摘要
由于受多种外界因素的影响,春节前后月度售电量的预测一直存在较大的误差。针对这一问题,提出了一种基于X12季节分解方法、ARIMA模型及因子分解机的综合预测模型。首先采用X12季节分解法将月度售电量历史数据分解为趋势分量、季节分量及随机分量;然后使用ARIMA模型和历史数据平均法分别对趋势分量、随机分量和季节分量进行预测;最后应用因子分解机算法对历史1—3月份售电量占季比与对应月份首日距春节当日的天数进行回归分析,以此对预测结果进行修正。结果表明:采用Eviews软件对历史数据进行统计与计算,将预测结果与传统ARIMA模型及TRAMO-SEATS模型进行对比,预测精度得到较大提高。
Due to a variety of external factors,there have been still errors in the prediction of monthly electricity sales before and after the Spring Festival.For this problem,a comprehensive model is proposed based on X12 seasonal adjustment,ARIMA model and factorization machine.Firstly,X12 seasonal method is used to decompose the historical data of monthly electricity sales into seasonal trend-cycle,seasonal factors and irregular component;then ARIMA model and historical data average method are used to predict the trend component,irregular component and season component respectively;finally,the factor decomposition machine algorithm is used to make regression analysis on the seasonal ratio of sales volume of historical 1,2 and 3 months and the days from the first day of corresponding month to the Spring Festival Day,so as to modify the prediction results.Eviews software is used for historical data statistics and calculation and the prediction accuracy is greatly improved as it compared the traditional ARIMA model with TRAMO-SEATS model.
作者
王安东
王东涛
叶剑华
WANG An-dong;WANG Dong-tao;YE Jian-hua(Tianjin Key Laboratory of Information Sensing and Intelligent Control,Tianjin University of Technology and Education,Tianjin 300222,China;School of Automation and Electrical Engineering,Tianjin University of Technology and Education,Tianjin 300222,China)
出处
《天津职业技术师范大学学报》
2020年第4期42-47,共6页
Journal of Tianjin University of Technology and Education
基金
天津市自然科学基金资助项目(18JCQNJC74500)
天津市津南区科技计划项目(20161508).
关键词
月度售电量
预测模型
因子分解机
X12季节调整
春节调整
monthly electricity sales
forecasting model
factorization machine
X12 seasonal adjustment
Spring Festival adjustment
