Fault detection performances analysis for stochastic systems based on adaptive threshold

A sliding mode control method is developed in this study for application to a class of underactuated systems with bounded unknown disturbance and sensor and actuator faults. In the proposed method, a robustness item compensates for the bounded unknown disturbance and a Nussbaum function realises sensor and actuator faults tolerance simultaneously, and all signals of the system are proven to be bounded. A radial basis function (RBF) neural network is developed to estimate the unknown functions of the system. Finally, Hurwitz stability analysis is conducted to guarantee the stability of the closed-loop system. Simulations are conducted wherein a coupled motor driving system is placed under the proposed control laws to validate this approach.

Detection, isolation and identification of multiple actuator and sensor faults in nonlinear dynamic systems: Application to a waste water treatment process

A vehicle health monitoring system based on analytical redundancy, is developed for automated passenger vehicles. A residual generator and a residual processor are designed together to detect and identify actuator and sensor faults of the Buick LeSabre rapidly. The residual generator includes fault detection filters and parity equations. It uses the control commands and sensor measurements to generate the residuals, which have a unique static pattern in response to each fault. Then, the residual processor interrogates the residuals by matching them to one of several known patterns. It computes the probability of each hypothesis conditioned on the history of residuals. The fault detection latency is reduced by integrating the design of the residual generator and the residual processor. The vehicle health monitoring system is evaluated in real-time on a Buick LeSabre. The vehicle sensor and actuator faults are simulated artificially by the computer or created manually by the driver. In one experiment, a real intermittent sensor fault occurred and was immediately detected and identified. The real-time evaluation demonstrates that the vehicle health monitoring system can detect and identify actuator and sensor faults under various disturbances and uncertainties with almost minimal detection latency.

A multi-engine cluster of a launcher operates in extreme conditions, making the tasks of detection and isolation complex, with only a precise group of faults with small magnitudes worth being detected and isolated. To address this, a model-based Fault Detection and Isolation (FDI) system is developed for a multi-engine cluster using Structural Analysis (SA). A model of a multi-engine cluster composed of three engines is established, where different sensor and actuator faults are considered. After analyzing the model using SA, more than 16,000 residual generator candidates are obtained. An algorithm is then proposed to find subsets of residuals with minimal cardinality capable of isolating all faults. Each subset generates a unique fault signature for each fault, enabling fault isolation. The selection of the best subsets is based on the new definition of the Subset Sensitivity Index (SSI), according to which the subsets composed of the most sensitive residuals have the highest SSI values. The residual sensitivity is calculated using a new method based on the signal-to-noise ratio (SNR), where a new definition of the Residual Sensitivity Index (RSI) is given. Furthermore, a new algorithm is presented for fault isolation. Suppose a theoretical fault signature is not observed. In that case, the algorithm uses the knowledge of the residual sensitivity (RSI) to define which residual did not behave as expected and predicts which fault is active in the system. Monte Carlo simulations are conducted under ten different faults to analyze the performance of the proposed FDI system, including actuators and sensor faults.

In this paper, a robust unknown input observer-based actuator fault detection and isolation (FDI) scheme is applied on an example aircraft fluid thermal management system (FTMS). System dynamics are modeled using a graph-based approach. A linearized state space model is used in a bank of unknown input observers (UIOs) to generate a set of structured robust (in the sense of disturbance decoupling) residuals. Simulation results show successful actuator FDI in the presence of unknown inputs.

Both estimation and accommodation of actuator faults is required to increase the robustness of control systems. Estimation helps in the early detection and characterization of faults, and allows for the scheduling of preventative maintenance before the onset of catastrophic failures. Fault accommodation helps mitigate the effect of faults or age-dependant component degradation, and allows for the design of control systems which incorporate low-cost components. We show that the integral action of the proportional-integral (PI) form of both the Luenberger and adaptive observers can be used to estimate and accommodate actuator faults. Our simulations show that the accurate estimate of fault induced disturbances by the integral action allows a PI observer-based controller to outperform a LQR when unknown actuator faults are present. When adaptation is required to estimate unknown plant parameters, similar results are obtained.

This paper presents the concept of a convoy of autonomous electric vehicles which will be able to continue its mission in the presence of some actuator or sensor faults. The autonomous vehicle is called RobuCar which is constituted by four independently driven wheels and two independently adjustable steering angles. Firstly, the authors develop longitudinal and lateral controllers for the vehicles operating normally. The objective is to follow a reference trajectory and to maintain safe spacing between vehicles. Secondly, the authors develop some strategies based on fault detection and isolation results in order to reconfigure the convoy of vehicles in presence of faults. Three categories have been considered: sensor or actuator fault which has not altered the trajectory; fault that affects the velocity or disturbs the trajectory; faults that make impossible for the vehicle to continue its mission. Simulation results are given at the end to show the efficiency of the approach.

The fault detection problem is investigated for discrete-time linear uncertain systems. Instead of designing a fault detection system from the viewpoint of observer design for robust residual generation, an adaptive threshold approach is proposed to attain robustness against disturbance and norm-bounded model uncertainty. The main goal of the research is to develop a threshold design method that could establish an appropriate trade-off between false alarms and missed fault detection in the presence of model uncertainty. For this purpose, the H∞ optimization technique is adopted in the linear matrix inequality framework to compute the unknown parameters of an adaptive threshold. It is shown that the proposed fault detection system based on an adaptive threshold depends only on the system parameters and the control input of the monitored system. It is independent of robust residual generator designs in traditional observer-based fault detection systems. The effectiveness of the proposed approach is verified on two well-known benchmark systems: a direct-current motor and three tank systems. Several types of faults are successfully detected in both applications.

