基于累积灰色径向基函数模型的建筑能耗预测  

Building Energy Consumption Prediction Based on Trigonometric Grey Radial Basis Function Model

作　　者：白燕[1] 贺引娥 武璐璐 BAI Yan*;HE Yine;WU Lulu(School of Science,Xi'an University of Architecture and Technology,Xi'an 710055,Shaanxi,China)

机构地区：[1]西安建筑科技大学理学院,陕西西安710055

出　　处：《制冷技术》2023年第4期46-53,共8页Chinese Journal of Refrigeration Technology

基　　金：陕西省自然科学基金(No.2017JM5019);陕西省建设厅科技发展计划(No.2019-K34);陕西省教育科学规划课题(No.SGH18H111)。

摘　　要：针对公共建筑逐月能耗呈现周期性震荡的特点,本文提出了基于函数组合变换的累积灰色径向基函数(TGM-RBF)神经网络预测模型,以实现对公共建筑能耗的预测。采用西安地区某公共建筑逐月能耗模拟数据及2010年—2017年中国建筑业能耗实测数据,分别构建累积TGM-RBF模型、累积TGM(1,1)模型和GM(1,1)模型对建筑能耗进行预测,并对预测结果进行了对比。结果表明,组合模型较单一模型的预测精度更高,预测精度分别提高了7.31%、6.46%以及0.95%、0.83%。Aiming at the public building with the characteristic of oscillations of the monthly building energy consumption,a trigonometric grey model radial basis function(TGM-RBF)neural network prediction model is proposed based on function combination transformation to realize the prediction of public building energy consumption.Using the monthly energy consumption simulation data of a public building in Xi'an area and the actual energy consumption data of China's construction industry from 2010 to 2017,the cumulative TGM-RBF model,cumulative TGM(1,1)model and GM(1,1)model are constructed respectively The building energy consumption is predicted,and the predictive results are compared.The results show that the prediction accuracy of the combined model is higher than that of the single model,with the prediction accuracy increased by 7.31%,6.46%and 0.95%,0.83%,respectively.

关 键 词：建筑能耗 函数组合变换 累积法 径向基函数神经网络 

分 类 号：TB611[一般工业技术—制冷工程] TU111.195[建筑科学—建筑理论]