaneous estimation of the fault, state and associated uncertainty intervals of a uncertain Takagi-Sugeno (TS) system. The fault estimation is obtained using a TS interval observer augmenting the system state with the fault and considering the system uncertainty in a bounded context. A set of Linear Matrix Inequalities (LMIs) have been derived to design the TS interval observer. With the purpose of illustrating the performance of TS interval observer for fault and state estimation, a case study based on a Proton Exchange Membrane (PEM) fuel cell is used.

International audience

This paper proposes a method to track the system mode and diagnose a hybrid system without building an entire diagnoser off-line. The method is supported by a hybrid automaton (HA) model that represents the hybrid system continuous and discrete behavioral dynamics. This model is built on request through parallel composition of the component HA models. Diagnosis is performed by interpreting the events and measurements issued by the physical system directly on the HA model. This interpretation allows us to construct the useful parts of the diagnoser developing only the branches that are required to explain the occurrence of incoming events. The resulting diagnoser adapts to the system operational life and is much less demanding in terms of memory storage than the entire diagnoser. In addition to this feature, the proposed framework subsumes previous works in that it copes with both structural and nonstructural faults. The method is validated by the application to a case study based on the sewer network of the city of Barcelona.

In this paper, a data detection/correction approach is proposed for a real environmental monitoring system, in order to provide a reliable dataset when sensor faults occur. This is the case of communication faults that may prevent the acquisition of a complete dataset, which is of paramount importance in order to successfully apply further system tasks such as fault diagnosis. Sensor detection/correction method presented here is based on the combined used of spatial and time series models. Spatial models take advantage of the physical relation between different variables emplaced in the system (temperature sensors here) while time series models take advantage of the temporal redundancy of the measured variables, by means of Holt-Winters models here. The proposed approach is successfully applied to the rock collapse forecasting system in the Torrioni di Rialba located in Lombardy (Italy).

In this paper, a robust fault diagnosis problem for nonlinear systems considering both bounded parametric modeling errors and noise is addressed using parity-equation-based analytical redundancy relations (ARR) and interval constraint satisfaction techniques. Fault detection, isolation, and estimation tasks are considered. Moreover, the problem of quantifying the uncertainty in the ARR parameters is also addressed. To illustrate the usefulness of the proposed approach, a case study based on the well-known wind turbine benchmark is used.

This paper presents a data-driven prognostic strategy for failure prediction and computing the remaining useful life (RUL) using an autoregressive (AR) model combined with the recursive least squares (RLS) algorithm. The proposed method not only provides an estimation of the remaining useful life (RUL), but also a confidence interval based on modeling the uncertainty as a probabilistic Gaussian variable. To illustrate the performance of the proposed approach, a conveyor belt system that uses an AC electric motor to move a cart from one end to the other is used.

This paper proposes a fault diagnosis method using a timed discrete-event approach based on interval observers that improves the integration of fault detection and isolation tasks. The interface between fault detection and fault isolation considers the activation degree and the occurrence time instant of the diagnostic signals using a combination of several theoretical fault signature matrices that store the knowledge of the relationship between diagnostic signals and faults. The fault isolation module is implemented using a timed discrete-event approach that recognizes the occurrence of a fault by identifying a unique sequence of observable events (fault signals). The states and transitions that characterize such a system can directly be inferred from the relation between fault signals and faults. The proposed fault diagnosis approach has been motivated by the problem of detecting and isolating faults of the Barcelona's urban sewer system limnimeters (level meter sensors). The results obtained in this case study illustrate the benefits of using the proposed approach in comparison with the standard fault detection and isolation approach.