基于自动聚类模型的输电线路外力破坏预警预测

Early warning prediction of external force destruction in transmission lines based on automatic clustering model

外力破坏事件已成为严重威胁架空输电线路安全稳定运行的主要因素,给防御、预警工作带来一定的困难。针对传统的聚类方法聚类中心难以准确确定、易受异常点影响的问题,提出了一种基于自动聚类模型的输电线路外破数据分析方法,对外力破坏数据从时间和空间纬度进行分析。该算法首先通过Canopy算法初始聚类中心,采用削弱不符合正态分布的异常数据权值的思想,利用优化的K-means算法进行聚类处理,最终通过实验分析证明了该算法的有效性及高效性。本文算法能够应用于电力信息系统的GIS模块,实现分析结果的时空可视化,为找到输电线路外力破坏发生原因、进行预警预测提供有力的决策支持。

The external force destruction has become a major threat to the safe and stable operation of overhead transmission lines,bringing difficulties to the defense and early warning work. In order to solve the problem that the traditional clustering center is difficult to accurately determined and susceptible to abnormal points,an automatic clustering method for data analysis work of transmission lines was presented,and external damage data was analyzed from time and space latitude. Firstly,the cluster center was initialized in this method by using Canopy algorithm.Then,the optimized K-means algorithm was used to perform clustering. Finally,the effectiveness of this method was proved by experimental analysis. This method will be applied to the GIS module in the power information system,which can realize the spatio-temporal visualization of the analysis results and provide powerful decision support for finding cause of the external force damage of the transmission line.