Fault detection in continuous industrial processes is crucial to ensure production, quality, safety, environmental and economic aspects. Reliable fault detection systems have become even more necessary due to increasingly stringent regulations. The complexity of processes, on the one hand, and the availability of data, on the other, have spread the adoption of the data-driven approach for industrial fault detection. Deep Reinforcement Learning (RL) has recently been applied for this purpose, mainly in rotating machinery and rolling bearings. In chemical processes, most RL applications are in process control. The main contribution of this work lies in proposing a reinforcement learning (RL)-based framework for fault detection in continuous chemical processes. The challenges are related to the large number of variables, nonlinear relationships and changing operational dynamics, to name a few. The classic Tennessee Eastman Process (TEP) benchmark was used as a case study. The results were compared with Principal Component Analysis (PCA), commonly used for chemical process monitoring, dynamic PCA (DPCA), and neural networks. Compared to PCA, the RL approach showed considerably lower missed detection rate (MDR) and time to detection (TTD) while maintaining reasonable false alarm rate (FAR) for all hard-to-detect faults, contributing to early fault detectability. Similar results were obtained with regard to DPCA. Compared to the supervised approach, RL proved to be more flexible regarding model parameterization, which can be very useful for practical aspects of industrial fault detection. The results show that RL-based approaches can be very useful for fault detection in continuous, non-linear, and time-varying systems, such as chemical processes. For example, with regard to adaptive alarm thresholds and the detection of known and unknown faults. The latter represents a major challenge in industrial practice.

In this study, the problem of fault detection for discrete systems is considered. For control inputs, unknown bounded disturbances and sensor faults, an observer‐based fault detection filter is designed based on a descriptor system method such that the error dynamic system is convergent, the effect of the disturbances on the residuals satisfies the H ∞ performance index and the effect of the faults on the residuals satisfies the H − performance index. By a Lyapunov functional approach, a sufficient condition for the solvability of the fault detection problem is expressed in terms of linear matrix inequalities. A numerical example shows the effectiveness of the proposed method.

This paper develops a novel fault reconstruction (FR) method and an FR-based fault-tolerant control (FTC) scheme for systems suffering from both sensor and actuator faults based on the combination of a Luenberger-like reduced-order observer and an interval observer. Firstly, by introducing an output transformation, an auxiliary output that is able to decouple the sensor fault is obtained. Secondly, for addressing the external disturbance and actuator fault, a multiple unknown input (MUI) is formed, and a reduced-order observer that is able to decouple the MUI is constructed. Consequently, asymptotic convergence estimations of the state and the sensor fault can be accomplished. Thirdly, in order to obtain the asymptotic convergence actual FR (AFR), an interval observer is designed. After this, an algebraic connection of the MUI and the state error estimation is given, and, based on the algebraic relationship, an algebraic MUI reconstruction (MUIR) method is proposed. Finally, an FTC scheme is developed by using the state estimation and MUIR. Under the FTC, the closed-loop system is asymptotically stable even if it suffers from sensor and actuator faults simultaneously. Theoretical analysis demonstrates that the observer-based FTC mechanism satisfies the separation principle. At last, two simulation examples are given to verify the effectiveness of the proposed methods.

In this article, boundary control strategy is adopted to suppress the vibration of a flexible three‐dimensional (3D) marine riser simultaneously against both sensor and actuator faults. The flexible 3D marine riser is a distributed parameter system modeled as partial differential equations. During normal operation of the system, all the system components will behave optimally. However, if the sensor and actuator appear fault, the system will work abnormally. While the sensor is faulty, the output signal of the sensor will have errors, and this error signal will be the input signal of the actuator. In terms of actuator fault, once the actuator appears fault problem, the actuator should make up for the fault and continue working for the sake of safety. In this work, the fault degree of sensors and actuators are unknown and the boundary control schemes using Nassbaum function are developed for fault compensation. The system stability is proven by introducing an appropriate Lyapunov function. Finally, the simulation results verify the accuracy of the proposed control laws.

Fault detection, isolation and estimation for Takagi–Sugeno nonlinear systems

Design Method of Fault Detector for Injection Unit

This paper presents the design and analysis of a methodology for detecting and isolating multiple sensor faults in large-scale interconnected nonlinear systems. The backbone of the proposed decentralized methodology is the design of a local sensor fault diagnosis agent dedicated to each interconnected subsystem, without the need to communicate with neighboring agents. Each local sensor fault diagnosis agent is responsible for detecting and isolating multiple faults in the local set of sensors. The local sensor fault diagnosis agent consists of a bank of modules that monitor smaller groups of sensors in the corresponding local sensor set. The detection of faults in each of the sensor groups is conducted using robust analytical redundancy relations, formulated by structured residuals and adaptive thresholds. The multiple sensor fault isolation in each local sensor fault diagnosis agent is realized by aggregating the decisions of the modules and applying a diagnostic reasoning-based decision logic. The performance of the proposed diagnostic scheme is analyzed with respect to sensor fault detectability and multiple sensor fault isolability. A simulation example of two interconnected robot manipulators is used to illustrate the application of the multiple sensor fault detection and isolation methodology.