基于双隐层DRNN的自整定PID解耦控制  

Self - tuning PID Decoupling Controller Based on DRNN with Double Hidden Layers

作　　者：吴星刚[1] 姜长洪[1] 姜楠[2] 

机构地区：[1]沈阳化工学院信息工程学院,辽宁沈阳110142 [2]沈阳化工学院计算机科学与技术学院,辽宁沈阳110142

出　　处：《计算机仿真》2005年第6期112-114,共3页Computer Simulation

摘　　要：随着科学技术的不断进步和发展,被控对象正变得越来越复杂,而人们对其控制精度的要求却日益提高,这样就产生了复杂性和精确性的尖锐矛盾。智能自适应控制是解决上述问题的有效方法之一。该文针对强耦合带延时多输入输出(MIMO)非线性离散系统难以实现解耦的问题,推导了双隐层DRNN的权值学习算法,实现了基于双隐层DRNN结构与动态BP网络的加速算法的在线自整定PID解耦控制,仿真表明该方案具有良好的动态、静态性能以及很强的自适应性。With the development of science and technology, the controlled systems have became more and more complicated, but the accuracy requirement to the controlled systems have increased. Thus there are some conflicts between the complexity and accuracy. Intelligent adaptive control is an effective method for resolving this problem. Aiming at the difficult decoupling problem in non - linear discrete system with time - delay, strong coupling and MI-MO , this paper deduced the weight learning algorithm of DRNN with double hidden layers, and an online self - tuning PID decoupling controller based on the structure of DRNN and dynamic BP acceleration algorithm is proposed. The simulation results show that the control system has perfect dynamic and static performances.

关 键 词：双隐层动态递归神经网络 解耦控制 多输人多输出系统 比例积分微分控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]