Multiple Fault Isolation Research Articles

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

The goal in many fault detection and isolation schemes is to increase the isolation and identification speed. This paper, presents a new approach of a nonlinear model based adaptive observer method, for detection, isolation and identification of actuator and sensor faults. Firstly, we will design a new method for the actuator fault problem where, after the fault detection and before the fault isolation, we will try to estimate the output of the instrument. The method is based on the formation of nonlinear observer banks where each bank isolates each actuator fault. Secondly, for the sensor problem we will reformulate the system by introducing a new state variable, so that an augmented system can be constructed to treat sensor faults as actuator faults. A method based on the design of an adaptive observers’ bank will be used for the fault treatment. These approaches use the system model and the outputs of the adaptive observers to generate residues. Residuals are defined in such way to isolate the faulty instrument after detecting the fault occurrence. The advantages of these methods are that we can treat not only single actuator and sensor faults but also multiple faults, more over the isolation time has been decreased. In this study, we consider that only abrupt faults in the system can occur. The validity of the methods will be tested firstly in simulation by using a nonlinear model of waste water treatment process with and without measurement noise and secondly with the same nonlinear model but by using this time real data.

Fuzzy Nonlinear Unknown Input Observer Design with Fault Diagnosis Applications

An approach for nonlinear unknown input observer (NUIO) design is proposed for a class of nonlinear systems representable by a Takagi—Sugeno (TS) fuzzy system. The proposed NUIO design for TS fuzzy systems is carried out for two cases: (1) the premise variables do not depend on the unmeasured state variables; and (2) the premise variables depend on the unmeasured state variables. Sufficient conditions for the existence of NUIOs are derived, and a linear matrix inequality (LMI)-based design strategy is presented for NUIO design purposes. The proposed NUIO design approach is then applied to solve actuator fault detection and isolation problems for nonlinear systems described by TS fuzzy systems. To this end, a system structure with two groups of inputs where one group of inputs is treated as unknown inputs is developed. Based on the system structure, a bank of NUIOs are then designed using the developed NUIO design approach in order to investigate the following fault diagnosis problems. (1) How can the NUIOs be used for detecting faults? (2) Under what conditions is it possible to isolate single and/or multiple faults? (3) What is the maximum number of faults that can be isolated simultaneously? (4) How can multiple fault isolation be achieved? In this article we present a NUIO-based fault-detection scheme for problem (1), give sufficient conditions for problem (2), determine the maximum number of faults that can be isolated for problem (3), and propose a fault-diagnosis scheme using a bank of NUIOs to solve problem (4). As an illustrative example, Lorenz’s chaotic system with multi-inputs is chosen to show the effect of the designed NUIOs and the proposed fault detection and isolation scheme. Simulation results show that accurate state estimation is achieved and actuator faults can be detected and isolated successfully.

Multiple sensor fault detection and isolation of an air quality monitoring network using RBF-NLPCA model

This paper presents a data-driven method based on non-linear principal component analysis to detect and isolate multiple sensor faults. The RBF-NLPCA model is obtained by combining a principal curve algorithm and two three-layer radial basis function (RBF) networks. The reconstruction approach for multiple sensors is proposed in the non-linear case and successfully applied for multiple sensor fault detection and isolation of an air quality monitoring network. The proposed approach reduces considerably the number of reconstruction combinations and allows to determine replacement values for the faulty sensors.

A New Adaptive Kalman Estimator for Detection and Isolation of Multiple Faults Integrated in a Fault Tolerant Control

For sequential jumps detection, isolation, and estimation in discrete-time stochastic linear systems, Willsky and Jones (1976) have developed the Generalized Likelihood Ratio (GLR) test. After each detection and isolation of one jump, the treatment of another possible jump is obtained by a direct state estimate and covariance incrementation of the Kalman filter originally designed on the jump-free system. This paper proposes to extend this approach from a state estimator designed on a reference model directly sensitive to system changes. We will show that the obtained passive GLR test can be easily integrated in a Fault Tolerant Control System (FTCS) via a control law designed in order to asymptotically reject the effect of sequential jumps.

Multiple sensor fault detection and isolation of an air quality monitoring network using RBF-NLPCA model

This paper presents a data-driven method based on nonlinear principal component analysis to detect and isolate multiple sensor faults. The RBF-NLPCA model is obtained by combining a principal curve algorithm and two three layer radial basis function (RBF) networks. The reconstruction approach for multiple sensors is proposed in the non linear case and successfully applied for multiple sensor fault detection and isolation of an air quality monitoring network. The proposed approach reduces considerably the number of reconstruction combinations and allows to determine replacement values for the faulty sensors.

