基于加和比例分配算法的中长期电量预测改进方法

Improved medium and long-term electricity forecasting method based on summation-apportionment algorithm

针对中长期电量预测工作中,年度、季度、月度预测结果不统一,以及1、2月供电量因受春节影响而导致预测误差较大的问题,提出了基于加和比例分配算法的电量预测改进方法。加和比例分配算法包括季度加和算法和比例分配算法2部分。首先利用组合预测算法对季度电量数据序列进行预测;然后利用季度加和算法对季度预测值进行加和得到年度预测值;接着利用比例分配算法对季度预测值进行分解,得到月度预测值,从而得到年度、季度、月度的统一预测结果。最后利用某供电局的实际供电量数据对该方法进行校验,结果证明该方法不仅能够得到统一的预测结果,且其预测精度较普通方法更高。

For medium and long term electricity consump- tion forecasting, the annual, quarterly, monthly forecasting re- sults can not be unified, and the forecasting results accuracy of January and February were reduced by the influence of Spring Fes- tival. So an improved medium and long term electricity forecasting method based on summation- apportionment algorithm was pro- posed. Quarterly summation algorithm and apportionment algo- rithm were included in the sunmmtion-apportionment agorithm. In the summation-apportionment agorithm, combination forecast ago- rithm was firstly used to forecast the quarterly data. Then the quar- terly summation algorithm was used to get the annual forecast re- suits by summing up the quarterly results. Finally, the quarterly re- suits were divided into monthly results by using the apportionment algorithm. After the above steps, the annual, quarterly, monthly forecasting resuhs were unified. At the end of this paper, actual electricity data was used to verify the improved electricity forecasting method. The results showed that this method can not only get the unified forecasting resuhs, but also improved forecasting accuracy.