A multiple observers and dynamic weighting ensembles scheme for diagnosing new class faults in wind turbines

This paper presents an incremental way to design the decision module of a diagnostic system by resorting to dynamic weighting ensembles of classifiers. The method is applied for sensor fault detection and isolation in a doubly fed induction generator for wind turbine application. Three sets of observers are combined to generate residuals that are robust to operating point changes. These signals are progressively fed into a dynamic weighting ensembles algorithm, called Learn⁺⁺.NC, for fault classification. The algorithm incrementally learns the residuals–faults relationships and dynamically classifies the faults including multiple new classes. It resorts to a dynamically weighted consult and vote mechanism to combine the outputs of the base-classifiers.     

Robust Isolation Of Sensor Failures

Sensor self‐validity check is a critical step in system control and fault diagnostics. In this paper, a robust approach to isolate sensor failures is proposed. First, a residual model for a given system is built off‐line and directly based on input‐output measurement data. The residual model outputs are called “primary residuals” and are zero when there is no fault. Most conventional approaches to residual model generation are indirect, as they first require the determination of state‐space or other models using standard system identification algorithms. Second, a new max‐min design of structured residuals, which can maximize the sensitivity of structured residuals with respect to sensor failures, is proposed. Based on the structured residuals, one can then isolate the sensor failures. This design can also be done in an off‐line manner. It is an optimization procedure that avoids local optimal solutions. Simulation and experimental results demonstrated the effectiveness of the proposed method.     

Structural analysis based sensors fault detection and isolation of cylindrical lithium-ion batteries in automotive applications

The battery sensors fault diagnosis is of great importance to guarantee the battery performance, safety and life as the operations of battery management system (BMS) mainly depend on the embedded current, voltage and temperature sensor measurements. This paper presents a systematic model-based fault diagnosis scheme to detect and isolate the current, voltage and temperature sensor fault. The proposed scheme relies on the sequential residual generation using structural analysis theory and statistical inference residual evaluation. Structural analysis handles the pre-analysis of sensor fault detectability and isolability possibilities without the accurate knowledge of battery parameters, which is useful in the early design stages of diagnostic system. It also helps to find the analytical redundancy part of the battery model, from which subsets of equations are extracted and selected to construct diagnostic tests. With the help of state observes and other advanced techniques, these tests are ensured to be efficient by taking care of the inaccurate initial State-of-Charge (SoC) and derivation of variables. The residuals generated from diagnostic tests are further evaluated by a statistical inference method to make a reliable diagnostic decision. Finally, the proposed diagnostic scheme is experimentally validated and some experimental results are presented.     

Isolation and handling of sensor faults in nonlinear systems

This work considers the problem of sensor fault isolation and fault-tolerant control for nonlinear systems subject to input constraints. The key idea is to design fault detection residuals and fault isolation logic by exploiting model-based sensor redundancy through a state observer. To this end, a high-gain observer is first presented, for which the convergence property is rigorously established, forming the basis of the residual design. A bank of residuals are then designed using a bank of observers, with each driven by a subset of measured outputs. A fault is isolated by checking which residuals breach their thresholds according to a logic rule. After the fault is isolated, the state estimate generated using measurements from the healthy sensors is used in closed-loop to maintain nominal operation. The implementation of the fault isolation and handling framework subject to uncertainty and measurement noise is illustrated using a chemical reactor example.     

Sensor fault diagnosis in a chemical process via RBF neural networks

Radial basis function (RBF) neural networks are investigated here for process fault diagnosis. The use of the output prediction error, between a neural network model and a non-linear dynamic process, as a residual for diagnosing actuator, component and sensor faults is analysed. It is found that this residual for a dependent neural model is less sensitive to sensor faults than actuator or component faults. This is confirmed in experiments for a real, multivariable chemical reactor. A scheme is then developed utilising a semi-independent neural model to generate enhanced residuals for diagnosing the sensor faults in the reactor. A second neural-network classifier is developed to diagnose the sensor faults from the residuals generated, and results are presented to demonstrate the satisfactory detection and isolation of sensor faults achieved using this approach.     

New nonlinear residual feedback observer for fault diagnosis in nonlinear systems

The increased complexity of plants and the development of sophisticated control systems have necessitated the parallel development of efficient Fault Detection and Isolation (FDI) systems. This paper discusses a model based technique, viz., observers for detecting and isolating parametric and sensor faults. In this paper, a novel diagonal nonlinear residual feedback observer is proposed which is valid for a certain class of nonlinear systems where, subject to other conditions, the state depends nonlinearly on the fault. A number of typical chemical engineering systems can be represented by models of this form. The structure of the observer ensures that the residuals are diagonally affected by the faults. Conditions for exact decoupling of residuals are presented and convergence of the observer in the presence of step faults is proved using Lyapunov like analysis. Multiple observers and a decision logic module are used for FDI when there are un-monitored faults. Results are presented from numerical simulations of an illustrative example and a typical chemical engineering system: a counter-current heat exchanger.     

