Sampled-data Model Free Adaptive Control For Nonlinear Continuous-time System

Conventional model free adaptive control methods are based on discrete-time systems,while most of the actual industrial systems are continuous-time systems.Based on this problem,this dissertation proposes a series of sampled-data model free adaptive control methods,which extends the model free adaptive control from discrete-time systems to continuous-time systems.The main contents are as follows:For a class of nonlinear nonaffine continuous-time systems,this dissertation proposes a compact form sampled-data model free adaptive control method.The core idea of this method is to transform the original system into a sampled-data-based dynamical linearization data model by using an improved sampled-data dynamic linearization method,and the unknown nonlinearities and nonaffine structure of the system are compressed into a single time-varying parameter.Based on the proposed dynamic linearized data model,the sampled-data control law and the sampled-data parameter adaptive update law are given in this dissertation by solving two criterion functions.The proposed method can not only use the sampling period to improve the control performance,but also suppress the uncertainty through the sampled-data parameter adaptive update law.Furthermore,the proposed control method is data-driven and overcomes the problem of model dependence in traditional control methods.In addition,this dissertation also proposes partial form sampled-data model free adaptive control method and full form sampled-data model free adaptive control method by introducing the concept of pseudo-gradient vectors and more input and output information of past moments.For a class of nonlinear nonaffine continuous-time systems,this dissertation proposes a sampled-data local dynamic linearization method to transform the original system into a sampled-data-based local dynamical linearization data model.The uncertainty and unknown nonlinearity of the system are transformed into unknown time-varying parameter vectors and nonlinear residual terms.In this dissertation,the two unknown terms are estimated by the designed sampled-data parameter adaptive update law and sampled-data extended state observer,respectively.Based on the proposed local dynamic linearized data model,this dissertation proposes an observer based partial form sampled-data model free adaptive control method,which includes the sampling period and more input information of past moments to enhance the control effect of the system.In addition,this dissertation also proposes an observer based full form sampled-data model free adaptive control method.Theoretical analysis and simulation confirm the effectiveness of the above methods.