摘要
针对中国“三北”地区(东北、西北、华北)在供暖季利用电极式锅炉调峰控制困难的问题,建立了电极式锅炉电功率与主蒸汽压力的神经网络的预测模型。在详细分析了电极式锅炉原理的基础上,确定了神经网络的输入输出参数,建立了电极式锅炉的NARX和时延Elman神经网络模型,通过测试数据对两个模型进行了比较。结果显示,NARX神经网络在训练过程中更加节省时间,在预测时具有更小的预测误差。确定NARX神经网络为电极式锅炉的预测模型,能够较好的预测电功率与主蒸汽压力,反应电极式锅炉的动态特性,为电极式锅炉在调峰控制中提供了一定的理论基础。
Aiming at the difficulty of controlling the peaking control of the electrode boilers in the"Three North"area(northeast,northwest and north)in China,neural network prediction models for the electric power and main steam pressure of the electrode boiler were established.On the basis of detailed analysis of the electrode boiler,the input and output parameters of the neural network were determined,and the NARX and the delay Elman neural network models of the electrode boiler were established.The two models were compared through the test data.The results show that the NARX neural network can save time during the training process and has a smaller prediction error in prediction.We determined the NARX neural network as the prediction model of the electrode boiler,which can better predict the electric power and main steam pressure and reflect the dynamic characteristics of the electrode boiler.It provides a certain theoretical basis for the peak regulation control of electrode boilers.
作者
马进
强硕
袁雪峰
MA Jin;QIANG Shuo;YUAN Xue-feng(School of Control and Computer Engineering,North China Electric Power University,Baoding Hebei 071003,China)
出处
《计算机仿真》
北大核心
2019年第9期126-130,共5页
Computer Simulation
关键词
电极式锅炉
神经网络
预测模型
建模与仿真
动态特性
调峰
Electrode boiler
Neural network
Prediction model
Modeling and simulation
Dynamic characteristics
Peak-shaving
作者简介
马进(1971-),男(汉族),河北省保定市人,博士,副教授,主要研究领域为热力系统仿真建模、人工智能应用、核能发电技术;强硕(1991-),男(汉族),河北省保定市人,硕士研究生,主要研究领域为火电厂系统建模与仿真、神经网络应用;袁雪峰(1993-),男(汉族),山东省潍坊市人,硕士研究生,主要研究领域为火电厂系统建模与仿真。
