变电站为中心配电网电压等高线时序态势图片的聚类分析

Clustering Analysis of Voltage Contour Time-sequence Situation Pictures in Substation-centralized Distribution Network

基于电气接线图并按特定图形设计的时序态势图,其图形特征是网络结构、运行方式、接入源荷及各类电气元件行为、量测数据等动态潮流的一种高维融合映射。通过基于刻画态势图片的关键但数量有限的图形特征量的聚类分析,从中获取态势图中隐藏的规律和有价值的知识。文中提出变电站为中心的配电网电压等高线时序态势图片的特征向量及其计算算法;探索了基于图片特征向量的态势时序图片聚类分析方法,提出了聚类中心图片(典型图片)的图片相似度、时间离散度及其计算算法。以典型日96幅电压态势时序图为案例,介绍了所提方法的效果以及分析结果,并进行了近两个月的长时序图片聚类的初步应用分析。

The time-sequence situation picture based on the electrical wiring diagram is designed according to a specific diagram. Its characteristic vector is a high dimensional mapping of the fusion of dynamic power {lows such as network structure, operation mode, action of load and other equipment, and measurement data. Through the clustering analysis based on the limited number of core characteristic vectors used to describe the time-sequence situation pictures, the hidden rule and valuable knowledge in situation pictures can be acquired. As a trial, the characteristic vector o{ voltage contour situation picture o{ substation-centralized distribution network is defined and its calculating methods are proposed. The characteristic vector based clustering analysis method of time sequence situation pictures is explored. Several new measurement indices for the clustering classification, including picture similarity and time dispersion of clustered center pictures （typical pictures）, are proposed to help with best picture clustering. By taking the 96 voltage time-sequence situation pictures in a typical day as an example, the effect of the proposed method and the analysis results are described, The preliminary clustering analysis of situation pictures with a long time-sequence of approximately two months has been carried on.