Nonlinear Adaptive Decoupling Control Based On Multiple Models And Neural Networks

Adaptive decoupling control for a class of nonlinear multivariable uncertain systems is a challenging problem all the while. Supported by the National Fundamental Research Program of China (973), "Real-time intelligent control theory and algorithm research for complex industrial processes" (No. 2002CB312201), adaptive decoupling control methods based on multiple models and neural networks respectively for a class of nonlinear multivariable parameter-unknown and parameter-jumping systems are investigated in this thesis. The stability and performance of the closed-loop systems are established, and the simulation researches are developed. Theory analysis and simultation results show the effectiveness of the proposed methods. The main contributions of this thesis are listed as follows:(1) By extending the multiple model neural network adaptive control method of a class of nonlinear single variable systems with asymptotically stable zero dynamics to multivariable systems, a multivariable adaptive control method based on multiple models and neural networks is proposed. The stability and performance of the closed-loop system are established. The restriction that the higher order nonlinear term should be globally bounded is relaxed. The corresponding indirect adaptive control method is also proposed. The stability of the closed-loop system is established. Simulation researches for the above control methods are developed, and the results illustrate the effectiveness of the proposed methods.(2) Based on the above method, an adaptive closed-loop decoupling control method is proposed for a class of nonlinear multivariable systems possibly with unstable zero dynamics. The stability and performance of the closed-loop system are established. The proposed control method is composed of a linear robust adaptive closed-loop decoupling controller, a nonlinear neural network adaptive closed-loop decoupling controller and a switching mechanism. The corresponding indirect adaptive closed-loop decoupling control method is also proposed. The stability of the closed-loop system is established. Simulation researches for the above control methods are developed, and the results illustrate the effectiveness of proposed methods. Theory analysis and simulation results show that the above methods can realize approximately dynamic decoupling and steady state decoupling.(3) To realize dynamic decoupling control, by combining the open-loop decoupling compensator with the above adaptive closed-loop decoupling control strategy, an adaptive open-loop decoupling control method is proposed. The stability and performance of the closed-loop system are established. The corresponding indirect adaptive open-loop decoupling control method is also proposed. The stability of the closed-loop system is established. Simulation researches for the above control methods are developed, and the results illustrate the effectiveness of proposed methods.(4) For a class of nonlinear parameter-jumping systems, based on the above control method, an intelligent decoupling control method composed of a free-running neural network adaptive decoupling controller, a re-initialized neural network adaptive decoupling controller, N-2 fixed parameter neural network decoupling controllers and a switching mechanism is proposd. The stability and performance of the closed-loop system are established. On the background of an injector driven transonic wind tunnel system, simulation researches for the above control method are developed, and the results illustrate the dynamic decoupling between the stagnation total pressure and the Mach number in the test section can be realized under the conditions of step change of the Mach number.