Data‐based design of robust fault detection and isolation residuals via LASSO optimization and Bayesian filtering

In this paper, a data‐based approach for the design of structured residual subsets for the robust isolation of sensor faults is proposed. Linear regression models are employed to estimate faulty signals and to build a set of primary residuals. L₁‐regularized least squares estimation is used to identify model parameters and to enforce sparsity of the solutions by increasing the regularization weight. In this way, it is possible to generate a set of residuals generators with different fault sensitivity. Then, a residual selection procedure based on fault sensitivity maximization is proposed to extract a minimum size subset of structured residuals that allows for isolation of the faulty sensor. To overcome modelling uncertainty, a robust recursive Bayesian Filter has been employed to process, online, the distance of the residuals from nominal fault directions, providing a fault probability for each sensor. The proposed method has been validated by designing and testing a fault isolation scheme for six aircraft sensors using multi‐flight experimental data of a P92 Tecnam aircraft.     

A new method of sensor fault diagnosis for under-measurement system based on space geometry approach

Sensor fault cannot be converted to system equation under the condition of under- measurement system. Aiming to solve this problem, we present a new method which treats sensor fault as state variable to enforce fault diagnosis. Firstly, the system model of sensor fault is constructed by putting sensor fault into the state equation. Then, the residual generator is designed using the space projection operation to solve the relevant parameter matrices. Since the proposed algorithm satisfies one-to-one correspondence of faults and residuals, it can achieve single and multiple sensors FDI. Simulation results show the effectiveness of the proposed approach.     

基于稀疏分解残差的氢气传感器故障探测与辨识方法

针对传感器故障探测和诊断,提出了一种基于稀疏分解残差的氢气传感器故障探测和辨识方法.基于信号稀疏分解理论,对采集的传感器正常信号数据集,利用K奇异值分解(K-SVD)学习算法得到一超完备字典D;在字典上对非正常(故障)信号进行分解,根据稀疏分解的残差大小和范围完成对传感器故障的探测及辨识.实验结果表明:对氢气传感器的故障探测率和总辨识率分别达到98.75%和97.25%,可以有效地解决氢气传感器的故障探测和辨识.     

Data-driven sensor fault diagnosis systems for linear feedback control loops

This paper develops a sensor fault diagnosis (SFD) scheme for a multi-input and multi-output linear dynamic system under feedback control to identify different types of sensor faults (bias, drift and precision degradation), particularly for the incipient sensor faults. Feedback control, leading to fault propagation and disguised fault rectification, imposes the challenge on the data-driven SFD. With only available output data in closed loop, the proposed scheme comprises two stages of residual generation and residual evaluation. In the residual generation, a data-driven identification of the residual generator for the feedback control system is proposed. One class of parameters in the residual generator are estimated using process delays while another class of parameters describing the output dynamic are derived by the Bayes’ formula. The means and variances control charts of online calculated residuals are made to judge the root cause. Two case studies are performed to illustrate the effectiveness of the proposed method.     

Optimal residual design for fault diagnosis using multi-objective optimization and genetic algorithms

This paper develops a new approach to the design of optimal residuals in order to diagnose incipient faults based on multi-objective optimization and genetic algorithms. In this approach the residual is generated via an observer. To reduce false and missed alarm rates in fault diagnosis, a number of performance indices are introduced into the observer design. Some performance indices are expressed in the frequency domain to take account of the frequency distributions of faults, noise and modelling uncertainties. All objectives then are reformulated into a set of inequality constraints on performance indices. A genetic algorithm is thus used to search for an optimal solution to satisfy these inequality constraints on performance indices. The approach developed is applied to a flight control system example, and simulation results show that incipient sensor faults can be detected reliably in the presence of modelling uncertainty.     

Set-membership parity space hybrid system diagnosis

In this paper, diagnosis for hybrid systems using a parity space approach that considers model uncertainty is proposed. The hybrid diagnoser is composed of modules which carry out the mode recognition and diagnosis tasks interacting each other, since the diagnosis module adapts accordingly to the current hybrid system mode. Moreover, the methodology takes into account the unknown but bounded uncertainty in parameters and additive errors (including noise and discretisation errors) using a passive robust strategy based on the set-membership approach. An adaptive threshold that bounds the effect of model uncertainty in residuals is generated for residual evaluation using zonotopes, and the parity space approach is used to design a set of residuals for each mode. The proposed fault diagnosis approach for hybrid systems is illustrated on a piece of the Barcelona sewer network.     

