Sensor fault diagnosis based on energy balance evaluation: application to a metal processing

This paper deals with the design of a residual generator for fault detection and isolation in the dynamic closed-loop systems based on the balance of energy which "enters" and "leaves" plants. The main contribution of this paper consists in developing a suitable fault detection and isolation technique to detect faults in single-input single-output closed-loop system based on major signals without the requirement of an accurate static or dynamic model. Indeed, in the absence of conventional input-output models, the proposed method involves the on-line energy balance evaluation to detect a sensor fault. The application to the monitoring of a galvanizing line in steel industry shows the effectiveness of the suggested approach when a sensor fault occurs.     

Fault detection and classification in chemical processes based on neural networks with feature extraction

Feed forward neural networks are investigated here for fault diagnosis in chemical processes, especially batch processes. The use of the neural model prediction error as the residual for fault diagnosis of sensor and component is analyzed. To reduce the training time required for the neural process model, an input feature extraction process for the neural model is implemented. An additional radial basis function neural classifier is developed to isolate faults from the residual generated, and results are presented to demonstrate the satisfactory detection and isolation of faults using this approach.     

Multi-thresholds for fault isolation in the presence of uncertainties

Monitoring of the faults is an important task in mechatronics. It involves the detection and isolation of faults which are performed by using the residuals. These residuals represent numerical values that define certain intervals called thresholds. In fact, the fault is detected if the residuals exceed the thresholds. In addition, each considered fault must activate a unique set of residuals to be isolated. However, in the presence of uncertainties, false decisions can occur due to the low sensitivity of certain residuals towards faults. In this paper, an efficient approach to make decision on fault isolation in the presence of uncertainties is proposed. Based on the bond graph tool, the approach is developed in order to generate systematically the relations between residuals and faults. The generated relations allow the estimation of the minimum detectable and isolable fault values. The latter is used to calculate the thresholds of isolation for each residual.     

Fault diagnosis for a class of nonlinear systems via ESO

In this paper, a novel fault detection and identification (FDI) scheme for a class of nonlinear systems is presented. First of all, an augment system is constructed by making the unknown system faults as an extended system state. Then based on the ESO theory, a novel fault diagnosis filter is constructed to diagnose the nonlinear system faults. An extension to a class of nonlinear uncertain systems is then made. An outstanding feature of this scheme is that it can simultaneously detect and identify the shape and magnitude of the system faults in real time without training the network compared with the neural network-based FDI schemes. Finally, simulation examples are given to illustrate the feasibility and effectiveness of the proposed approach.     

Multiple sensor fault diagnosis for dynamic processes

Modern industrial plants are usually large scaled and contain a great amount of sensors. Sensor fault diagnosis is crucial and necessary to process safety and optimal operation. This paper proposes a systematic approach to detect, isolate and identify multiple sensor faults for multivariate dynamic systems. The current work first defines deviation vectors for sensor observations, and further defines and derives the basic sensor fault matrix (BSFM), consisting of the normalized basic fault vectors, by several different methods. By projecting a process deviation vector to the space spanned by BSFM, this research uses a vector with the resulted weights on each direction for multiple sensor fault diagnosis. This study also proposes a novel monitoring index and derives corresponding sensor fault detectability. The study also utilizes that vector to isolate and identify multiple sensor faults, and discusses the isolatability and identifiability. Simulation examples and comparison with two conventional PCA-based contribution plots are presented to demonstrate the effectiveness of the proposed methodology.     

Sliding mode observer based incipient sensor fault detection with application to high-speed railway traction device

This paper considers incipient sensor fault detection issue for a class of nonlinear systems with “observer unmatched” uncertainties. A particular fault detection sliding mode observer is designed for the augmented system formed by the original system and incipient sensor faults. The designed parameters are obtained using LMI and line filter techniques to guarantee that the generated residuals are robust to uncertainties and that sliding motion is not destroyed by faults. Then, three levels of novel adaptive thresholds are proposed based on the reduced order sliding mode dynamics, which effectively improve incipient sensor faults detectability. Case study of on the traction system in China Railway High-speed is presented to demonstrate the effectiveness of the proposed incipient senor faults detection schemes.     