WWTP diagnosis based on robust principal component analysis

Principal component analysis (PCA) is a powerful fault detection and isolation method. However, the classical PCA which is based on the estimation of the sample mean and variance-covariance matrix of the data is very sensitive to outliers in the training data set. Usually robust principal component analysis was applied to remove the effect of outliers on the PCA model. In this paper, a fast two-step algorithm is proposed. First, the objective was to find a robust PCA. Hence a scale-M estimator is computed using an iterative re-weighted least squares (IRWLS) procedure. This algorithm is initialized from a nearly robust variance-covariance estimate which tends to emphasize the contribution of close observations in comparison with distant observations (outliers). Second, structured residuals are used for multiple fault detection and isolation. These structured residuals are based on the reconstruction principle and the existence condition of such residuals is used to determine the detectable faults and the isolable faults. The proposed scheme avoids the combinatorial explosion of faulty scenarios related to multiple faults to consider. Then, this procedure is successfully applied for sensor fault detection and isolation of the hydraulic part of an activated sludge wastewater treatment plant (WWTP).

Multiple and simultaneous fault isolation with minimal fault indicating signals: a case study

This paper presents a formalized isolation technique for multiple and simultaneous failure diagnosis in the orbiter measurement unit during the rendezvous phase of the Mars Sample Return (MSR) mission. The MSR mission is a spacecraft mission whose aim is to bring tangible samples from Mars to Earth for analysis. The proposed methodology aims at combining judiciously available hardware redundancy-based fault indicating signals with analytic-based fault indicating signals (i.e. residuals). The paper deals with the problem of finding the minimal number of analytic-based fault indicating signals for a given isolation task. To cope with the diagnosis problem, logical tools derived from the set theory are used. Simulation results are presented in order to prove the effectiveness of the proposed method.

Multiple Fault Isolation to a Servo-Valve Controlled Motor Transmission System

Model based quantitative fault diagnosis procedure is based on generation of fault indicators, structural analysis and finally estimation of parameters. In this article, bond graph modelling is used to describe the system model and generate fault indicators by evaluating a set of Analytical Redundancy Relations (ARR), which are obtained from differentially causalled bond graph model to satisfy inverse system dynamics. Multiple fault case is realised by changing the parameters simultaneously. Thereafter the parameters related only to unstructured part of the fault subspace are estimated by optimising an objective function containing ARRs. Sometimes the derivative form of ARRs creates problem of singularity in estimation, and to avoid such a singularity problem we have estimated the corresponding parameters directly from constraint relations and thereafter those values are called in the main objective function to estimate rest of the parameters. The algorithm provides quicker fault isolation because it does not need several model simulations; thereby making it suitable for real-time process supervision.

Fault Detection and Isolation with Robust Principal Component Analysis

Fault Detection and Isolation with Robust Principal Component AnalysisPrincipal component analysis (PCA) is a powerful fault detection and isolation method. However, the classical PCA, which is based on the estimation of the sample mean and covariance matrix of the data, is very sensitive to outliers in the training data set. Usually robust principal component analysis is applied to remove the effect of outliers on the PCA model. In this paper, a fast two-step algorithm is proposed. First, the objective was to find an accurate estimate of the covariance matrix of the data so that a PCA model might be developed that could then be used for fault detection and isolation. A very simple estimate derived from a one-step weighted variance-covariance estimate is used (Ruiz-Gazen, 1996). This is a "local" matrix of variance which tends to emphasize the contribution of close observations in comparison with distant observations (outliers). Second, structured residuals are used for multiple fault detection and isolation. These structured residuals are based on the reconstruction principle, and the existence condition of such residuals is used to determine the detectable faults and the isolable faults. The proposed scheme avoids the combinatorial explosion of faulty scenarios related to multiple faults to be considered. Then, this procedure for outliers detection and isolation is successfully applied to an example with multiple faults.

A sliding mode observer-based strategy for fault detection, isolation, and estimation in a class of Lipschitz nonlinear systems

