基于输入输出数据的非线性系统建模与实验-控制理论与控制工程专业论文.docxVIP

Abstract With the improvement of complexity and control requirements of electromechanical systems, efficient control algorithm need to be used in controller design, and that is inevitably need to know the system model. However, for complex system, it is difficult to establish accurate mathematical model for the system. But a substantial amount of experimental data which are associated with such systems, and expert experience and knowledge usually can be obtained. In these premises, how to obtain the system model becomes an urgent problem in system control. Firstly, according to the experiments, it is found that the system model is always changed with the different of excitation signals. In order to obtain the accurate model of the electromechanical system, several common excitation signals are applied to establish models respectively based on the hardware-in-loop simulation platform. Then by comparing those models on different excitation signals, the most suitable excitation signal is acquired for constructing actual system. Then, for the general nonlinear phenomenon in projects, the series nonlinear models are studied in this paper. Firstly, orthonormal basis functions are used for constructing Hammerstein system and the method how to structure basis functions is improved. Secondly, blind identification approaches for Wiener systems and Hammerstein-Wiener systems are proposed. By special disposal approach of the input and output data, all internal variables can be recovered solely based on the output signal and parameters are estimated by the least squares. Finally, simulation and experimental results show the effectiveness of the method. At last, a selection adaptive algorithm for the construction of feedback nonlinear model is presented in this paper. Firstly, a set of nonlinear basis functions and the desired error performance are constructed .Secondly, combing the output error between the actual system and the reference the model with the performance indicat