Research On Additive Fault Observation And Fault Tolerant Control For T-S Fuzzy System

Control system is often affected by various fault signals during the process of operation, and research on the additive fault observation plays an important role in the practice due to the fact that it’s a typical representative of fault signals. On one hand, the observer design is aim to know the work condition of equipment; on another hand, it is important to design a fault tolerant controller based on the estimate information in order to make fault system meet desired performance. Most of actual systems have nonlinear characteristics, and T-S fuzzy method provides a comprehensive way to analysis the nonlinear system.When actuator or sensor failures occur in the controlled plant, fault signals and system states influence each other. If we still adopt the traditional state observer, not only the fault signal can’t be observed, but also the observed effect on system states will be greatly reduced. Due to the existing facts, this paper has studied actuator and sensor fault observers in detail.H∞ optimal performance with LMI catering to disturbance rejection characteristic of the fast adaptive observer for actuator fault has been studied based on the existing results.For sensor failure, two kinds of modified PD observer are considered, and an improved research catering to the first one is placed, which aims to make the error system satisfy both H∞ and regional pole assignment performances. Moreover, the second one for unstable systems is proposed which can observe fault signal and system state at the same time.State feedback fault tolerant controller based on observer and model reference fault tolerant controller are studied, both of them utilize the estimate signals, and two types of fault-tolerant controller design conditions are given. The fact should be pointed out is that all the research works use the T-S fuzzy model to approximate the primary nonlinear systems. Simulation experiments have illustrated the effectiveness of these novel methods. The conclusion is made about this thesis and the prospect is proposed in the end.