基于局部密度的最小生成树聚类算法及其在电力大数据的应用

Minimum Spanning Tree Clustering Based on Local Density and Its Application to Power Big Data

摘要:

电力大数据主要来源于电力生产和电能使用的发电、输电、变电、配电、用电和调度各个环节，,如何运用这些数据提高电力管理工作的智能化水平已经成为相关电力环节十分重要的研究课题之一。但现有电力大数据中用到的聚类方法却不能发现任意形状的数据集聚类类别（即类簇），这影响了电力大数据在应用中的计算精度与计算时长。本文提出了一种新的算法，即使用局部密度峰值和基于共享邻点的距离，更好地结合了密度与距离的关系，表示出数据之间的差异。使用局部密度峰值并用基于共享邻点的距离来构造最小生成树，然后重复切割最长的边，直到找到给定数量的簇。在电力大数据的的应用上的实验结果表明，该算法在电力大数据应用中具有良好的效果。

Abstract:

Power big data mainly comes from all aspects of power generation, transmission, transformation, distribution, power consumption and dispatching of power production and energy use. How to use these data to improve the intelligent level of power management has become one of the most important research topics of the related power links. However, the existing clustering methods used in power big data can not find clusters of arbitrary shape, which affects the calculation accuracy and calculation time in the application to power big data. A new algorithm is proposed, which uses the local density peak and the distance based on shared neighbor points to better combine the relationship between density and distance and express the differences between data. The minimum spanning tree (MST) is constructed by using the local density peak and the distance based on the shared neighbor, and then the longest edge is cut repeatedly until a given number of clusters is found. The experimental results show that the proposed algorithm has a good effect in the application to power big data.