摘要
通过相关性分析确定了集中供热系统换热站供/回水均温的影响因素,进一步依据最小二乘拟合计算得到预测模型中历史供热参数的最佳周期,同时结合室外空气温度和室内温度作为模型输入参数,运用Matlab仿真模拟软件建立广义回归神经网络(GRNN)、Elman递归神经网络(Elman)以及多元线性回归(MLR)预测模型,分别对未来18个时刻的供/回水均温进行仿真验证。分析预测结果发现,MLR预测模型的精度最高,GRNN预测模型精度略低于M LR,而Elman模型预测精度最低。
Through the correlation analysis,the influencing factors of the average temperature of supply and return water in heat exchange station of the central heating system are determined,and the optimal period of the historical heating parameters in the prediction model is further calculated based on the least squares fitting calculation.Meanwhile,combined with outdoor air temperature and indoor temperature as the input parameters of the model,the Generalized Regression Neural Network(GRNN),Elman recursive neural network(Elman) and Multiple Linear Regression(MLR) prediction model are established by Matlab simulation software to simulate and verify the average temperature of water supply and return at the next eighteen moments.The prediction results show that the accuracy of the MLR prediction model is the highest,the accuracy of the GRNN prediction model is slightly lower than that of the MLR,and the Elman model has the lowest prediction accuracy.
作者
朱佳
孙春华
齐成勇
夏国强
陈佳丽
ZHU Jia;SUN Chun-hua;QI Cheng-yong;XIA Guo-qiang;CHEN Jia-li(School of Energy and Environmental Engineering,Hebei University of Technology,Tianjin 300400,China;Hebei Gongda Keya Energy Technology Co.,Ltd.,Shijiazhuang 050000,China)
出处
《建筑节能》
CAS
2020年第7期9-13,71,共6页
BUILDING ENERGY EFFICIENCY
基金
“十三五”国家科技支撑计划资助项目(2016YFC0700707)
关键词
供热参数预测
广义回归神经网络
Elman递归神经网络
多元线性回归
prediction of heating parameters
generalized regression neural network(GRNN)
Elman recursive neural network
multiple linear regression(MLR)
