摘要
随着近年来建筑能耗的迅速增长,建筑节能成为可持续发展战略的一个重要问题,因此,构建一个可以快速精准预测建筑能耗的模型成为实现建筑节能的关键一步.本文结合集合经验模态分解,将LSSVR与ARIMA相融合,构建了一种基于时间序列多尺度分解的预测建筑用电负荷数据的EEMD-LSSVR-ARIMA混合模型.该模型通过EEMD将建筑用电负荷数据分解为多个频率不同的分量,使用LSSVR模型预测高频分量以及用电负荷数据与各分量之和的差值序列,使用ARIMA模型预测低频分量,最后将各分量的预测结果以及差值序列的预测结果叠加得到最终的预测结果.并通过某建筑的用电数据进行实验分析,通过与传统的ARIMA和EEMD-ARIMA模型以及基于残差的ARIMA-LSTM模型进行对比,实验结果表明,本文提出的模型预测精度达到了98%以上,与其他模型相比预测精度提升了将近2%.
With the rapid growth of building energy consumption in recent years,building energy conservation has become an important issue in the sustainable development strategy.Therefore,building a model that can quickly and accurately predict building energy consumption has become a key step to achieve building energy conservation.In a study reported in this paper,combined with empirical mode decomposition,LSSVR and ARIMA were integrated to construct a mixed model of EEMD-LSSVR-ARIMA based on time series multi-scale decomposition to predict building electricity data.With this model,the electricity load data of the buildingwas decomposed into a number of different-frequency components by EEMD,the LSSVR model was used to predict the high-frequency component,and the difference between load data and the sum of the components in the sequence,and the ARIMA model was used to predict the low-frequency component.Lastly,the predicted results of the components and those of the difference sequences were superimposed,and the final results were obtained.This method was employed to analyze the data of power consumption of a certain building,and was compared with the traditional ARIMA and EEMD-ARIMA models andthe residual-based ARIMA-LSTM model.The experimental results showed that the prediction accuracy of the model proposed in this paper was more than 98%,which was nearly 2%higher than that of theother models.
作者
鞠亚轩
张春雨
朱仁敬
朱习军
JU Ya-xuan;ZHANG Chun-yu;ZHU Ren-jing;ZHU Xi-jun(School of Information Science and Technology,Qingdao University of Science and Technology,Qingdao Shandong 266100,China)
出处
《西南大学学报(自然科学版)》
CAS
CSCD
北大核心
2020年第10期8-16,共9页
Journal of Southwest University(Natural Science Edition)
基金
山东省重点研发计划基金(2015GSF119016).
关键词
能耗预测
EEMD
LSSVR
ARIMA
energy consumption prediction
EEMD(Ensemble Empirical Mode Decomposition)
LSSVR(least square support vector regression)
ARIMA(Auto-regressive Integrated Moving Average)