论文部分内容阅读
分析了BP网络辨识器和控制器的常规训练方法存在的不足之处及其限制,提出了一种通过BP网格辨识器的灵敏度网络反传对象的输出误差来训练BP网络控制器的新方法,根据BP网络和自适应控制的相容性,提出了一种新的BP网络自适应控制结构,并利用BP网络辨识器的灵敏度网络训练BP网络控制器的思想,来调整BP网络控制器的参数,等离子喷徐过程控制模型的训练研究表明其效果良好。
This paper analyzes the shortcomings and limitations of the conventional training methods of the BP network identifier and controller, and proposes a new method to train the BP network controller through the output error of the backpropagation object of the BP network identifier , A new BP network adaptive control structure is proposed according to the compatibility between BP network and adaptive control. The idea of BP network controller trained by the sensitivity network of BP network identifier is used to adjust the BP network controller The parameter and plasma jet Xu process control model training research shows its effect is good.