Article

Nonlinear Observer-Based Fault Detection

04/1996;
Source: CiteSeer

ABSTRACT This communication deals with the problem of designing a nonlinear observer in order to achieve fault detection and localization for a wide class of nonlinear systems subjected to bounded nonlinearities. A dedicated nonlinear observer scheme (DNOS) for fault detection and identification in reconstructible systems is proposed. INTRODUCTION State observation of nonlinear dynamical systems is becoming a growing topic of investigation in the specialized literature. The reconstruction of the time behaviour of state variables remains a major problem both in control theory and process diagnosis. Researchers attention is being particularly focused on the design of adaptive observers for on-line process state estimation. There is increasing awareness that to ensure robustness in performance requires simpler and stable adaptive observer schemes. Linear systems have received considerable attention (Luenberger, 1966), (O'Reilly, 1983) leading to several stable adaptive observer systems (Kreisselm...

Download full-text

Full-text

Available from: José Ragot, Jan 17, 2013
0 Followers
 · 
221 Views
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: With the increasing demand for higher performance, safety and reliability of dynamic systems, fault diagnosis has received more and more attention. The observer-based strategy is one of the active research fields, which is widely used to construct model-based fault detection systems for technical processes which can be well modelled as linear time invariant systems. Fault diagnosis for nonlinear system is an active area of research. Observer-based fault detection includes two stages, residual generation and residual evaluation. The residual generation problems and residual evaluation problems for systems with only deterministic disturbances or stochastic disturbances have been widely separately studied. Recently some efforts have been made in the integrated design of fault detection systems for systems with deterministic disturbances and stochastic disturbances. Recently, successful results of applying Takagi-Sugeno (TS) fuzzy model-based technique to solve fault detection and isolation problems met in the nonlinear system have been achieved. With TS model, a nonlinear dynamic system can be linearised around a number of operating points. Each linear model represents the local system behaviour around the operating point. The global system behaviour is described by a fuzzy IF-THEN rules which represent local linear input/output relations of the nonlinear system. Applying the Takagi-Sugeno fuzzy model based technique to solve fault detection and isolation problems in the nonlinear systems is active area of research. The main contribution of this thesis is the design of robust fault detection systems based on Takagi-Sugeno fuzzy filters. There are a number of schemes to achieve robustness problem in fault detection. One of them is to introduce a performance index. It is function of unknown input signal and fault signal. For continuous time system, first, robust fault detection system will be designed for nonlinear system with only deterministic disturbance as unknown inputs. Second, robust fault detection system will be designed for nonlinear system with deterministic disturbance as unknown inputs and parameter uncertainties. Finally, robust fault detection system will be designed for nonlinear system with deterministic disturbance as unknown inputs and stated delay. Sufficient conditions for solving robustness problem are given in terms of Linear Matrix Inequalities (LMIs). For discrete time system, kalman filter design for nonlinear system is diffcult. In this thesis new fault detection approach will be presented for nonlinear system with only stochastic disturbance. Fault Detection (FD) system for each local subsystem is design by solving the corresponding Discrete-time Algebraic Riccati Equation (DARE). Optimisation algorithm based on minimizing the residual covariance matrix is used to obtain a robust FD system optimised for global system behaviour. The optimisation algorithm is established in terms of LMIs. The different robust fault diagnosis system are developed to detect sensor faults of vehicle lateral dynamic control systems.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Dans cette thèse nous proposons une nouvelle méthode d'isolation et d'identification de défaut singulier pour les systèmes dynamiques non linéaires. Cette méthode est basée sur la caractéristique de monotonicité de l'erreur de prédiction de l'observateur en fonction de la différence des paramètres. L'ensemble des valeurs admissibles de chaque paramètre est subdivisé en un certain nombre d'intervalles. On construit un observateur d'isolation pour chaque intervalle, cet observateur est initialisé dans l'intervalle considéré. Après l'occurrence du défaut, la valeur du paramètre défectueux doit être dans un des intervalles du paramètre. L'amplitude du résidu calculé par l'observateur d'isolation correspondant à cet intervalle (celui qui contient la valeur du paramètre défectueux) sera dans le domaine limité par deux seuils dynamiques à tout instant. Par contre, les résidus correspondant aux autres intervalles auront de grandes amplitudes et leurs évolutions ne sont pas limitées par les deux seuils dynamiques correspondants. Par conséquent l'intervalle contenant la valeur du paramètre défectueux peut être déterminé et le défaut est donc isolé et identifié. Différentes versions de cette méthode ont été développées : une première avec des seuils fixes, une seconde avec des seuils adaptatifs et une dernière sans seuils. On peut montrer que cette méthode a des points communs avec celle basée sur les observateurs adaptatifs. Cependant, cette dernière a un inconvénient majeur qui est la lenteur de sa vitesse d'isolation. C'est pour pallier ce problème, que nous proposons cette méthode In this thesis we propose a new method of isolation and identification of singular fault for nonlinear dynamic systems. This method is based on the characteristic of monotonicity of the observer prediction error according to the difference of the parameters. The whole of the admissible values of each parameter is subdivided in a certain number of intervals. One builds an isolation observer for each interval, this observer is initialized in the considered interval. After the fault occurrence, the value of the faulty parameter must be in one of the parameter intervals. The amplitude of the residue calculated by the isolation observer corresponding to this interval (that which contains the faulty parameter value) will be in the field limited by two dynamic thresholds at any time. On the other hand, the residues corresponding to the other intervals will have great amplitudes and their evolutions are not limited by the two corresponding dynamic thresholds. Consequently the interval containing the value of the faulty parameter can be determined and the fault thus is isolated and identified. Various versions of this method were developed: a first with fixed thresholds, one second with adaptive thresholds and a last without thresholds. One can show that this method has common points with that based on the adaptive observers. However, this last has a major disadvantage which is the slowness of its isolation speed. It is to mitigate this problem, that we propose this method
  • Source