基于ANN的自适应PID控制器的仿真研究及单片机实现探讨

Simulation Study of Adaptive PID Controller Based on ANN andDiscussion of Its Implementation on Single Chip Computer

提出一种基于ＡＮＮ（人工神经网络）的自适应ＰＩＤ控制器．通过ＢＰ网络的自学习把增量式数字ＰＩＤ控制算法的参数存贮在神经元间连接权值上．不管由于任何原因使控制品质有所下降时，ＢＰ网络能自动修正权值，表现出对环境的自适应能力．最后，对本控制器的８０９８单片机实现进行探讨．

An adaptive PID(proportional integral and differential) controller based on ANN (artificial neuralnet) is presented in this paper. By self-learning of back propagation (BP) net,the parameters of incrementaldigital PID control algorithm can be stored at interneuronal connected weights. Despite of the decrease of control quality for whatever reasons,BP net will be able to modify automatically the weights and will show itsadaptive ability against environment. The implementation of this controller on MCS-8098 single chip computeris discussed finally.