神经网络预测控制在加热炉炉温控制中的仿真研究  

Simulation of Neural Network Predictive Control in the Heating Furnace Temperature Control

作　　者：薛美盛[1] 方醒 闵天 秦宇海 XUE Mei-sheng1, FANG Xing1 , MIN Tian1 ,QIN Yu-hai2(1 School of Information Science and Technology, University of Science and Technology of China 2. Jiangsu Panvieo Energy Saving Technology Co. , Ltd.)

机构地区：[1]中国科学技术大学信息科学技术学院 [2]江苏庞景节能科技有限公司

出　　处：《化工自动化及仪表》2018年第8期590-594,共5页Control and Instruments in Chemical Industry

摘　　要：针对加热炉处于干扰环境下的不确定性、非线性、大滞后和大惯性特点,采用径向基神经网络(Radial Basis Function,RBF)建立加热炉炉温预测模型,以实现炉温预报。同时设计以L-M(LevenbergMarquardt)优化算法为基础的控制器对加热炉炉温进行滚动控制。针对L-M优化算法对初始值敏感的问题,采用RBF逆神经网络动态确定算法初始值。仿真结果表明所提方法在不同工况下均具有较快的调节时间和较小的超调量。Considering the heating furnace＇s uncertainty,nonlinearity,large lag and great inertia in a disturbed environment,RBF（ radial basis function） neural network was adopted to establish the predictive model of furnace temperature. Meanwhile,a controller which based on L-M（ Levenberg-Marquardt） optimization algorithm was designed to control furnace temperature. Considering the L-M algorithm＇s sensitiveness to the initial value,the RBF inverse neural network was adopted to dynamically calculate initial value of the algorithm.Simulation results show that,this proposed method has faster setting time and smaller overshoot under different operating conditions.

关 键 词：加热炉炉温 RBF神经网络 逆神经网络 L-M优化算法 扰动 炉温预测 滚动控制 调节时间 超调量 降低能耗 

分 类 号：TH862[机械工程—仪器科学与技术]