多故障的奇偶方程-参数估计诊断方法

提出一种将奇偶方程与参数估计相结合的多故障诊断方法。构造了一种新的奇俩方程，其产生的残差仅对一个传感器故障和一个执行器故障敏感。将传感器和执行器故障模型表示成刘度因子和偏差的形式，应用卡尔曼滤没方法对各故障模型参数进行估计。某型号飞机控制系统的仿真结果表明，新方法能对传感器故障和执行器故障同时存在的线性系统进行诊断，有效地估计出各故障的模型参数。     

基于状态观测器的双馈风电机组变流器开关管开路故障检测

针对双馈风电机组变流器可靠性较差,一旦变流器开关管发生故障,整个风电机组可能遭到严重损坏甚至导致停机的问题,提出一种基于状态观测器的双馈风电机组转子侧变流器开关管开路故障检测方法.根据双馈风电机组拓扑结构,结合其感应电机和变流器进行整体建模;构建状态观测器以实现对转子侧电流的在线估计并生成残差;将残差与设定的阈值比较,判断是否有故障发生.仿真结果证明该方法可行且有效.     

FDI based on an adaptive observer for current and speed sensors of PMSM drives

This paper deals with a new method of current and speed sensors faults detection isolation (FDI) and identification for a permanent magnet synchronous motor (PMSM) drives. A new state variable is introduced so that an augmented system can be constructed to treat PMSM sensor faults as actuator faults. This method uses the PMSM model and a bank of adaptive observers to generate residuals. The residuals results are used for sensor fault detection. A logic algorithm is built in such a way to isolate and identify the faulty sensor for a stator phase current fault after detecting the fault occurrence. Simulation results are presented to illustrate the functionality of theoretical developments. Experimental results with 1.1-kW PMSM have validated the effectiveness of the proposed FDI method. The experimental implementation is carried out on powerful dSpace DS1103 controller board based on the DSP TMS320F240.     

基于模型的曲轴位置传感器故障诊断研究

为了对汽车发动机转速传感器作出快速准确的诊断,结合曲轴信号处理原理,在MATLAB/simulink中设计基于卡尔曼滤波器残差检验的OBD诊断模块,实现了曲轴位置传感器无信号、信号错误等故障诊断功能。针对卡尔曼滤波器法无法排除掉非故障性信号输入的缺点,在卡尔曼滤波法基础上加入逻辑诊断,通过条件判断以启动诊断,加入决策逻辑以确定故障状态。验证结果表明,卡尔曼滤波器法能够准确快速识别两种传感器故障,但在非故障性信号时无法准确判断故障。加入诊断逻辑的改进卡尔曼滤波法能够准确排除非故障性输入,保证诊断的准确性。     

Real-Time Model-Based Fault Detection and Isolation for UGVs

The paper presents a model-based sensor fault detection and isolation system applied in real-time to unmanned ground vehicles. Structural analysis is applied on the nonlinear model of the vehicle for building the residual generation module, followed by an ad-hoc residual evaluation module for detecting single and multiple sensor faults. The overall proposed diagnosis scheme has been tested in real-time on a real mobile robot in an outdoors environment and for different tasks. The obtained experimental results are satisfactory in terms of diagnosis performance and real-time implementation.     

一种神经网络预测器在传感器故障诊断中的应用

讨论基于神经网络预测器的传感器故障诊断问题.介绍了传感器故障诊断技术的发展,提出了一种基于神经网络在线学习的传感器故障实时诊断的模型.通过比较三种前馈神经网络的预测残差确定网络类型.介绍了网络的学习规则,说明了在线学习的能力.最后,通过电厂高加热器的几个温度传感器的实际数据为例说明了此方法的实效性.     

Invariant‐set‐based fault diagnosis in Lure systems

In this paper, we present an invariant‐set‐based method for actuator and sensor fault detection and isolation in Lure systems. The Lure plant is controlled by an observer‐based feedback tracking controller, designed for the nominal (fault‐free) system. Suitable residual signals are constructed from measurable system outputs and estimates associated with the nominal observer. Faults are diagnosed by online contrasting the residual signal trajectories against sets of values that the residuals are shown to attain under healthy or faulty operation. These values are obtained via set‐invariance analysis of the system closed‐loop trajectories. Copyright © 2013 John Wiley & Sons, Ltd.     

Sensor fault diagnosis of nonlinear processes based on structured kernel principal component analysis

A new sensor fault diagnosis method based on structured kernel principal component analysis （KPCA） is proposed for nonlinear processes. By performing KPCA on subsets of variables, a set of structured residuals, i.e., scaled powers of KPCA, can be obtained in the same way as partial PCA. The structured residuals are utilized in composing an isolation scheme for sensor fault diagnosis, according to a properly designed incidence matrix. Sensor fault sensitivity and critical sensitivity are defined, based on which an incidence matrix optimization algorithm is proposed to improve the performance of the structured KPCA. The effectiveness of the proposed method is demonstrated on the simulated continuous stirred tank reactor （CSTR） process.