摘要
月售电量是具有趋势性、季节性和随机性的非平稳负荷,直接预测难度较大。为解决该问题,结合X12乘法模型与差分自回归移动平均(ARIMA)模型提出一种新的月售电量预测方法。首先,用X12乘法模型将历史月售电量分解为趋势分量、季节周期分量和随机分量,其中趋势分量用ARIMA模型预测,季节周期分量和随机分量分别用加权法和平均法预测;然后,用乘法模型将上述3个分量的预测值还原为最终的月售电量预测值。该方法可避免直接预测月售电量时不同分量间的相互干扰,提高预测精度;最后用重庆市铜梁区实际数据进行仿真分析。仿真结果表明,相对于ARIMA和季节ARIMA模型对月售电量序列直接建模预测的方法,所提方法具有更高的预测精度。
It is difficult to forecast the monthly electricity consumption directly because the monthly electricity consumption is non-stationary load which contains a trend,seasonality and randomness. In order to solve this problem,a new forecasting method of monthly electricity consumption is proposed based on the multiplication model of X12 method and autoregressive integrated moving average（ARIMA）model. First of all,the multiplication model of X12 method is applied to decompose the electricity series to trend component,seasonal cycle component and random component. ARIMA model is used to forecast the trend component,weighting method and average method are used to forecast the seasonal cycle component and random component respectively. Then,the multiplication model of X12 method is applied to fuse above three predictive values as the final predictive value of the monthly electricity consumption. This method can avoid the interference of different components when it predicts the monthly electricity consumption directly,and thus improve prediction accuracy. Finally,simulation analysis is made with the actual data of Tongliang district in Chongqing. The results show that the proposed method has higher prediction accuracy compared with the method which forecast monthly electricity consumption directly using ARIMA model and seasonal ARIMA model.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2016年第5期74-80,共7页
Proceedings of the CSU-EPSA
关键词
X12乘法模型
差分自回归移动平均模型
月售电量预测
分解
还原
multiplication model of X12
autoregressive integrated moving average（ARIMA）model
prediction of monthly electricity consumption
decompose
fuse
