摘要
在一般的基于神经网络的智能ＰＩＤ控制基础上，针对受控模型参数、系统设定值以及扰动幅值在大范围内变化的情况，探讨并建立了ＢＰ反传算法学习速率与受控模型参数、系统设定值以及扰动幅值的关系，神经网络控制器可在线调整学习速率，使控制系统具有较强的适应能力和较好的调节品质．
This paper studies the automatic tuning of PID controller based on Neural Networks, and establishes the relationships between the learning rate of BP backpropagation and controlled model′s parameters, set point and disturbance of the system. The simulation study indicates that the system not only posseses better robustness and fast following performence but also has capability of achieving better control character by adjusting the learning rate on line when controlled model′s parameters,set point and disturbance of the system change over a wider ange.〖KH*2D]
出处
《北京理工大学学报》
EI
CAS
CSCD
1997年第4期481-487,共7页
Transactions of Beijing Institute of Technology