This article investigates the design and application of a sliding mode observer (SMO) strategy for actuator as well as sensor fault detection, isolation, and estimation (FDIE) problem for a class of uncertain Lipschitz nonlinear systems. Actuator FDIE is addressed by regrouping the system's inputs into a structure suitable for SMO design. Similarly, by filtering the regrouped outputs, a similar system structure can be developed for sensor FDIE problem. Once in the suitable form and under certain assumptions, nonlinear SMOs are proposed for actuator and sensor FDIE. A systematic LMI-based design approach for the proposed SMO is presented. Additionally, the article addresses four problems, namely: (P1) What are the conditions for isolating single and/or multiple faults? (P2) What is the maximum number of faults that can be isolated simultaneously? (P3) How should one design SMO-based FDI approach in order to achieve multiple fault isolation using as few observers as possible? (P4) How can one estimate the shape of the faults? To solve the above problems, a new concept called fault isolation index (FIX) is proposed for actuator and sensor FDIE. It is proved that fault isolation can only be achieved if FIX ≠ 0, and also that the maximum number of faults that can be isolated is equal to FIX. Using the proposed fault isolation strategy and by treating some healthy inputs or outputs as unknown inputs, a systematic FDIE design scheme using a bank of nonlinear SMOs, which provides a solution for the four problems is provided. An example is used to illustrate the proposed ideas. The simulation results show that the proposed FDIE scheme can successfully detect and isolate both slowly and fast-changing actuator faults. It is also shown that accurate estimation of actuator faults can be achieved.

Sensitivity bond graph approach to multiple fault isolation through parameter estimation

In model-based quantitative multiple fault detection and isolation (FDI), fault disambiguation is based on parameter estimation. In this paper, the fault hypothesis is generated by evaluating a set of analytical redundancy relations (ARRs) and parameter values corresponding to the unstructured part of the fault subspace are estimated by minimizing a function of the ARRs. Process and measurement uncertainties are handled by using a passive approach for robust FDI. Bond graph modelling is used to describe process models and to derive the ARRs. The bond graph model of the process is differentially causalled and it is then converted into a diagnostic bond graph form. The diagnostic bond graph is further converted into its corresponding sensitivity bond graph form, which gives the residual sensitivity to parametric changes. The developed algorithm provides quicker fault isolation because only a few parameters are estimated and it does not need several model simulations, thereby making it suitable for real-time process supervision.

DESIGN OF STRUCTURED RESIDUALS USING INTERVAL MODELS: APPLICATION TO MULTIPLE SEQUENTIAL FAULT ISOLATION IN SENSOR NETWORKS

In this paper, the design of a structured residual set to deal with multiple sequential fault isolation is proposed in case that the underlying model describing the system to be monitored is an “interval model”. In industrial applications, this type of fault is the most common since the continuous operation of systems/processes is required. An example based on the rain-gauge sensor network of the Barcelona sewer system will be used to illustrate the effectiveness of the proposed approach.

Unscented 칼만필터를 이용한 관성센서 복합 고장검출기법

2자유도 관성센서는 두 입력축이 기계적으로 연관되어 있기 때문에 해당하는 관성센서의 두 입력축에 동시에 고장이 발생할 확률이 매우 높다. 따라서 하드웨어 여분만으로 고장검출 및 분리를 수행하기 위해서는 최소한 4개의 관성센서를 사용하여야 한다. 2자유도 관성센서를 3개 중첩해서 사용하는 경우 기존의 하드웨어 여분기법으로는 고장검출은 가능하나 고장분리가 불가능하다. 본 논문에서는 이러한 문제점을 개선하기 위해서 비선형 필터인 Unscented Kalman Filter를 이용하여 얻은 정보를 해석적 여분으로 활용하여, 하드웨어 여분과 해석적 여분을 동시에 고려한 복합 고장검출기법을 제안하였다. 제안한 복합 고장검출기법의 성능을 검증하기 위해서 비선형 항공기 수치 시뮬레이션을 수행하였다. In two-degree of freedom(TDOF) inertial sensors, two axes are mechanically correlated with each other. Fault source of one axis sensor may affect the other axis sensor, and therefore multiple fault detection and isolation(FDI) technique is required. Conventional FDI techniques using hardware redundancy need four TDOF inertial sensors for FDI. In this study, three TDOF inertial sensor redudancy case is considered, where conventional FDI technique can detect the fault, but cannot isolate the fault sensor. Hybrid FDI technique is proposed to solve this problem. Hybrid FDI technique utilizes the analytic redundancy by utilizing the unscented kalman filter as well as hardware redundancy for FDI. To verify the effectiveness of the proposed FDI technique, numerical simulations are performed using six degree of freedom nonlinear aircrft dynamics.

