On Event-Triggered Control For Dynamic Systems Subject To Actuator Saturation

In practical systems,the plant is directly controlled by actuators.Therefore,actu-ator is an indispensable part of the whole system.However,the output of the actuator is often bounded due to physical limitation.This phenomenon can be described by saturation nonlinearity.Usually,saturation would cause degradation of the required performance,or even make the closed-loop system unstable.At present,one of the most important challenge for network control systems is resources saving.Event-triggered scheme is a functional method,which can reduce load of the network and save communication resources.Its basic idea is that signal of the system is not transmitted periodically,it transmits until the variable of the system reaches a prescribed condition.In this way,occupation of the communication channel would reduce.In this thesis,we take saturation and resources saving into consideration simulta-neously.For general linear systems and Markov jump systems,different methods are utilized for event-triggered controller design.First,we propose the state feedback control law to deal with saturation under event-triggered scheme for systems where the states can be fully accessed.Compared with the traditional time-triggered scheme,the number of data transmission is greatly decreased.Moreover,the optimization algorithm is presented for estimating the domain of attraction.In addition,the event condition involves information whether saturation occurs or not.By doing so,the number of event is further reduced.Second,for systems whose states cannot be measured fully,state feedback control law is not expected.In this situation,dynamic output feedback control with event-triggered scheme is applied to stabilize systems subject to actuator saturation.An ex-ponential term is introduced in the given event condition so that Zeno behavior can be excluded.Meanwhile,pole placement strategy is adopted to deal with high gain of the controller.Third,the static anti-windup compensation is investigated for event-triggered scheme with saturation nonlinearity.The obtained results are compared with none anti-windup compensation.To find a larger domain of attraction,the static anti-windup compensator gain,which can be seen as a free variable,is designed using an iterative algorithm.In the sequel,dynamic anti-windup compensation is utilized for guaranteeing sta-bility of the closed-loop system under event-triggered scheme.Due to coupling of vari-ables,projection theorem is used to derive the equivalent condition by transformation.Even though the analysis procedure is somewhat complex,dynamic anti-windup com-pensation has extra degrees of freedom for obtaining a larger domain of attraction.Finally,many practical systems can be modeled as Markov jump system.The most important issue of these system is transition probability.In many cases,all the element in transition probability matrix are assumed to be known.In practice,however,it is difficult or even impossible to obtain all the exact values of transition probabilities,i.e.,transition probability matrix contains unknown elements.With event-triggered scheme,sufficient conditions are proposed to stabilize these systems subject to actuator satura-tion for these two cases.