采用非线性模块的BP神经网络PID水位预测控制  

PID Prediction Control of Water Level with BP Neural Network and Nonlinear Module

作　　者：郭清[1] 孙蓉[2] 徐立芳[1] 唐明[1] GUO Qing;SUN Rong;XU Lifang;TANG Ming(Engineering Training Centre,Harbin Engineering University,Harbin 150001,China;College of Intelligent Science and Engineering,Harbin Engineering University,Harbin 150001,China)

机构地区：[1]哈尔滨工程大学工程训练中心,哈尔滨150001 [2]哈尔滨工程大学智能科学与工程学院,哈尔滨150001

出　　处：《实验室研究与探索》2022年第8期128-133,共6页Research and Exploration In Laboratory

基　　金：教育部产学合作协同育人项目(2021101383001、202101058004);黑龙江省教育科学“十四五”规划重点课题(GJB1422103)。

摘　　要：因非线性时变控制系统造成的滞后,压水堆核电厂蒸汽发生器(SG)水位控制会引起的较大惯性或耦合,对此提出一种基于非线性模块的BP神经网络PID控制预测模块。采用BP神经网络模型作为非线性预测模块,调用SG水位PID控制运行数据作为非线性预测模块的训练样本;提取BP神经网络模型的输出数据作为BP神经网络PID的输入数据,多重优化BP神经网络的权值及阈值,缩小修正差值逼近系统整定值;构建SG水位非线性模块的预测数值与系统运行数据之间的非线性映射关系,修正带有预测判断方向的阈值,实现在线动态调整输出PID各参数的最优值。实验结果表明,预测模块可有效缩短系统的稳定阶跃响应周期,具有较高的计算精度。To solve the problem of large inertia or coupling caused by nonlinear time-varying control system of steam generator water level control in PWR nuclear power plant,this paper puts forward aprediction method based on BP neural network PID control of nonlinear module.Firstly,a BP neural network modelis established as a nonlinear prediction module,it transfersthe PID controlling operation data of steam generator water level to train the nonlinear prediction module;then extractsthe output data of BP neural network model as the input data of BP neural network PID.The weights and the threshold of BP neural networkare optimized toreduceand correct the difference.Finally,nonlinear mapping relationship isbuilt between the nonlinear module predicted value and the system operation data for the steam generator water level.The threshold is corrected with prediction judgment direction to realizeonline dynamic adjustment output ofeachPID parameterto reach its optimal value.The experimental results show that the proposed method can effectively shorten the system stable step response period,it has higher degree of convergence accuracy.

关 键 词：BP神经网络 非线性模块 PID控制 蒸汽发生器 水位预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]