一种基于双层聚类分析的负荷形态组合识别方法

A New Combinational Electrical Load Analysis Method Based on Bilayer Clustering Analysis

区别于传统用户用电行为分析方法,提出一种以聚类算法为基础的双层聚类分析方法。该方法结合给出的内、外层变随机设置为有目的选取初始聚类中心的选取规则,解决了聚类算法受初始聚类中心随机选取的影响,其收敛容易陷入局部最小化的问题。利用余弦相似度形态相似作为外层聚类的判据、欧式距离相近作为内层聚类的判据,对不需要经过归一化处理的用户用电轨迹向量进行分类。最后对某地区电力用户日负荷曲线进行算例分析,结果表明:双层聚类组合方法能把不同负荷形态及其大、小用户准确识别出来,实现了地区负荷形态的自动分类识别功能,证明了上述方法的有效性和优越性。

Different from traditional electricity customer behavior analysis method,this paper presents a combinational load profile recognition method with two layers based on improved clustering algorithm.In order to overcome defect of traditional clustering algorithm,sensitive to initial conditions and usually leading to local minimum results,selection rules for two layers of initial cluster centers are proposed.It changes cluster center selection from random set to purposeful selection.On outer layer,it uses cosine similarity function as its performance evaluation index; on inner layer,Euclidean distance function is used as evaluation index.For the proposed method,it is not necessary to normalize load curve vector in analysis.This method is utilized to analyze calculation examples and daily electricity consumption curves of whole customers in a certain area.The method can automatically recognize cluster number,load categories and load distributions.The result verifies effectiveness and superiority of the proposed method.