Fault Detection And Tolerant Control For Stochastic Distribution Systems Based On Fuzzy Modeling

With the higher requirements for stochastic control and process control in the aspect of theory and practice, the output probability density functions (PDFs) in different actual processes are viewed as a controlled objective and further analysis a variety of modeling and control methods, has become a novel and challenging task and is called stochastic distribution system control problem. On this basis, this dissertation focus on the modeling and control methods of stochastic distribution systems with fault and investigates the fault detection, diagnosis and tolerant control problem under the fuzzy modeling framework by combining with fuzzy identification performance. The main contributions of this dissertation are summarized as follows:(1) This dissertation introduces the background knowledge and the development situation in domestic and foreign of non-Gaussian stochastic distribution systems, fault detection and tolerant control, fuzzy modeling and so on. This dissertation also analyzes some theoretical problems and gives the structural arrangements.(2) This dissertation investigates the fault detection and diagnosis problem for stochastic distribution systems based on fuzzy modeling. By introducing T-S fuzzy model, the nonlinear relationship between the weight dynamics and the control input can be described approximately, and the gray-box modeling problem in stochastic distribution systems can be solved. On this basis, the fault detection filter and adaptive fault detection filter are designed based on the output PDFs, and the criterion of fault detection is given by solving linear matrix inequalities (LMIs) and a combination of Lyapunov function analysis. Moreover, the method can determine whether the system exists fault by comparing the residual with the value of threshold, meanwhile can achieve fault tracing and measure the size of a fault. Simulation results show the effectiveness of the proposed control method.(3) This dissertation studies the fault diagnosis and tolerant control problem for stochastic distribution systems based on fuzzy modeling. By combining T-S fuzzy weight dynamic models with projection algorithm, the adaptive fault detection filter can be designed such that the bounded restrictions of fault observation value can be solved and can achieve fault tracing and measure the size of a fault. Furthermore, the stability of the closed-loop system with fault can be ensured by designing the feedback control input with fault tolerance and combining with linear matrix inequalities (LMI) optimization method. Simulation results show the effectiveness of the proposed control method.(4) This dissertation studies the fault detection problem for discrete-time stochastic distribution systems based on the two-step fuzzy modeling. By combining with the fuzzy logic systems and T-S fuzzy models, the nonlinear identification and approximation problem in stochastic distribution systems can be solved. Furthermore, the criterion of fault detection is given based on the discrete-time T-S fuzzy dynamic model and the corresponding threshold can also be calculated. By comparing the residual signal with the value of threshold, we can determine whether the system has fault.(5) The general summary is given and the further prospect is also proposed in the dissertation. In this dissertation, the fuzzy modeling problems as well as the fault detection, diagnosis and tolerant control for non-Gaussian nonlinear stochastic distribution systems can be solved.