Robust fault detection for switched positive linear systems with time-varying delays

This paper investigates the problem of robust fault detection for a class of switched positive linear systems with time-varying delays. The fault detection filter is used as the residual generator, in which the filter parameters are dependent on the system mode. Attention is focused on designing the positive filter such that, for model uncertainties, unknown inputs and the control inputs, the error between the residual and fault is minimized. The problem of robust fault detection is converted into a positive L₁ filtering problem. Subsequently, by constructing an appropriate multiple co-positive type Lyapunov–Krasovskii functional, as well as using the average dwell time approach, sufficient conditions for the solvability of this problem are established in terms of linear matrix inequalities (LMIs). Two illustrative examples are provided to show the effectiveness and applicability of the proposed results.     

APPROACH TO FAULT ON-LINE DETECTION AND DIAGNOSIS BASED ON NEURAL NETWORKS FOR ROBOT IN FMS

ＡＰＰＲＯＡＣＨＴＯＦＡＵＬＴＯＮＬＩＮＥＤＥＴＥＣＴＩＯＮＡＮＤＤＩＡＧＮＯＳＩＳＢＡＳＥＤＯＮＮＥＵＲＡＬＮＥＴＷＯＲＫＳＦＯＲＲＯＢＯＴＩＮＦＭＳＳｈｉＴｉａｎｙｕｎＺｈａｎｇＺｈｉｊｉｎｇＷａｎｇＸｉｎｙｉＺｈｕＸｉａｏｙａｎＳｃｈｏｏｌｏ...     

基于奇异值分解的故障检测技术及其应用

针对基于未知输入观测器的诊断方法在诊断多故障时具有局限性，提出了一种基于奇异值分解的诊断方法，这种方法通过奇异值分解将不同故障对系统残差的影响进行分离，给出了多故障检测与分离的方法，仿真证明这种方法对于多故障诊断有很好的效果。     

Sensor fault detection and isolation via high-gain observers: application to a double-pipe heat exchanger

This paper deals with fault detection and isolation (FDI) in sensors applied to a concentric-pipe counter-flow heat exchanger. The proposed FDI is based on the analytical redundancy implementing nonlinear high-gain observers which are used to generate residuals when a sensor fault is presented (as software sensors). By evaluating the generated residual, it is possible to switch between the sensor and the observer when a failure is detected. Experiments in a heat exchanger pilot validate the effectiveness of the approach. The FDI technique is easy to implement allowing the industries to have an excellent alternative tool to keep their heat transfer process under supervision. The main contribution of this work is based on a dynamic model with heat transfer coefficients which depend on temperature and flow used to estimate the output temperatures of a heat exchanger. This model provides a satisfactory approximation of the states of the heat exchanger in order to allow its implementation in a FDI system used to perform supervision tasks.     

Fault diagnosis and accommodation of a three-tank system based on analytical redundancy

This paper investigates the application of a fault diagnosis and accommodation method to a real system composed of three tanks. The performance of a closed-loop system can be altered by the occurrence of faults which can, in some circumstances, cause serious damage on the system. The research goal is to prevent the system deterioration by developing a controller that has some capabilities to compensate for faults, that is, the fault accommodation or fault-tolerant control. In this paper, a two-step scheme composed of a fault detection, isolation and estimation module, and a control compensation module is presented. The main contribution is to develop a unique structured residual generator able to isolate and estimate both sensor and actuator faults. This estimation is of paramount importance to compensate for these faults and to preserve the system performances. The application of this method to the three-tank system gives encouraging results which are presented and commented on in case of various kinds of faults.     

A novel scheme for current sensor faults diagnosis in the stator of a DFIG described by a T-S fuzzy model

Diagnosis of current sensor faults (CSF) for doubly fed induction generators (DFIGs) is of paramount importance for the reliable power generation of DFIG-based wind turbines (WT). In this paper, a new scheme is developed for current sensors faults diagnosis in the stator of a DFIG-based WT. The nonlinear model of the DFIG is first transformed into an equivalent Takagi-Sugeno (T-S) fuzzy model. Secondly, using this model, a novel fault detection and isolation (FDI) algorithm is proposed. This algorithm is based on a bank of Luenberger observers for residuals generation combined with a new proposed residual vector. Furthermore, a new binary decision logic is used for CSF isolation. Stability analysis of the observer bank is analyzed using a Lyapunov theorem, which allows deriving sufficient stability conditions by solving a system of Linear Matrix Inequalities (LMIs). A simulation study is carried out to assess the performance and the effectiveness of the new FDI scheme.     

