Research On Data-Driven Control Method Of Nonlinear Systems Based On Dynamic Linearization Technique

With the rapid development of modern science and technology,the scale and com-plexity of industrial control systems are also increasing.It is becoming more and more difficult to study the control of production equipment by relying on the traditional mecha-nism model or the identification model.It is also difficult to get the accurate mathematical model of the system through the traditional mechanism modeling method,and it usual-ly takes considerable time and energy.However,the actual system produces and stores a large number of on-line and off-line input and output(I/O)data,which allows peo-ple to think about how to use these data to replace the mechanism model and design the corresponding data driven control methods.After more than twenty years of continuous development,the model free adaptive control methods have been successfully applied to industrial process control,information physical system,motor control,transportation,e-conomy and military and so on,and also achieved good results.In recent years,with the development of network communication technology and data transmission capability,the traditional time sampling control method due to the periodic transmission and updat-ing of control signals at every moment makes a large number of unnecessary redundant information transmitted,which will lead to the wave of the limited network resources.Therefore,it is very useful to study the event triggered control problem of nonlinear sys-tems by using online input and output data,when the system models are unknown.On the other hand,practical engineering systems are often limited in some applications,such as steady-state error,overshoot and convergence rate speed.How to effectively solve the prescribed performance control of data driven nonlinear systems has become an important research topic.On the basis of summarizing the previous results,we use neural network,even-t triggered technology,adaptive method,discrete sliding mode technology and dynamic linearization technique as the basic tools.In this dissertation,we consider the nonlinear discrete-time systems.The model free adaptive control methods are combined with adap-tive event triggered controller,discrete adaptive sliding mode strategy respectively.The main results of this paper are proved theoretically,and the simulation experiments are car-ried out on the actual model of the continuous stirred reactor system and the steam water heat exchanger system.The results show the effectiveness of the proposed methods.The full text of this dissertation is divided into eight chapters,the main contents of each chapter are given as follows:Chapters 1-2 systematically introduce and analyze the background and development of the model free adaptive technique and its related control methods.The preliminary knowledge and research methods related to this paper are also provided.In Chapter 3,a data driven pseudo gradient parameter estimation method based on the observer and adaptive event triggered method is designed for the unknown nonlin-ear discrete time systems.The proposed adaptive event triggered control strategy can effectively solve the known restrictions of the existing methods on the controlled object model.Compared with the traditional time sampling method,the transmission of input and output data and the weight updating rule of neural network are only working at the event triggered time.Therefore,the design method can improve the utilization efficiency of communication resources and solve the problem of excessive computation load and data transmission load.Finally,a simulation example and a comparison with the existing results are given to illustrate the effectiveness and advantages of the proposed method.In Chapter 4,on the basis of the third chapter,we discuss the adaptive event triggered control of nonlinear systems.First,the adaptive event triggered condition and neural net-work adaptive controller collaborative design method are proposed by using on-line input and output data.Moreover,the proposed method guarantees the stability and conver-gence of the closed-loop system.Compared with the existing time triggered method,this method can effectively reduce the number of transmission times of input and output data and guarantee the tracking performance of the system.Finally,an example of continu-ous stirred tank reactor system is used to illustrate the effectiveness and superiority of the proposed method.Chapter 5 considers the discrete time nonlinear system.An cooperative design method based on the transformation error strategy and the discrete adaptive sliding mod-e controller is proposed in the case that the system model is completely unknown.In order to solve the prescribed performance control problem,the transformation error strat-egy is used to convert the tracking error problem with the constraint into an equivalent unconstrained problem.The proposed method can ensure the steady-state error,the su-per harmonic convergence rate and meet the given performance requirements at the same time.Finally,a simulation example is given to illustrate the effectiveness of the design method.Chapter 6 on the basis of the fifth chapter,based on asymmetric prescribed perfor-mance function,a data driven adaptive sliding mode control method is presented.First,for the unknown nonlinear discrete-time system,in order to deal with the tracking error constraint problem,the constrained original error is converted to an equivalent uncon-strained condition.Then,a discrete adaptive sliding mode controller is designed by using the input and output data of the system,to ensure that the tracking error converges to a predefined region.Finally,a simulation example is given to illustrate the effectiveness of the proposed algorithm.Chapter 7 considers the same nonlinear system,a model free adaptive control strat-egy based on reinforcement learning is designed by using the execution evaluation neural network control structure when the system model is completely unknown.Compared with the existing methods,the obtained design conditions have better tracking performance.In addition,the Lyapunov function is used to prove that the closed-loop system is uniformly ultimately bounded.Finally,a simulation example is given to illustrate the effectiveness of the conclusions obtained.Chapter 8 summarizes the results of the dissertation and point out the future research topics in relevance.