Decentralized Control Of Adaptive Neural Networks For Non-strict Feedback Large Systems With Input Delay And Saturation

With the development of modern industrial intelligence,the nonlinear large-scale systems widely existing in industrial production and social environment have attracted the attention of many scholars.Obviously,compared with the general nonlinear systems,the research on this kind of more complex nonlinear large-scale systems is not only full of challenges,but also has higher application value.In addition,input delay and input saturation are two common phenomena in practical dynamic systems.Due to the existence of the input delay and the saturation,the stability and performance of systems are inevitably affected.On the basis of the existing studies,this thesis researches the state feedback and output feedback decentralized control problems of nonstrict-feedback nonlinear large-scale systems with input delay and saturation by the adaptive neural network method and backstepping technique.The following are the main contents of this thesis:In the first chapter,the research background,significance and status of nonlinear large-scale systems are introduced respectively,and then the organization of the whole thesis is given.In the second chapter,the preparatory knowledge of decentralized control,adaptive control,Radial Basis Function(RBF)neural networks,backstepping technique,Lyapunov stability theorem and the main lemma used in this thesis are introduced respectively.In the third chapter,under the assumption of the system states being measurable,the problem of decentralized control via state feedback of a class of nonstrict-feedback nonlinear large-scale systems with input delay and saturation is studied.In the process of control design,coordinate transformation with an integral term is introduced to deal with the input delay and a smooth nonlinear function is used to approximate the saturation function.In addition,the unknown nonlinear function generated in control design is approximated by RBF neural networks.A decentralized controller is constructed by combining the adaptive neural network control method and backstepping technique.The stability analysis is carried out by using Lyapunov stability theory.The theoretical analysis results show that the designed decentralized controller can ensure that tracking errors converge to a small neighborhood of the origin,and all signals of the closed-loop systems are bounded.Furthermore,the validity of this decentralized control scheme is verified by the results of two simulation examples.In the fourth chapter,under the assumption that the system states are unmeasurable,but the output variable is measurable,an observer-based decentralized output-feedback control scheme is proposed for a class of nonstrict-feedback nonlinear large-scale systems with time-varying input delay,saturation and unknown virtual control gains.In the process of control design,an auxiliary system is introduced to compensate for the effect of time-varying input delay.Furthermore,the convex combination technique is used to overcome the difficulty in control design caused by the unknown virtual control gains,and an effective state observer is constructed to estimate the system states.Then a decentralized output-feedback controller is designed by using the adaptive neural backstepping control approach.By Lyapunov stability theory,it is proved that the designed decentralized output-feedback controller can ensure that tracking errors converge to a small neighborhood of the origin and other closed-loop states are bounded.Subsequently,the numerical simulation is used to further verify the validity of this decentralized output-feedback control scheme.The fifth chapter summarizes the main content of this thesis and discusses the future research direction.Compared with the existing research results,the main contributions of this thesis are as follows:(1)Each subsystem function has the nonstrict-feedback form,and the interconnections are the functions of the whole large-scale systems' state variables.Therefore,the large-scale systems considered in some literature are just the special cases of the systems under our consideration;(2)The coordinate transformation method with integral term and the auxiliary system method are used to solve the input delay problem of nonlinear large-scale systems,respectively;(3)It provides a systematic method to construct state observer for the nonlinear systems with unknown time-varying virtual control gains.