Multiple model-based detection and estimation scheme for gas turbine sensor and gas path fault simultaneous diagnosis

In this paper, a multiple model (MM)-based detection and estimation scheme for gas turbine sensor and gas path fault diagnosis is proposed, which overcomes the coupling effects between sensor faults and gas path faults, and simultaneously realizes an accurate diagnosis of sensor and gas path faults. First, an adaptive fault detection and isolation (FDI) framework based on the MM method was established to detect and isolate sensor faults and gas path faults. Then, a fault amplitude estimation method was proposed according to the FDI results, and a fault validation method based on the Chi-square test was proposed to confirm the actual fault. Finally, hardware in the loop (HIL) simulation platform was established to validate the effectiveness of the proposed method. Several simulation case studies were conducted based on a two-shaft marine gas turbine with common gas path faults and sensor faults. The simulation results show that the proposed method can accurately diagnose the fault and estimate the corresponding fault amplitude when both the sensor fault and the gas path fault coincide.

     

单输出自适应模糊逻辑系统在传感器故障分类与诊断中的应用

本文提出了一种基于泛函构造思想及其辨识的故障分类诊断新方法。为了实现传感器故障分类,首先在系统各模型（包括故障与正常模型）与设定分类标志之间进行泛函构造,然后利用自适应模糊逻辑系统的辨识功能,对构造好的泛函进行辨识,最终利用单输出实现分类功能。仿真实验证实了本文方法的正确性和有效性,同时还表明该方法对噪声有很强的抑制能力。     

An improved PCA method with application to boiler leak detection

Principal component analysis (PCA) is a popular fault detection technique. It has been widely used in process industries, especially in the chemical industry. In industrial applications, achieving a sensitive system capable of detecting incipient faults, which maintains the false alarm rate to a minimum, is a crucial issue. Although a lot of research has been focused on these issues for PCA-based fault detection and diagnosis methods, sensitivity of the fault detection scheme versus false alarm rate continues to be an important issue. In this paper, an improved PCA method is proposed to address this problem. In this method, a new data preprocessing scheme and a new fault detection scheme designed for Hotelling's T2 as well as the squared prediction error are developed. A dynamic PCA model is also developed for boiler leak detection. This new method is applied to boiler water/steam leak detection with real data from Syncrude Canada's utility plant in Fort McMurray, Canada. Our results demonstrate that the proposed method can effectively reduce false alarm rate, provide effective and correct leak alarms, and give early warning to operators.     

Sensor fault detection and isolation system for a condensation process

This article presents the design of a sensor Fault Detection and Isolation (FDI) system for a condensation process based on a nonlinear model. The condenser is modeled by dynamic and thermodynamic equations. For this work, the dynamic equations are described by three pairs of differential equations which represent the energy balance between the fluids. The thermodynamic equations consist in algebraic heat transfer equations and empirical equations, that allow for the estimation of heat transfer coefficients. The FDI system consists of a bank of two nonlinear high-gain observers, in order to detect, estimate and to isolate the fault in any of both outlet temperature sensors. The main contributions of this work were the experimental validation of the condenser nonlinear model and the FDI system.     

A combined mono- and multi-turbine approach for fault indicator synthesis and wind turbine monitoring using SCADA data

The monitoring of wind turbines using SCADA data has received lately a growing interest from the fault diagnosis community because of the very low cost of these data, which are available in number without the need for any additional sensor. Yet, these data are highly variable due to the turbine constantly changing its operating conditions and to the rapid fluctuations of the environmental conditions (wind speed and direction, air density, turbulence, …). This makes the occurrence of a fault difficult to detect. To address this problem, we propose a multi-level (turbine and farm level) strategy combining a mono- and a multi-turbine approach to create fault indicators insensitive to both operating and environmental conditions. At the turbine level, mono-turbine residuals (i.e. a difference between an actual monitored value and the predicted one) obtained with a normal behavior model expressing the causal relations between variables from the same single turbine and learnt during a normal condition period are calculated for each turbine, so as to get rid of the influence of the operating conditions. At the farm level, the residuals are then compared to a wind farm reference in a multi-turbine approach to obtain fault indicators insensitive to environmental conditions. Indicators for the objective performance evaluation are also proposed to compare wind turbine fault detection methods, which aim at evaluating the cost/benefit of the methods from a production manager’s point of view. The performance of the proposed combined mono- and multi-turbine method is evaluated and compared to more classical methods proposed in the literature on a large real data set made of SCADA data recorded on a French wind farm during four years : it is shown than it can improve the fault detection performance when compared to a residual analysis limited at the turbine level only.     

