基于对角递归神经网络整定的PID解耦单元机组负荷控制系统  

Load Control System of Thermal Power Sets Based on Self-tuning PID Decoupling Control with Diagonal Recurrent Neural Network

作　　者：刘红军[1] 韩璞[1] 于希宁[1] 

机构地区：[1]华北电力大学自动化系,保定071003

出　　处：《动力工程》2004年第6期809-812,818,共5页Power Engineering

摘　　要：针对火电厂单元机组具有多变量强耦合、非线性及参数时变的受控对象,提出了基于对角递归神经网络整定的PID解耦控制方法,其主要特点是能够提供一个对角递归神经网络来辩识系统模型,进而对PID控制器参数进行整定,实现多变量解耦控制。通过对火电机组负荷控制系统的设计和仿真研究,表明系统达到了动态近似解耦、静态完全解耦和无静差跟踪,并具有响应速度快,鲁棒性好等特点。A PID decoupling control method is proposed on the basis of diagonal recurrent neural network (DRNN) for thermal power sets which are featured by multiparious variables, strong coupling, nonlinear and parameter time-varying parameters. The DRNN is able to identify system models, tune PID controller parameters and therewith realize multivariable decoupling control as required. Simulation results of the load control system of a thermal power set show that ,with DRNN, quasi-dynamic decoupling and complete static decoupling can be attained, simultaneously featured by zero static error, quick response and strong robust capability. Figs 5 and refs 6.

关 键 词：自动控制技术 单元机组 对角递归神经网络(DRNN) 解耦控制 PID控制 负荷控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]