基于混合聚类分析的区域短期用电量预测方法研究

Research on Forecasting Method of Regional Short-term Electricity Consumption Based on Hybrid Cluster Analysis

提出了一种基于功能聚类、极限学习机和混合模型的区域短期用电量预测方法。该方法使用功能聚类算法对用电量曲线进行分组,随后针对聚类分组使用极限学习机模型进行用电量预测,最后使用线性回归方法对独立的分组模型极限进行混合实现对区域客户的短期整体用电量预测。此外该方法还使用温度分区策略提高聚类分组的合理性。实验表明该方法能够提高最终用电量预测的准确性。

This paper proposes a regional short-term electricity consumption prediction method based on functional clustering,extreme learning machines and hybrid models.This method uses a functional clustering algorithm to group power consumption curves,then uses a limit learning machine model to predict power consumption for clustering grouping,and finally uses a linear regression method to mix independent grouping model limits to achieve short-term regional customer overall electricity consumption forecast.In addition,the method uses the temperature partitioning strategy to improve the rationality of clustering and grouping.Experiments show that this method can improve the accuracy of the final electricity consumption prediction.