Research On Hybrid Fault-tolerant Control For Markovian Jump System

In this thesis,a series of hybrid fault-tolerant controller design methods for Markovian jump systems are given,which make the systems can obtain the suitable control strategy under the influence of difference failures through multi-threshold alarm system.The proposed methods mainly involve robust control theory,fault-tolerant control theory and adaptive control theory.According to different types of failures which the systems are suffered,the corresponding control methods are divided as follows:(1)When invisible fault or incipient fault appears in the system,which is denoted as?(t)?Rq,and such fault has little impact on the system.Thus,the mixed passive/H? hybrid controller is invoked to inhibit the influence of this type of failure.(2)When the actuator partial failure occurs,by using free-weighting matrix approach,Schur complement lemma and S-procedure lemma to obtain the passive fault-tolerant controller against the failure through LMI technology,moreover,this controller will stabilize the system.(3)When the actuator bias failure occurs,augmented matrix approach is adopted to separate the fault information from the system state,then,a new type of state variable with the system state is established.Furthermore,a fault observer is designed to estimate the infomiation of the actuator bias failure,moreover,compensating the influence of the failure through invoking the estimate value into the controller.Since the estimation error is inevitable,it can be regarded as a new type of disturbance that can be inhibited by invoking the H controller.(4)When the system suffers the sudden change of model structure,we will adopt adaptive control method.Combined with the above research results,a multi-threshold alarm system is designed,moreover,a new type of hybrid fault-tolerant control is given based on this alarm system.The advantage of the multi-threshold alarm system is that it can avoid the false switching caused by the false alarm signal,thereby ensuring the control performance of the system.