HYBRID FAULT DETECTION AND ISOLATION METHOD FOR UAV INERTIAL SENSOR REDUNDANCY MANAGEMENT SYSTEM

Redundant system with three two-degree-of-freedom (2-DOF) inertial sensors is one of the possible candidates for the UAV inertial sensor redundancy management system. In the conventional FDI techniques using hardware redundancy, at least four 2-DOF inertial sensors are needed to detect and isolate the faulty sensor. Since two input axes of 2-DOF inertial sensors are mechanically correlated with each other, the fault of one axis sensor can affect the fault of the other axis sensor. When three sensors are used, singular direction problem can be occurred in detecting process, where false alarm may be declared such that two sensors have bias fault although only one sensor is out of order. Therefore, the study of multiple fault detection and isolation (FDI) technique is required to deal with these problems. In this study, a hybrid FDI technique is proposed for multiple FDI of three 2-DOF sensor system. The proposed FDI algorithm is based on hardware redundancy and is combined with an analytic redundancy by utilizing the unscented Kalman Filter. Numerical simulations are performed to verify the effectiveness of the proposed FDI technique.

Simultaneous Fault Diagnosis and Robust Model Selection in Multiple Linear Models Framework

In this paper, the main goal is to design an approach that performs fault detection and isolation in non linear system. Fault diagnosis is established by regarding system as an interpolation of multiple linear time invariant stochastic models and not as a single global model. In multiple model framework, the purpose of the paper consist of generate a robust model selection of the "best" representative linear model. Fault diagnosis method presented here is based on a bank of decoupled Kalman filters. The proposed method allows detection, isolation and estimation of multiple faults which appear simultaneously or in a sequential way. Robust model selection is obtained by regarding the residual vector insensitive to faults. Performances of the method are tested on a simulation example.

Adaptive Filter Design for FDI in Nonlinear Systems Based on Multiple Model Approach

In this paper, fault diagnosis for a large class of nonlinear systems is developed: nonlinear systems are supposed to be represented via a multiple model approach based on a convex combination of linear time invariant (LTI) models. Under fault isolation conditions, the main contribution is to design an adaptive filter which makes possible multiple faults detection, isolation and estimation over the whole operating range of nonlinear system. According to nonlinear representation via a gain scheduling interpolation method (successfully used in the control design even with fast varying parameters), the design is performed by extending robust fault isolation filter methods in LTI systems to nonlinear systems. The stability conditions of the adaptive filter are developed. These conditions result in convex linear matrix inequalities (LMIs) that can be solved efficiently with optimization techniques. The performances of the methods are demonstrated using a nonlinear system simulation example.

Combining AI, FDI, and Statistical Hypothesis-Testing in a Framework for Diagnosis

A new framework for model based diagnosis is presented using ideas from AI, FDI, and statistical hypothesis testing. The isolation mechanism is based on AI methods, and the main advantage is that multiple faults are handled implicitly. Thus, no special care for isolation of multiple faults is needed. The methods for residual generation, developed in the field of control theory (FDI), can within the framework be fully utilized. Since the framework is also based upon statistical hypothesis testing, it is suitable for problems including noise.

On-Line Fault Isolation Algorithm for Industrial Processes

The paper presents the solutions of main problems that appear in diagnostics of large scale processes in chemical, petrochemical, pharmaceutical, power and other industry utilised in applied diagnostic software. In addition to description of reasoning rules based on fuzzy logic and taking into account the dynamics of symptoms forming the following topics where raised: decomposition of diagnostic system, dynamical creation of FI threads and multiple fault isolation assuming single fault scenarios.

Sequential testing algorithms for multiple fault diagnosis

We consider the problem of constructing optimal and near-optimal test sequences for multiple fault diagnosis. The computational complexity of solving the optimal multiple-fault isolation problem is super exponential, that is, it is much more difficult than the single-fault isolation problem, which, by itself, is NP-hard. By employing concepts from information theory and AND/OR graph search and by exploiting the single fault testing strategies of Pattipati et al. (1990), we present several test sequencing algorithms for the multiple fault isolation problem. These algorithms provide a trade-off between the degree of suboptimality and computational complexity. Furthermore, we present novel diagnostic strategies that generate a diagnostic directed graph, instead of a traditional diagnostic tree, for multiple fault diagnosis. Using this approach, the storage complexity of the overall diagnostic strategy reduces substantially. The algorithms developed herein have been successfully applied to several real-world systems.

Multiple Fault Detection and Isolation Using the Haar Transform, Part 2: Application to the Stamping Process

The sheet metal drawing operation is a complex manufacturing process involving more than forty process variables. The intricate interaction among these variables affect the forming tonnage which is measured by strain gages mounted on the press. A fault is said to occur when any of these process variables deviate beyond their specified limits. Current detection schemes based on thresholding do not fully exploit the information in the tonnage signals for the detection and isolation of multiple fault condition. It is thus an excellent case study for demonstrating the implementation of the detection methodology presented in Part 1. By partitioning the tonnage signature into disjoint segments, mutually exclusive sets of Haar coefficients can be used to isolate faults in each stage of the process.