严寒地区低碳建筑能耗预测方法仿真  

Simulation of Low Carbon Building Energy Consumption Prediction Method in Severe Cold Area

作　　者：瞿萧羽 刘子恒 王沫[3] QU Xiao-yu;LIU Zi-heng;WANG Mo(Jilin University of Architecture and Technology,School of Architecture and Planning,Changchun Jilin 130000,China;Environmental and Architectural Engineering Design and Research Branch,Jilin Traffic Planning and Design Institute,Changchun Jilin 130000,China;Academy of Fine Arts,Hainan Normal University,Haikou 570100,China)

机构地区：[1]吉林建筑科技学院建筑与规划学院,吉林长春130000 [2]吉林省交通规划设计院环境与建筑工程设计研究分院,吉林长春130000 [3]海南师范大学美术学院,海南海口570100

出　　处：《计算机仿真》2023年第11期261-266,共6页Computer Simulation

基　　金：吉林建筑科技学院基金项目(校科字[2022]006PTKJ);林省职业教育与成人教育教学改革研究课题(2022ZCY240);2023年度吉林省教育厅科学技术研究项目(JJKH20231359KJ)。

摘　　要：严寒地区建筑耗能较大,绿色化程度较低。为了解决严寒地区低碳建筑能耗预测精度低、速度慢的问题,提出算法组合优化建模的方法,将支持向量回归算法与寒区环境特征数据分析算法有机结合,构建PR-KM-SVR寒区低碳建筑能耗预测回归模型。模型首先采用z-score法,将能耗数据与环境数据归一化处理成无量纲;然后采用Pearson相关系数分析方法,对数据集进行分析降维处理,提高模型训练速度;接着利用改进K-Means算法对数据进行聚类分析,提高模型预测精度;并构建基于RBF核函数的PR-KM-SVR建筑能耗预测回归模型;最后采用十折交叉验证的方法寻得最优gamma参数与惩罚因子参数分别为0.487、7.502。仿真结果表明,PR-KM-SVR模型的R2与MSE评价指标较优,分别达到0.882、0.908,且对建筑能耗预测平均精度超过95%。提出的PR-KM-SVR寒区低碳建筑能耗预测回归模型应用于人机交互界面开发,便于管理人员操作,有利于能源管理系统发挥作用。The buildings in cold regions consume a lot of energy and have a low degree of greenness.In order to solve the problems of low accuracy and slow speed in the prediction of low-carbon building energy consumption in cold regions,this paper proposes an algorithm combination optimization modeling method,which combines the support vector regression algorithm with the data analysis algorithm of cold region environment characteristics,and constructs the PR-KM-SVR regression model for the prediction of low-carbon building energy consumption in cold regions.Firstly,the model uses the z-score method to normalize the energy consumption data and environmental data into di-mensionless;then uses the Pearson correlation coefficient analysis method to analyze and reduce the dimensionality of the data set to improve the model training speed;then uses the improved K-Means algorithm to cluster the data to improve the model prediction accuracy.The PR-KM-SVR building energy consumption prediction regression model based on RBF kernel function is constructed.Finally,the optimal gamma parameter and penalty factor parameter are found to be 0.487 and 7.502,respectively.The simulation results show that the R2 and MSE evaluation indexes of the PR-KM-SVR model are better,reaching 0.882 and 0.908,respectively,and the average accuracy of building energy consumption prediction is more than 95%.The PR-KM-SVR model for predicting energy consumption of lowcarbon buildings in cold regions proposed in this paper is applied to the development of human-computer interaction interface,which is convenient for managers to operate and is conducive to the role of energy management system.

关 键 词：寒区环境 能耗预测 算法组合 

分 类 号：TU111.195[建筑科学—建筑理论]