Research On Model Reference Adaptive Control Of Switched Systems Based On Neural Network

As a special case of hybrid systems,switched systems have a deep background in engineering applications.Especially with the scientific and technological progress in recent years,the control idea of "switching" has been widely application in the field of automatic control,and thus the switched systems have been gained higher attention.In recent years,scholars have continuously reported the research results of switched systems,however,most of them are about deterministic systems.While,in the actual engineering application of the control system,there are often uncertainties,mainly manifested in the dynamic performance of the system changes,the mathematical model established for the system with unknown or random factors.The interaction between the uncertainties and the continuous and discrete dynamic of switched systems makes the dynamic performance of switched systems more complex,which makes the existing analysis and design methods for switched systems cannot be applied directly.Therefore,the study of switched systems with uncertainties has both theoretical and practical significance.Based on neural networks,this thesis studies the model reference adaptive control problem of uncertain switched systems.Uncertainties are parameterized using neural network.By the design of the switching controller and switching adaptive law,the model reference adaptive tracking control problem of switched systems is studied under the average dwell time method,and the state constraints problem of the model reference adaptive switched systems is investigated under the common Lyapunov function method.The main work is outlined below:(1)Under the slow switching,the model reference adaptive tracking control problem of the switched systems based on neural network is studied.Firstly,the switched reference model is established and the stability is analyzed.Secondly,the appropriate switching controller and the switching adaptive law are designed.Then the average dwell time(ADT)method is used to prove the stability of the switched systems according to the multi-Lyapunov stability theory of the switched systems.Finally,the effectiveness of proposed method is illustrated by an aircraft example.(2)Under the arbitrary switching signal,the model reference adaptive state constraints problem of the switched systems based on the neural network is studied.Firstly,the switched reference model is established and the stability is analyzed.Secondly,for the state constraints,a switching controller with switching adaptive law is designed,and the proof that the state tracking error can satisfy the pre-set constraint condition is given by using the common Lyapunov method.Finally,the effectiveness of this method is verified by an example of aircraft.