摘要
PID控制由于其结构简单、稳定性好而被广泛应用,然而在实际的工业过程中,许多被控过程机理复杂,具有高度非线性。在实际PID控制中就要求,PID参数不仅整定不依赖对象数学模型,并且能在线调整,以满足实时控制的要求。对角回归神经网络是一种具有反馈回路的动态神经网络,本文提出了一种基于对角回归神经网络的PID控制器结构,利用对角回归型神经网络辨识控制量权值调整中的未知Jacobian信息,在线整定PID控制器参数。仿真实验中,将此法与基于径向基函数神经网络PID控制效果进行对比,结果显示基于对角回归型神经网络的自适应PID控制器在抗干扰、设定值跟踪动态响应和鲁棒性上都有较明显的改善。
PID control has been widely used because of its simple structure, stable performance. However, in actual industrial processes, many complex controlled mechanism has highly nonlinear.This requires that PID parameters setting not only do not rely on mathematical models, and can be adjusted on line to meet the requirements of real-time control. This paper presents a new type of adaptive PID controller using diagonal recurrent neural network which is dynamic network with reactive structure.The diagonal recurrent neural network is used to identify the undetermined Jacobian information in adusting the PID parameters. In simulation research ,the control effect of this controller is compared with the effect of the controller based on radial basis function neural network.The results prove that this controller based on diagonal recurrent neual network has better adaptation,anti-jamming ability, dynamic and robustness characteristics.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2012年第9期1056-1059,共4页
Computers and Applied Chemistry