Research On Fault Diagnosis And Fault-Tolerant Control For T-S Fuzzy Systems

Ensuring the stable operation of control system is a vital prerequisite for its work,but in reality,a variety of faults brought enormous challenges to the smooth running of the system.Therefore,the research of fault diagnosis and fault-tolerant control for control system is of great importance in practice.In the actual situation,most of control systems contain nonlinear dynamic characteristics,T-S fuzzy method provides an important way for the research of nonlinear system due to the better approximation of nonlinear dynamics.Fault diagnosis and fault-tolerant control based on T-S fuzzy model has important practical significance.First,when faults occur in the system,detection of faults as early as possible will avoid a lot of unnecessary losses.In the model-based approach,it is desired that the fault detection observer is as sensitive to fault and robust against disturbance as possible.To this end,sufficient conditions for stable T-S fuzzy model-based observer design with H_/H? performance are derived in terms of linear matrix inequalities in continue and discrete cases respectively.Then,by constructing a set-theoretic description of fuzzy systems,the equivalence class theory in set theory is applied for designing observer and controller for less conservative results.With the information the designed observer get of actuator faults in the system,the fault-tolerant controller is designed to compensate the effect of faults and guarantee stability of the closed-loop dynamic system.Finally,the problem of fault-tolerant control for discrete T-S fuzzy system based on reduced order fault estimation observer is solved.Discrete T-S fuzzy system is reduced in order by the state transformation,then,the reduced-order fault estimation observer is proposed to get the information of the actuator faults,we use this information to design the dynamic output feedback fault-tolerant controller to compensate for the effect of faults and guarantee stability of the system.Simulation results are presented to illustrate the effectiveness of the proposed method.