本文已被:浏览 1908次 下载 611次
投稿时间:2019-10-07
投稿时间:2019-10-07
中文摘要: 在楼宇短期负荷预测中,针对单一预测模型难以充分学习负荷时间序列中的特性问题,提出了一种基于自回归差分移动平均-长短期记忆神经网络(ARIMA-LSTM)组合模型的楼宇负荷预测方法。首先,根据灰色关联度选取相似日时间序列数据为训练样本;然后,利用ARIMA模型预测负荷,并将原始数据和ARIMA预测数据之间的误差视为非线性分量;最后,通过LSTM神经网络对误差序列进行校正,得到楼宇短期负荷的最终预测值。通过对上海市某楼宇的预测效果分析,并将其与ARIMA模型、LSTM模型和ARIMA-SVM组合模型进行对比,验证了所提方法能够有效控制预测误差,提高楼宇负荷预测精度。
中文关键词: 楼宇短期负荷预测 自回归差分移动平均模型 长短期记忆神经网络 时间序列 灰色关联度
Abstract:Aiming at the problem that a single forecasting model is difficult to fully learn the characteristics of the time series,this paper proposes a combined model based on ARIMA-LSTM forecasting method.Firstly,the data of similar daily time series are selected as training samples according to the grey correlation degree;then the load is predicted by ARIMA,and the error between the original data and the ARIMA prediction data is regarded as a nonlinear component.The Long-Short Term Memory Network(LSTM) corrects the error sequence to obtain the final predicted value of the building's short-term load.Through the analysis of the prediction effect of a building in Shanghai,and the comparison with ARIMA,LSTM model and ARIMA-SVM combined model,it is proved that the method can effectively control the prediction error and improve the building load forecasting accuracy.
keywords: short-term load forecasting of buildings autoregressive integrated moving average model(ARIMA) long-short term memory(LSTM) time series grey correlation degree
文章编号:20196011 中图分类号:TM715 文献标志码:
基金项目:国家自然科学基金青年科学基金(51607111);上海市科学技术委员会地方院校能力建设项目(15160500800)。
作者 | 单位 | |
李鹏辉 | 上海电力学院 自动化工程学院 | 1075834029@qq.com |
崔承刚 | 上海电力学院 自动化工程学院 | |
杨宁 | 上海电力学院 自动化工程学院 | |
陈辉 | 上海电力学院 自动化工程学院 |
引用文本:
李鹏辉,崔承刚,杨宁,等.基于ARIMALSTM组合模型的楼宇短期负荷预测方法研究[J].上海电力大学学报,2019,35(6):573-579.
LI Penghui,CUI Chenggang,YANG Ning,et al.Research on Short-term Building Load Forecasting Method Based on ARIMALSTM Combination Model[J].Journal of Shanghai University of Electric Power,2019,35(6):573-579.
李鹏辉,崔承刚,杨宁,等.基于ARIMALSTM组合模型的楼宇短期负荷预测方法研究[J].上海电力大学学报,2019,35(6):573-579.
LI Penghui,CUI Chenggang,YANG Ning,et al.Research on Short-term Building Load Forecasting Method Based on ARIMALSTM Combination Model[J].Journal of Shanghai University of Electric Power,2019,35(6):573-579.