摘要
提出一种基于RBF辨识神经网络算法的改进神经网络PID控制,应用最优控制理论中的二次型性能指标加入到控制算法中的加权系数学习修正部分,将RBF与单神经元相结合构成PID控制器,通过Matlab对指定对象仿真控制,得到了良好的效果。比起以往的神经网络PID控制,改进的算法在学习速度和实时性都得到了提高。
An improved neural network PID control based on RBF neural network is brought forward. Quadratic model in optimization is used as the guideline in the regulation of the coefficient. By combining them a PID controller is formulated. Simulating by Matlab on an object, a good result is gained. Compared with previous neural network PID control, the improved arithmetic has better real time capability and learning rate.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2005年第z2期345-346,共2页
Chinese Journal of Scientific Instrument
关键词
神经网络
PID控制
单神经元
仿真
Neural network PID control Single nerve cell Simulating
