使用计量数据和聚类算法检测非技术损失

Using Measurement Data and Clustering Algorithms to Detect NTL

减少非技术损失(NTL)是实施智能电网所带来的潜在利益的重要组成部分。提出了一种基于智能电表数据的聚类算法来检测窃电和其他原因所导致的非技术性损失。通过对智能电表采集的数据进行聚类,提取正常用电行为的数据原型。然后对待检测数据样本和正常数据的聚类中心之间的距离进行计算,如果距离明显,则将其分类为NTL数据样本。最后对四种不同的异常用电指标进行空间分析,使分类结果更易于可视化。实验表明,基于GA聚类算法的NTL检测方法具有优于同类检测方法的性能。

Reducing NTL is an important part of the potential benefits of implementing a smart grid.This paper proposes a clustering algorithm based on smart meter data to detect non-technical losses caused by electricity theft and other causes.By synthesizing the data collected by the smart meter,the data prototype of the normal power usage behavior is extracted.The distance between the test data sample and the cluster center of the normal data is then calculated,and if the distance is significant,it is classified as an NTL data sample.Finally,spatial analysis of four different abnormal power consumption indicators makes the classification results easier to visualize.Experiments show that the NTL detection method based on GA clustering algorithm has better performance than similar detection methods.