For Fault Detection And Isolation Research Articles

Variance-capturing forward-forward autoencoder (VFFAE): A forward learning neural network for fault detection and isolation of process data

Data-driven models have emerged as popular choices for fault detection and isolation (FDI) in process industries. However, real-time updating of these models due to streaming data requires significant computational resources, is tedious and therefore pauses difficulty in fault detection. To address this problem, in this study, we have developed a novel forward-learning neural network framework that can efficiently update data-driven models in real time for high-frequency data without compromising the accuracy. The neural network parameters are updated using a suitably constructed forward-forward learning algorithm instead of the traditional backpropagation algorithm. Firstly, we develop a variance-capturing forward-forward autoencoder (VFFAE) for FDI. Further, we showcase that the previously trained VFFAE model can be quickly adapted to incoming data which demonstrate the efficacy of the proposed framework. We have three process case studies to validate the proposed approach, namely, the Tennesse-Eastman dataset, nuclear power flux dataset, and wastewater plant dataset, to validate the proposed approach. Our findings demonstrate that within the initial 90 s, the model underwent 90 updates using a forward-forward approach and only 10 updates using backpropagation-based methods without compromising accuracy. This highlights the model’s capacity to effectively handle streaming data during the modeling process.

On the design of an unknown input observer to fault detection, isolation, and estimation for uncertain multi-delay nonlinear systems

In this article, an unknown input observer (UIO) is proposed for fault detection and isolation (FDI) for uncertain nonlinear time-delay systems with multiple discrete delays. The presented UIO can be able to isolate both actuator and sensor faults and estimate them to identify the shape, time of occurrence, amplitude, and other features of the faults. Furthermore, the proposed FDI method is robust to both input/state and output disturbances. At first, the system dynamics and faults are augmented and then UIO is driven. A delay-independent Lyapunov–Krasovskii functional is utilized to guarantee the stability of the proposed UIO-based FDI method. The terms of stability and UIO gains are obtained by solving some feasible linear matrix inequalities (LMI). The proposed method is simulated on two interconnected well-mixed non-isothermal continuous stirred tank reactors (CSTRs) with recycling where there are three parallel, irreversible elemental exothermic reactions. Two point-wise time delays in the output temperature and concentration in both CSTRs are considered. Furthermore, parametric uncertainties in the modeling are investigated. The simulation and numerical results show the superiority of the proposed UIO for fault detection, isolation, and estimation in the presence of multiple discrete time delays, model uncertainty, and input/state and output disturbances.

Expert system for FDI of dc motor faults using structured residuals design technique

A major concern in many electrical drives is the reliability of sensors and actuators. In the paper, the usage of the Drools expert system (ES) for Fault detection and isolation (FDI) of the additive actuator and sensor DC Motor faults using the Structured residuals design technique (SRDT) is presented. The SRDT is used to obtain essential knowledge about the system. Afterward, an expert system that can isolate faults based on the developed structure matrix and generated residuals is designed. Accordingly, following the structure matrix each residual becomes able to answer to a desired subset of faults and stands insensitive to the others. The proposed method is successfully applied in an analyzed laboratory system and can be used for online FDI.

Multiple fault detection and isolation using artificial neural networks in sensors of an internal combustion engine

Technology advances rapidly in vehicles with internal combustion engines; therefore, detecting vehicles’ faults becomes more complex, primarily due to the increased usage of sensors and actuators to improve engine performance. Specialized instruments for detecting vehicle faults can sometimes show errors in multiple sensors when there is only one issue in the system, and they are not responsible for isolating faults. Then, it must add superior algorithms for fault detection and isolation (FDI) in the Internal Combustion Engine to these diagnostic instruments. For this reason, this research focuses on designing an individual and multiple fault detection and isolation system based on artificial neural networks for an internal combustion engine. The proposed scheme detects individual and multiple faults in the throttle position sensors (TPS), mass air-flow (MAF), and manifold absolute pressure (MAP). The FDI system uses five artificial neural networks (ANN). Each ANN estimates the value of two sensors using the engine speed and the air–fuel ratio (AFR) signal to generate analytical redundancy. When a sensor fault occurs, the FDI system substitutes the faulty signal with a healthy signal estimate provided by an ANN. Suppose a second fault arises in another sensor. The FDI system can replace the faulty signal with the ANN’s estimated signal, allowing the internal combustion engine to run even with multiple faults.

