基于卷积神经网络的医院建筑公共照明能耗预测研究  

Research on Prediction of Public Lighting Energy Consumption in Hospital Building Based on Convolutional Neural Network

作　　者：张昊冲 王晶晶[1] 钱怡佳 ZHANG Hao-chong;WANG Jing-jing;QIAN Yi-jia(The Second Affiliated Hospital of Soochow University,Suzhou 215004 China)

机构地区：[1]苏州大学附属第二医院,江苏苏州215004

出　　处：《自动化技术与应用》2024年第4期99-102,146,共5页Techniques of Automation and Applications

基　　金：中国工程管理协会“十四五”规划重点课题(JKY15440)。

摘　　要：为提升医院建筑公共照明能耗预测准确性,研究一种基于卷积神经网络的医院建筑公共照明能耗预测方法。对历史公共照明能耗数据实施缺失数据填补、孤立值检测与处理,以此作为输出变量;通过计算灰色关联度选取医院建筑公共照明能耗影响因子,作为输入变量。基于卷积神经网络构建预测模型,以输入、输出变量为基础,完成模型训练,完成实际能耗的预测。结果表明:所研究预测方法的拟合优度值达到极大值,最高可达到0.92,说明该方法的预测结果更为准确,与真实情况更为贴近,具有一定应用价值。In order to improve the accuracy of the prediction of public lighting energy consumption in hospital buildings,this paper studies a prediction method of public lighting energy consumption in hospital buildings based on convolutional neural network.It carries out missing data filling,isolated value detection and processing for historical public lighting energy consumption data as output variables.The influence factors of public lighting energy consumption in hospital buildings are selected as input variables by calculating the grey correlation degree.The prediction model is constructed based on convolutional neural network.Based on input and output variables,the model training is completed and the actual energy consumption is predicted.The results show that the goodness of fit value of the prediction method studied reaches the maximum value,which can reach 0.92,indicating that the prediction result of this method is more accurate,closer to the real situation,and has certain application value.

关 键 词：卷积神经网络 能耗预测模型 公共照明 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]