Fault Diagnosis And Fault Tolerant Control For Uncertain Linear Systems

With the increasing demand of reliability and safety for complex dynamic systems,fault diagnosis and fault tolerant control have been widely applied in many fields.This thesis studies the fault diagnosis and fault tolerant control for uncertain linear systems.The main contributions of the thesis are summarized as follows:A fault detection(FD)filter design is developed for linear uncertain polytopic systems in finite frequency domains.The H?/H_FD scheme is applied to minimize the effect of disturbances and maximize the sensitivity of faults.By decomposing unavailable state variables,a new adaptive mechanism is developed to design the FD filter with time-varying gains.Moreover,a tangent hyperplane technique is introduced to solve the non-convex problem involved in H_performance,and the FD filter design conditions with some parameter fixed are then given in terms of linear matrix inequalities.The effectiveness and advantages of the proposed method are illustrated through some simulation examples.A fault tolerant consensus controller design is presented for linear multi-agent systems(MAS)with completely unknown system model.Based on an important factorization of Laplacian matrix,a novel system named as weighted edge dynamics is introduced,and it is then proved that the stabilizing controller design for the transformed weighted edge dynamics is equivalent to the consensus controller design for the initial MAS.Within this framework,an efficient data-based algorithm is given to derive the fault tolerant stabilizing controller,and the consensus of the initial MAS can be further ensured under the designed control protocol.Some simulation examples on LC oscillator network are given to illustrate the effectiveness of theoretical results.