基于动态神经网络的自整定PID控制策略

A Control Method for PID Optimizing Parameters Based on Dynamic RBFNN

提出了在采用最近邻聚类算法在线构造RBF(NN1)正向辨识器,并在线辨识被控对象的Jacobian阵的基础上,引入RBF(NN2)对PID控制器参数进行在线调整的算法。该算法可以实现PID控制参数的自动在线整定和优化。通过和BP网络调整控制器参数的方法的对比和仿真实验证明,该方法控制精度高、响应速度快,并且具备较强的自适应性和鲁棒性。

A new method is proposed,in which parameters of the PID controller are optimized online by RBF （NN2）based on constructing forward RBF （NN1） identifier online to identify the Jacobian matrix of the controlled plant by means of adopting the nearest ne ighbors clustering algorithm.The deep analyzing and comparison to BP net are made.The simulation results demonstrate that the method has perfect dynamic performances,quick response speed,and upper precision,and possesses great adaptability and robustness.