基于因素分解法和季节ARIMA模型的居民生活用电预测对比分析

Comparative Analysis of Household Electricity Consumption Forecast Based on Factor Decomposition Method and Seasonal ARIMA Model

居民生活用电是居民家庭生活能源消耗的主要原因之一,对居民生活用电量的准确预测对能源保障和节能减排具有重要意义。居民的生活用电量往往受到长期趋势、季节因素、随机因素等多个因素的影响,合理选择预测方法是提高预测精度的关键。本文基于2013年10月-2019年9月的合肥市居民生活用电月度数据,分别采用因素分解法、加法季节ARIMA模型和乘法季节ARIMA模型三种方法建立模型并进行对比分析与预测,并结合2019年10月-12月数据进行预测效果检验。结果表明乘法季节ARIMA模型在样本区间的拟合效果最好;因素分解法对未来的预测效果最好且更易解释;在各因素存在交互作用的情况下,乘法季节ARIMA模型拟合效果和预测效果均优于加法季节ARIMA模型。

Household electricity consumption is one of the main causes of household energy consumption,Accurate prediction of it is of great significance to energy security,energy conservation and emission reduction.Household electricity consumption is often affected by long-term trend,seasonal factors and random factors.Reasonable selection of forecasting methods is the key to improve the forecasting accuracy.Based on the monthly data from October 2013 to September 2019 of household electricity consumption of Hefei,this paper uses factor decomposition method,the seasonal ARIMA model of addition and the seasonal ARIMA model of multiplication.The results show that the seasonal ARIMA model of multiplication has the best fitting effect in the sample interval;The factor decomposition method has the best predictive effect in the future and it is easier to be explained;In the case of interaction between various factors,the seasonal ARIMA model of multiplication has better fitting effect and predictive effect than the seasonal ARIMA model of addition.