A Cost-Efficient MCSA-Based Fault Diagnostic Framework for SCIM at Low-Load Conditions

In industry, electric motors such as the squirrel cage induction motor (SCIM) generate motive power and are particularly popular due to their low acquisition cost, strength, and robustness. Along with these benefits, they have minimal maintenance costs and can run for extended periods before requiring repair and/or maintenance. Early fault detection in SCIMs, especially at low-load conditions, further helps minimize maintenance costs and mitigate abrupt equipment failure when loading is increased. Recent research on these devices is focused on fault/failure diagnostics with the aim of reducing downtime, minimizing costs, and increasing utility and productivity. Data-driven predictive maintenance offers a reliable avenue for intelligent monitoring whereby signals generated by the equipment are harnessed for fault detection and isolation (FDI). Particularly, motor current signature analysis (MCSA) provides a reliable avenue for extracting and/or exploiting discriminant information from signals for FDI and/or fault diagnosis. This study presents a fault diagnostic framework that exploits underlying spectral characteristics following MCSA and intelligent classification for fault diagnosis based on extracted spectral features. Results show that the extracted features reflect induction motor fault conditions with significant diagnostic performance (minimal false alarm rate) from intelligent models, out of which the random forest (RF) classifier was the most accurate, with an accuracy of 79.25%. Further assessment of the models showed that RF had the highest computational cost of 3.66 s, while NBC had the lowest at 0.003 s. Other significant empirical assessments were conducted, and the results support the validity of the proposed FDI technique.

Analysis and implementation of certain multivariate statistical process monitoring Tools for fault detection and isolation (FDI) task in a laboratory scale shell-tube heat exchanger

Early detection and diagnosis of faulty events in industrial processes can represent economic, social and environmental profits. When the process has a great quantity of sensors or actuators, the Fault Detection and Isolation (FDI) task is very difficult. Advanced statistical based FDI methods are extensively used for fault detection and isolation purposes. In this work, three multivariate statistical techniques such as neural network based Principal Component Analysis (PCA), neural network-based Fischer Discriminant Analysis (FDA) and Correspondence Analysis (CA) was applied to the multivariate data extracted from laboratory scale shell and tube heat exchanger. Performance metric such as detection delay, estimated time of occurrence of fault, misclassification rate was computed for those three techniques for the detection and isolation of biases in sensors and actuators. Correspondence Analysis was proven to perform better when compared to PCA and FDA. CA was observed to perform FDI with minimal detection delay (which is less than or equal to 7 seconds) and lower misclassification rate (which is less than or equal to 6%) in case of both sensor & actuator faults. PCA and FDA showed significant detection delay and missed alarm rate for single and multiple fault identification.

A novel integrated framework for fault diagnosis with application to process safety

A novel integrated framework for Fault Detection and Isolation(FDI) is proposed, with applications to process safety, by sequentially integrating model-free(MFA) and model-based(MBA) approaches. The MFA includes Limit Checking/Visual and Plausibility analysis, Artificial Neural Network and Fuzzy Logic, Adaptive Neuro-Fuzzy Inference System. The MBA uses a Linear Parameter-Varying model to handle a wide class of generally-nonlinear physical systems. The adaptive Kalman filter(KF) residuals are used for FDI in the MBA. Novel emulators, cascaded with the system during off-line data acquisition, system identification and Bayes’measure of belief computation for each FDI scheme, have their parameters perturbed at each operating point, to mimic unforeseen operational scenarios, thus covering all operating regions. Critical information about the presence/absence of a fault is quickly gained via the faster FDI scheme. A more accurate subsystem’s status is unfolded sequentially by the slower FDI scheme. The final decision on the fault status is obtained using a weighted Bayes classifier fusion scheme meeting the critical requirements of high(low) probability of correct decision (false alarm). Implications of FDI in process safety/environment protection are discussed. This framework is successfully evaluated on simulated and physical systems, including benchmarked laboratory-scale two-tank system, by detecting and isolating sensor, actuator and leakage faults.