Model-based fault diagnosis in continuous dynamic systems

Traditional fault detection and isolation methods are based on quantitative models which are sometimes difficult and costly to obtain. In this paper, qualitative bond graph (QBG) reasoning is adopted as the modeling scheme to generate a set of qualitative equations. The QBG method provides a unified approach for modeling engineering systems, in particular, mechatronic systems. An input-output qualitative equation derived from QBG formalism performs continuous system monitoring. Fault diagnosis is activated when a discrepancy is observed between measured abnormal behavior and predicted system behavior. Genetic algorithms (GA's) are then used to search for possible faulty components among a system of qualitative equations. In order to demonstrate the performance of the proposed algorithm, we have tested it on a laboratory scale servo-tank liquid process rig. Results of the proposed model-based fault detection and diagnosis algorithm for the process rig are presented and discussed.     

Design of H(infinity) robust fault detection filter for linear uncertain time-delay systems

In this paper, the robust fault detection filter design problem for linear time-delay systems with both unknown inputs and parameter uncertainties is studied. Using a multiobjective optimization technique, a new performance index is introduced, which takes into account the robustness of the fault detection filter against disturbances and sensitivity to faults simultaneously. The reference residual model is then designed based on this performance index to formulate the robust fault detection filter design problem as an H(infinity) model-matching problem. By applying robust H(infinity) optimization control technique, the existence condition of the robust fault detection filter for linear time-delay systems with both unknown inputs and parameter uncertainties is presented in terms of linear matrix inequality formulation, independently of time delay. In order to detect the fault, an adaptive threshold which depends on the inputs is finally determined. An illustrative design example is used to demonstrate the validity of the proposed approach.     

Model-based fault detection and isolation in steer-by-wire vehicle using sliding mode observer

Steer-by-Wire system (SbW), in which the conventional mechanical linkages between the steering wheel and the front wheel are removed, is suited to active steering control, improving vehicle stability, dynamics and maneuverability. And SbW is implemented to autonomous steering control to assist the driver. However, the SbW vehicle contains unsolved important problems about fault tolerant function. For example, it is the detection of sensor fault and multiplicative fault simultaneously. Fault detection and isolation (FDI) is essential in fault-tolerant problems, and conventional FDI for SbW was based on Kalman filter. But this method has weak robustness and cannot detect sensor fault and multiplicative fault simultaneously. We propose a novel model-based fault detection and isolation method using sliding mode observer in the SbW vehicle, which contains measurement of sensor fault and multiplicative fault. The effectiveness of the proposed method is verified by simulations. This paper was recommended for publication in revised form by Associate Editor Kyoungsu Yi Jae-Sung Im was born in Busan, Korea in 1978. He received his B.S. and M.S. degrees in Mechanical Engineering from Pukyong National University, Korea, in 2003 and 2005, respectively. He then received his Ph.D. degree from Kumamoto University, Japan, in 2009. His interests are in vehicle dynamics, robust control, fault detection and isolation, and man-machine interface. Fuminori Ozaki received the B.S. and M.S. degrees from the Department of Computer Science, Kumamoto University, Japan, in 1998 and 2000. In 2000, he joined OMRON Corporation, Kyoto, Japan, where he developed semiconductor manufacturing equipment. His current interests include EPS control and KANSEI engineering. Tae-Kyeong Yue received the B.S. and M.S. degrees from Pukyong National University, Korea, in 1998 and 2000, respectively. He received the Ph.D. degree from Kumamoto University, Kumamoto, Japan in 2003. He is working in the Korea Ocean Research and Development Institute (KORDI), Korea. His interests are fault detection and isolation, decentralized control and control of deep-sea mining system. Shigeyasu Kawaji received his Master of Engineering in Electrical Engineering and Doctor of Engineering in Control Engineering from Kumamoto University and Tokyo Institute of Technology, Japan, in 1969 and 1980, respectively. He joined the Department of Electronic Engineering of Kumamoto University, Japan, where he is presently as a full professor. He is the Director of System Integration Laboratory. He is presently the President of Advanced Health Laboratory Ltd. His current research interest includes robust control, intelligent control mechatronics and robotics, fusion of medicine and engineering, and automotive mechatronic systems.