基于改进的BP神经网络模型参考自适应控制

摘要

Because of some problems of the model reference adaptive control based on the traditional BP algorithm,such as the real-time performance, low precision, slow training and so on, and combining the improved BP algorithm and the reversibility of the nonlinear system, this paper proposes the model adaptive control based on the improved bi-phase weight adjusting algorithm to train neural networks. Using the proposed algorithm, the network structure of the system identification and controller is simple and precise. The simulation results show the algorithm presents excellent effects in identifying and controlling the nonlinear system,and it is suitable for engineering process.%针对传统BP算法的神经网络模型参考自适应控制实时性差、精度不高、收敛慢等不足,结合BP改进算法和非线性系统的可逆性,提出了基于改进的双向权值调整BP算法的神经网络模型参考自适应控制.基于此算法设计的系统辨识器和控制器的网络结构简单,精度高,仿真结果表明该算法的辨识和控制效果均很理想,可应用于工程实际.