基于弹性网络模型的月度用电量预测方法

由于现有月度用电量预测所选影响因素较少,无法较为全面地反映与用电量强关联的因素,同时针对高维数据变量筛选和高精度预测等突出难题,文中提出了一种弹性网络用电量预测模型。为了考虑更为全面的影响因素,建立了用电量、气象、经济、交通4类,共340个变量的数据集。首先对8年96个点的高维变量数据进行弹性网络因子筛选,然后使用Granger因果关系分析找出了用电量数据与其它数据的关联关系,对一年范围内的全社会月度用电量使用弹性网络进行预测,预测结果的平均绝对百分误差为3.07%。为验证该模型的有效性,对比向量自回归(VAR)模型,反向传播(BP)模型和最小绝对值收缩和选择算子(Lasso)预测的效果,验证了文中所提方法预测精度较高。

Monthly electricity consumption forecasting method based on elastic network model

The expansion of power grid scale and the improvement of power system automation level provide a data foun-dation and practical basis for accurate power consumption forecast based on big data technology. Aiming at the prominent problems such as high-dimensional data variable screening and high-precision prediction, this paper proposes an elastic network electricity consumption prediction model, which uses 340 variables for power con-sumption, economy, transportation and meteorology, and monthly data of 96 time points. Perform correlation analysis. Elastic network factor screening for high-dimensional variables, and Granger causality analysis to find out the dependence of electricity consumption data and other data, predicting the monthly electricity consumption of the whole society in a year, and the average absolute value of the prediction results. The percentage error is con-trolled within 2%, and a good prediction effect is obtained. Compared with VAR model, BP model and Lasso, the prediction results show that the method selected in this paper has high precision, which verifies the feasibility and effectiveness of the method.