Inverse fuzzy fault model for fault detection and isolation with least angle regression for variable selection

Fault detection is paramount in industrial processes because expensive reparations can be avoided, and the normal flow of operations is not disrupted. However, it is difficult to use a model-based method for fault detection in complex systems. Thus, a data-driven approach is implemented in the Tennessee Eastman process for fault detection and isolation (FDI). As a contribution, this paper proposes an inverse fuzzy fault model to detect and isolate faults. To reduce the amount of data to process, the least angle regression is applied for variable selection. To compare the detection and isolation times obtained using the fuzzy fault model, a fuzzy classifier is described; where the signals are preprocessed with the wavelet transform to highlight the faulty signals. The inverse fuzzy fault model has only four fuzzy rules and shows a smaller isolation time than the required using the fuzzy classifier.

Model-Based Fault Detection and Isolation in DC Microgrids Using Optimal Observers

DC microgrids require advanced protection techniques for fault detection and isolation (FDI). In this work, an FDI method able to respond to different types of component faults is developed based on system modeling. First, the state-space representation of a multiterminal dc microgrid with component faults is derived. Then, an FDI function based on <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mathcal {H}}_{-}/{\mathcal {H}}_{\infty }$ </tex-math></inline-formula> observers is designed. To achieve the desired selectivity in fault isolation, the linear matrix inequality (LMI) optimization approach is adopted in the observer design. The performance of the proposed FDI method is verified under the real-time (RT) simulation of a three-terminal low-voltage dc microgrid and with a small-scale laboratory dc grid. The proposed FDI method is proved to be effective to detect and isolate different faults in dc microgrids with a response time of 1 ms.

Joint Distribution-Based Test Selection for Fault Detection and Isolation Under Multiple Faults Condition

Effective and comprehensive fault information can reduce the maintenance cost of a system. The selection of a proper test is important to obtain accurate information for fault detection and isolation (FDI) once a fault occurs. In real systems, the FDI task is made difficult by the correlations in the physical behaviors of the components and the possible coexistence of multiple faults. In this article, a new type of test selection model based on deep joint distribution is proposed, which takes into account the dependence and ambiguity groups. The optimal test can, then, be selected by an improved discrete binary particle swarm optimization (IBPSO) algorithm. The strengths of the proposed optimal test selection strategy are: 1) it can greatly improve the accuracy of test selection; 2) it considers the influence of multiple faults; 3) the dependence problem of the test outcomes can be addressed; and 4) it is computationally efficient and capable of selecting optimal test points meeting the requirements of real-time FDI. The application to a negative feedback circuit demonstrates the performance of the proposed method.

Design of linear residual generators for fault detection and isolation in nonlinear systems

A systematic method for the design of disturbance-decoupled linear residual generators for fault detection and isolation (FDI) in nonlinear systems is developed. Necessary and sufficient conditions for the existence of linear residual generators for nonlinear systems are derived. As long as these conditions are satisfied, we obtain explicit design formulas for the residual generator, with eigenvalue assignment. The proposed formulation and results provide a nonlinear generalisation of standard FDI methods for linear systems. The method is applied to three case studies: a bio reactor, a continuous stirred tanked reactor (CSTR) and a process network.

A Neural Adaptive Approach for Active Fault-Tolerant Control Design in UAV

Faults in aircraft actuators can cause serious issues in safety. Due to the autonomous nature of the unmanned aerial vehicles (UAVs), faults can lead to more serious problems in these systems. In this paper, a new active fault tolerant control (FTC) system design for an UAV is presented. The proposed design uses a neural network adaptive structure for fault detection and isolation (FDI), then, the FDI signal combined with a nonlinear dynamic inversion technique is used to compensate for the faults in the actuators. The proposed scheme detects and isolates faults in the actuators of the system in real-time without the need of controller reconfiguration in presence of faults in the actuators. The simulation results of the designed FTC system when it is applied to WVU YF-22 UAV show that the proposed design can successfully detect and isolate the faults and compensate for their effect.

