摘要
BP网络与PID控制器相结合,可以实现对PID控制器参数的优化调整。但是BP网络的隐含层层数和神经元节点数的选取尚无定则,需要反复的计算论证才能确定;并且网络连接权重初值选取为随机值,难以保证系统初始运行的稳定。本文提出一种将BP神经网络与PID控制规律融合的新方法--PID神经网络,该方法控制结构简单、系统参数物理意义明确,同时又克服了上述网络的诸多缺点。将该方法应用于对发动机油门开度的仿真控制,仿真结果表明该控制器大大改善了发动机油门控制系统的性能,仿真效果良好。
PID parameters can be adjusted with BP neural network,yet it has some disadvantages,for instance,it is difficult to choose the number of hidden layer and neuron nodes,and requires repeated calculations to determine the argument;Network connection weights are randomly selected as the initial value,so this will make it difficult to guarantee the stability of the system in initial operation.This paper uses a new method--PID neural network which combining the BP neural network and PID control theory,to control the engine throttle.Its control construction is simple and the system parameters physics significance is clear,moreover it has overcome many shortcomings of the BP neural network.The simulation effect shows that the control in engine throttle system based on PID neural network is good in flexibility and robustness.
出处
《自动化技术与应用》
2010年第8期1-4,共4页
Techniques of Automation and Applications
基金
黑龙江省自然科学基金(编号F2008003)
