船用核动力装置二回路PID神经网络解耦控制研究  

Research of PID neural networks decoupling control of marine nuclear power plant

作　　者：孙建华[1] 汪伟[2] 郑可为 

机构地区：[1]华中科技大学控制科学与工程系,湖北武汉430074 [2]武汉第二船舶设计研究所,湖北武汉430064 [3]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2007年第6期656-659,668,共5页Journal of Harbin Engineering University

摘　　要：直流蒸发器的船用核动力装置是一个非线性、时变及多变量强耦合的被控对象.针对该被控对象设计了改进型PID神经网络控制系统,用于船用核动力装置多变量解耦控制,该神经网络控制器不需要对系统进行辨识,在调整权值的学习过程中使控制系统具有良好的解耦控制性能.仿真结果表明,直流蒸发器压力和汽轮机转速控制之间协调性好,并具有响应速度快、鲁棒性好等特点.A marine nuclear power plant with a once-through steam generator （OTSG） is a controllable object with nonlinear, time-variant and strong multiple variants coupling features. A PID neural network （PIDNN） control system consists of three hidden layers of proportional （P）, integral （I）, and derivative （D） neurons. A PIDNN＇s weights are adjusted by back-propagation algorithms and these are improved for the multiple variants decoupling control of a marine nuclear power plant. The proposed neural network controller doesn＇t need to identify the system. The control system can achieve good decoupling control performance during the process of learning to adjust network weights. The simulation results prove that this OTSG controller coordinates pressure and turbine rotational speed smoothly, with quick response and robust capability.

关 键 词：船用核动力装置 PID神经网络 解耦控制 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]