计算机应用与软件2013，Vol.30Issue(5)：129-131,136,4.DOI:10.3969/j.issn.1000-386x.2013.05.037
短期燃气预测的信息熵组合模型研究
STUDY ON INFORMATION ENTROPY COMBINATION MODEL FOR SHORT-TERM GAS LOAD PREDICTION
摘要
Abstract
Forecasting city gas load has an important role for the intelligent gas system in smart city and is a challenging work as well.Aiming at the characteristic of periodic gas load in our region,and in order to improve the accuracy of prodiction,in this paper we put forward an information entropy-based forecasting method in which the ARIMA and BP neural network parallelly combine the model.Based on preprocessing the original data on its outliers,this method first forecasts with ARIMA method,decomposes the gas load TS into trend TS and seasonal TS and model them respectively； then it adopts difference training method in BPNN to forecast the load； finally,it combines the two methods based on the information entropy theory so as to give the prediction on the daily gas load in next few days.The contrast experiment results of the above three kinds of method verify the feasibility and effectiveness of the information entropy combination model of the short-term gas prediction.关键词
信息熵/差分自回归移动平均模型/BP神经网络/时间序列预测Key words
Information entropy/ Autoregressive integrated moving average (ARIMA) model / Back propagation neural network / Time series forecasting分类
信息技术与安全科学引用本文复制引用
祖广磊,陆黎明,徐晓钟..短期燃气预测的信息熵组合模型研究[J].计算机应用与软件,2013,30(5):129-131,136,4.基金项目
上海市科学技术委员会项目(11510502400) （11510502400）
