Research On Intelligent Predictive Control And Its Applications Based On MATLAB Mixed Simulation Platform

The predictive control has emerged as a powerful practical control technique especially in the process industry since the 1960s. It conquers some kinds of problem in the industrial process, which would be a handicao of the modern control theory faced. In order to adapt to the control of complex industrial process, intelligent predictive control has started to present as one of the new research domain of model predictive control.The studies and applications of intelligent predictive are discussed in this dissertation. Two new kinds of control strategies are present in this paper, and some are applied in the mixed simulation platform. The simulation results show these algorithms have good control performance. There are six chapers in this paper, and the main contents and conclusions are summaeized as follows:The fore two chapters of this paper, it completely reviews the development, present state and perspectives of inteligent predictive control. And the main characteristics of model predictive control and the algorithm of generalized predictive control are also discussed. At last it introduces the structure and rationale of neural networks and its application in the predictive control field.In chapter 3, the implementation and principle of mixe simulation platform are introduced in detail. It contains the system configuration of mixed platform, the process of the real-time code generation by using RTW, the function of the real-time kernel and its support to I/O board. A simulation platform based on PC for designing and testing virtual controller is constructed, which successfully realizes real-time control with a PC under the windows.In chapter 4, it introduced the nonlinear predictive controller based a single neural network. By enhancing the traditional cost function, a new multi-step predictive cost function with a differential item is provided, with which we design the predictive controller for SISO and MIMO nonlinear systems. At last these algorithms are applied in some real system of the mixed simulation platform, and they all acquire valid control.In chapter 5, a PID neural network is used to complete the optimizing calculation. It provides a kind of association scheme between PID neural network and predictive control, and deduces the method of forward calculation and weight-correction of PIDNN in detail.In chapter 6, the conclusion of the thesis, the work and the future research and problem are presented.