Multi-Fault Diagnosis of an Aero-Engine Control System Using Joint Sliding Mode Observers

An aero-engine is a complex aerodynamic thermal system, which can operate in extreme environments for long periods. It is crucial to diagnose any faults of the aero-engine control system accurately. At present, most aero-engine control system fault diagnosis schemes suffer from large interference, significant chattering, and low estimation accuracy. To diagnose multi-faults of the control system effectively, we introduce and investigate a new fault diagnosis scheme in this paper, which uses joint sliding mode observers. First, we develop a mathematical model for multi-faults in the control system, which can describe actuator and sensor faults in detail. Then, we design the joint sliding mode observers for fault detection and isolation (FDI), using the sliding mode variable structure term to reduce the coupling effect. Finally, during the fault estimation process, we use a pseudo-sliding form to reduce the chattering problem and suppress the impact of interference, which leads to an accurate estimation of the multi-fault characteristics. The simulation results show that, the proposed scheme can effectively detect and isolate faults, which enables superior timeliness and accuracy compared to a conventional sliding mode observer scheme. During the process of fault estimation, the effect of chattering is reduced, which shows the advantages of strong sensitivity and high estimation accuracy.

A new approach for the diagnosis of different types of faults in dc–dc power converters based on inversion method

This paper presents theory, a new approach and validation results for fault detection and isolation (FDI) in dc–dc power converters, based on inversion method. The developed method consists on the inversion-based estimation of faults and change detection mechanisms adapted to the power converters context. With the inverse model of a switched linear system, we have designed a real-time FDI algorithm with an integrated fuzzy logic scheme which detects and isolates abrupt changes (faults) at unknown time instants. A smoothing strategy is used to attenuate the effect of unknown disturbances and noise that are present at the outputs of this inverse model. Once the fault event is detected, a dedicated fuzzy-logic-based scheme is proposed to isolate the four types of faults: switch, voltage and current sensor, and capacitor. The performance of the proposed method is verified experimentally to detect and isolate the mentioned faults in the dc–dc boost power converter.

A fault detection and isolation technique using nonlinear support vectors dichotomizing multi-class parity space residuals

Parity equation based residual generation with a statistical decision-making scheme has been proved to be effective for fault detection and isolation (FDI) in process systems. However, to implement a statistical decision-making scheme on residuals generated through parity equations, a multivariate probability distribution function of the residuals must be identified for each fault class. Often, the statistical properties of these residuals depend on the noise characteristics associated with the measurement systems, in addition to modeling errors. However, identification of the correct probability distribution function for the noise is a nontrivial task. The issue is further complicated due to the effects of the noise propagating through the transformation designed for minimizing the effects of the normal operating envelope and disturbances. To overcome such problems, a support vector machine (SVM) classification algorithm can be used in lieu of a statistical decision-making scheme. Optimal fault classifications can be achieved under the SVM framework using properly designed hyper-planes. To process the residuals using SVMs, several additional issues have to be dealt with, such as the non-linear classification problem, over-fitting issues, and multi-class fault classification. The objective of this paper is to describe a framework within which SVMs can be effectively integrated with parity equation based residual generation schemes. The developed framework has been evaluated on a physical process control platform under various fault scenarios. It has been shown that the scheme is feasible and particularly suitable for situations where it is difficult to obtain probability distribution functions for all potential fault classes.

Fault Detection and Isolation in Electrical Machines using Deep Neural Networks

Condition and health monitoring of electrical machines during dynamic loading is a common, yet challenging problem in main battle tanks. Existing methods address this issue by extracting various features which are subsequently used in a classifier to isolate faults. However, this approach relies on the feature set being extracted and therefore most of the time does not provide expected accuracy in identification of faults. In this work, we have used convolution neural network that utilises the original time domain measurements for fault detection and isolation (FDI). Results from experimental studies indicate that the proposed approach can perform FDI with more than 95\% accuracy using commonly available current measurements.

