Fault Analysis And Simulation Research For Stochastic Distribution Systems Based On Two-step Fuzzy Modeling

The complex stochastic processes with random noises, exogenous disturbances and system faults exist in many practical applications, such as industrial control systems, aeronautics and astronautics, etc. Among these stochastic processes, non-Gaussian stochastic distribution system has become a hot point in the field of automatic control because of more complex external environment and lacking of necessary model information. Especially, when the system is affected by unknown fault and external disturbance, how to ensure the stability of the system is particularly important. This thesis studies two-step fuzzy modeling method and further discusses fault diagnosis, fault-tolerant control and simulation realization problem of designed algorithm. The main research work of this thesis is listed as follows:The two-step fuzzy modeling framework is proposed for non-Gaussian stochastic distribution processes based on the approximation capability of fuzzy logic systems and T-S fuzzy models. Under this framework, the fault diagnosis problem is discussed. By combining the designed fuzzy filter with adaptive projection algorithm, the unknown fault can be well estimated and the bounded of fault estimation can also be achieved. Based on convex optimization theory and Lyapunov analysis process, the filter gain is computed and the stability of error systems can be guaranteed when the controlled system exist unknown fault. Simulation examples compare the dynamical approximation process between B-spline function with fuzzy logic system. Simulation results show the satisfactory performance of fault diagnosis.The fault-tolerant control problem of stochastic distribution systems is discussed under the framework of two-step fuzzy modeling. The time-varying fault is considered in complex systems, the corresponding fuzzy filter and the adaptive regulation are designed based on T-S fuzzy weight dynamical models. The filter gain and the controller gain are computed by using convex optimization theory. Furthermore, the fuzzy controlled input with fault-tolerant performance is constructed such that the error systems and closed-loop T-S weight dynamics still stable under the influence of time-varying fault based on Lyapunov analysis process. By using the Simulink simulation platform, the corresponding control flow chart is designed. Simulation results can embody the satisfactory performance of fault diagnosis and fault-tolerant control.The fault diagnosis problem for stochastic distribution systems is discussed under the framework of two-step fuzzy modeling when system faults and exogenous disturbances exist simultaneously. On one hand, by constructing an expansion observer, the exogenous disturbances can be estimated. On the other hand, by designing an adaptive filter with projection algorithm, the unknown fault can be estimated and the bounded of fault estimation can also be achieved. Furthermore, by using convex optimization theory, the observer gain and the filter gain can be computed such that the controlled systems can be ensured to stable and favorable dynamical performance under the dual influence of disturbances and faults. When considering constant faults and harmonic disturbances, the corresponding simulation results are given to show the efficiency of the proposed approach.