基于聚类和随机矩阵理论的用电行为刻画方法

针对配网大数据应用背景下难以建立用户用电行为刻画模型的问题,提出一种基于聚类和随机矩阵理论的电力用户用电行为刻画方法.首先利用K-means聚类法对海量用户用电特征数据进行分析,根据不同用电模式对用户进行群体划分.然后基于随机矩阵理论建立用户用电行为分析模型,利用各用户群体的经济数据、气候数据及电力价格数据等辨识与用户群体用电量相关联的因素,实现对电力用户用电行为的刻画.最后通过甘肃省武威市电网实际用电数据验证所提方法的有效性和准确性,为电力精准营销和制定电力需求侧响应策略提供数据支撑.

Research on characterization of electricity consumption behavior based on clustering and random matrix theory

In order to solve the problem that it is difficult to establish a model to describe the user’s electricity consumption behavior under the background of big data application in distribution network, a method based on clustering and random matrix theory is proposed to describe the user’s electricity consumption behavior. Firstly, the K-means clustering method is used to analyze the power consumption characteristic data of massive users, and the users are divided into groups according to the clustering results under different power consumption modes. Then, the model of user electricity behavior analysis is established based on the random matrix theory, through the analysis of the economic data, climate data and electricity price data of each user group, the factors related to the power consumption of the user group are identified, and describes the power user’s electricity consumption behavior. Finally, the effectiveness and accuracy of the proposed method is verified by the actual power consumption data of Wuwei power grid in Gansu Province, which provides data support for accurate power marketing and power demand side response strategy formulation.