Novel Non-Model-Based Fault Detection and Isolation of Satellite Reaction Wheels Based on a Mixed-Learning Fusion Framework

Abstract This paper suggests a model-free framework for Fault Detection and Isolation (FDI) of satellite reaction wheels for the first time. The proposed FDI method is based on multi-classifier fusion with diverse learning algorithms and configured in a parallel form where a unique module simultaneously performs both detection and isolation tasks. In other words, a multi-classifier-based arrangement is presented on the basis of Mixed Learning strategy where four classic and well-practised classification schemes including Random Forest, Support Vector Machine, Partial Least Square, and Naive Bayes are incorporated into FDI module in order to make a decision on the occurrence of a fault and its location. Extensive simulation results with a high-fidelity nonlinear spacecraft simulator considering gyroscopic effects, measurement noise, and exogenous aerodynamic disturbance signals show that the proposed FDI scheme can cope with faults affecting reaction wheel torques and obtain promising FDI performances in most of the designed scenarios.

Fault-tolerant VSI for stand-alone DFIG feeding an isolated DC load

ABSTRACTIn this article, a fault-tolerant voltage source inverter (VSI) is proposed for a novel topology of stand-alone doubly fed induction generator (DFIG) feeding an isolated DC load. In this topology, stator and rotor sides of DFIG are connected to DC-bus through a diode rectifier and VSI, respectively. The fault-tolerant VSI includes a redundant leg connected by bidirectional switches in order to replace the faulted leg and improve the reliability of the proposed system. The field oriented control strategy is adopted to control the d and q-axis rotor currents in order to maintain the voltage and the frequency at the output of the generator constant. A novel, easy and fast approach for fault detection and isolation (FDI) of open-switch damage in insulated gate bipolar transistor-based VSI is proposed in this study. This approach is developed using mathematical transformations such as a hysteresis detector, an integrator and a trigger. The FDI algorithm proposed here does not require the knowledge of the system model and is independent from its complexity. Simulations results are illustrated for a 3.7 kW DFIG feeding a DC load with open-switch fault in VSI that confirm the concepts proposed in this study.

Model-free fault detection and isolation of a benchmark process control system based on multiple classifiers techniques—A comparative study

This paper presents a combined data-driven framework for fault detection and isolation (FDI) based on the ensemble of diverse classification schemes. The proposed FDI scheme is configured in series and parallel forms in the sense that in series form the decision on the occurrence of fault is made in FD module, and subsequently, the FI module coupled to the FD module will be activated for fault indication purposes. On the other hand, in parallel form a single module is employed for FDI purposes, simultaneously. In other words, two separate multiple-classifiers schemes are presented by using fourteen various statistical and non-statistical classification schemes. Furthermore, in this study, a novel ensemble classification scheme namely blended learning (BL) is proposed for the first time where single and boosted classifiers are blended as the local classifiers in order to enrich the classification performance. Single-classifier schemes are also exploited in FDI modules along with the ensemble-classifier methods for comparison purposes. In order to show the performance of proposed FDI method, it was also tested and validated on DAMADICS actuator system benchmark. Besides, comparative study with the related works done on this benchmark is provided to show the pros and cons of the proposed FDI method.

Stabilization of linear impulsive systems under dwell-time constraints: Interval observer-based framework

The problem of interval observer design is studied for a class of linear impulsive systems. Ranged and minimum dwell-time constraints are considered under detectability assumption. The first contribution of this paper lies in designing interval observers for linear impulsive systems under ranged and minimum dwell-time constraints, and investigating positivity of the estimation error dynamics in addition to stability. Several observers are designed oriented on different conditions of positivity and stability for estimation error dynamics. The boundedness of the estimation error (input-to-state stability property) and the observer stability conditions are stated as infinite-dimensional linear programming problems. Next, an output stabilizing feedback design problem is discussed, where the stability is checked using linear matrix inequalities (LMIs). Efficiency of the proposed approach is demonstrated by computer simulations for a commercial electric vehicle equipped with a low power range extender fuel cell, a bouncing ball, an academic linear impulsive system and for Fault Detection and Isolation (FDI) and Fault-Tolerant Control (FTC) of a power split device with clutch for heavy-duty military vehicles.