基于改进kmeans++算法的用户分类与电价政策影响分析

Classification of users and analysis of influence of electricity price policy based on improved kmeans++ algorithm

以上海市张江某工业园大工业用户为研究对象,采用大数据方法分析典型用户用电与电价政策的相关性。针对传统kmeans聚类算法的缺点,提出基于复合距离的kmeans++算法,考虑用电曲线的空间相近性与形态相似性,并优化初值选择,具有更好的聚类效果。采用该算法对大工业用户进行聚类,得到大工业具有5种典型行业用户并分析其原因。基于典型行业用户变压器容量和用电量数据,计算各典型行业用电量标杆值并分析其在新电价政策实施前后的波动情况,明确电力公司后续提升工作的重点方向并对不同用户的用电提出合理意见。

The large industrial users of an industrial park in Zhangjiang, Shanghai are studied to analyze the relevance between power consumption benchmark value of typical industry users and power price policy using big data technology. To solve problems of traditional kmeans clustering algorithm, a kmeans ++ algorithm based on composite distance is proposed, which takes into account the spatial and morphological similarities of electricity curves, and optimizes the selection of initial values, so clustering effect is better. This algorithm is used to cluster large industrial users. Finally,five typical industrial users are found in large industry and their reasons are analyzed. Based on the transformer capacity and power consumption data of typical industries, the power consumption benchmark values of each typical industry are calculated and their fluctuations before and after the implementation of the new tariff tphoeli ckye ya rdei raenctailoynze odf. fTohlleoswe-hueplp weolrekc traincidt yp ucto mfoprawnaireds rteo aisdoennatbiflye opinions to different users.