Font Size: a A A

Research On Multiple Model Adaptive Control Strategy For Nonlinear Discrete-time Systems

Posted on:2016-03-27Degree:DoctorType:Dissertation
Country:ChinaCandidate:M HuangFull Text:PDF
GTID:1228330461961346Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
With the wide application and intensive study of the computer control theory, most of these complex system control problems can be converted into the control problems of the nonlinear discrete-time systems. Since most of the actually industrial plants suffer from the properties of nonlinear, multiple operating points, fast time-varying and so on, traditional control methods are difficult to obtain satisfactory control effect. Neural network based nonlinear adaptive control methods can identify and control nonlinear systems successfully. However, due to the complexity of the structure and the nonlinear maps of a multilayer neural network, the stability analysis of this method has always been very difficult. The multiple model adaptive control method can guarantee the stability of the systems and improve the control performance simultaneously. However, this method assumes that the higher order nonlinear term of the nonlinear system is globally bounded and the zero dynamic system of the nonlinear system is globally uniformly asymptotically stable, which limits the application of this method. To apply the multiple model adaptive control method to a wider range, modified multiple model adaptive control methods are proposed to relax the assumptions of the nonlinear systems and expand the application scope of this method gradually. The main contributions of this thesis are listed as follows:(1) A novel multiple model adaptive control method is proposed for a class of nonlinear discrete-time systems described by a nonlinear autoregressive moving average model. This controller is composed of a nonlinear robust adaptive controller, a neural network based nonlinear adaptive controller and a switching mechanism. This method relaxes the assumption that the nonlinear term of the system is globally bounded to zero-order proximity bounded and the application scope of this control system is expanded. This method is also applied to the nonlinear multi-variable systems through adding reasonable assumptions of the system interaction matrix and the transfer function. In this paper, the bounded-input-bounded-output stability of the control system and the convergence of the tracking errors are proved and the simulation verification is given.(2)To further expand the application range of the multiple model adaptive control method, an artificial neural network based multiple model adaptive control method is proposed for a class of linear bounded nonlinear discrete-time systems. In this method, a linear model and an artificial neural network based nonlinear model are used to identify the system simultaneously. At every system instant, the controller with good performance is chosen to control the system by the switching mechanism. A robust adaptive law with the normalization is introduced to this method to update the parameters of the linear model and the boundedness assumption of the system nonlinear term is relax to linear boundedness. The assumption that the system has a globally uniformly asymptotically stable zero dynamic system is also removed. The boundedness of all the signals and the convergence of the tracking errors in the closed-loop system are proved and the simulation verification is given.(3)A multiple model direct adaptive control method is proposed for a class of differential linear bounded nonlinear discrete-time systems. An incremental model and an adaptive law with dynamic normalization is introduced to relax the boundedness assumption of nonlinear term to differential linear boundedness. When the zero dynamic system of this system is unstable, a pole-placement control strategy based nonlinear multiple model indirect adaptive control method is proposed. The bounded-input-bounded-output stability of the closed-loop system is proved.(4)A nonlinear multiple model adaptive control method is proposed for a class of uncertain nonlinear discrete-time systems with unknown parameters belonging to a finite set. This method is composed of limited multiple nonlinear candidate controllers and a switching mechanism. Firstly, the input-to-state stability of a class of nonlinear discrete-time switching systems with an average dwell-time switching signal is proved, when all the subsystems of the switching system are input-to-state stable. Secondly, the boundedness of the states of the nonlinear discrete-time system can be proved when the proposed method can guarantee the input-to-state stability of each subsystem. The simulation experiments are given to verify the proposed algorithm.
Keywords/Search Tags:Nonlinear discrete-time system, Switching system, Multiple model adaptive control, Input-to-state stability, Zero-order proximity bounded, Linear bounded, Differential linear bounded
PDF Full Text Request
Related items