摘要
针对常规PID控制器对于复杂的、动态的和不确定的系统控制还存在着许多不足之处,提出一种基于模糊RBF神经网络的PID自适应控制方法。首先用改进的模糊K-均值聚类算法初始化RBF神经网络的隐层节点中心和基函数宽度,再采用梯度法优化RBF神经网络自适应地整定PID的三个参数。仿真结果表明该学习算法的实用性和有效性。
Analyses the traditional PID controller weak point in controlling the complicated, dynamic and uncertain system. In order to achieve the goal of improving traditional PID controller, presents a novel approach of PID adaptive control based on constructs RBF neural networks with the optimized K-means algorithm is proposed in virtue of the disadvantage of conventional PID control. Firstly uses a modified fuzzy K-means clustering algorithm to initialize the RBF neural network hidden layer nodes and the width of basis functions, then the gradient method opti- mizes the RBF neural network to adaptively tuning the three parameters of the PID. The simu- lation results show the practicality and effectiveness of the new algorithm.
出处
《现代计算机》
2012年第19期3-6,共4页
Modern Computer