基于kNN算法的智能电网5G海量接入数据异常检测  

Abnormality detection of 5G massive access dataof smart grid based on kNN algorithm

作　　者：林舒嫄 林晓敏 欧亚 阚双星 莫裕全 LIN Shuyuan;LIN Xiaomin;OU Ya;KAN Shuangxing;MO Yuquan(Fujian Power Trading Center Co.,Ltd.,Fuzhou,350001,China;Fujian Power Supply Service Co.,Ltd.,Fuzhou,350001,China;Shanghai Hill Management Consulting Co.,Ltd.,Shanghai,201210,China)

机构地区：[1]福建电力交易中心有限公司,福建福州350001 [2]福建省供电服务有限责任公司,福建福州350001 [3]上海希尔企业管理咨询股份有限公司,上海201210

出　　处：《粘接》2024年第2期155-158,共4页Adhesion

摘　　要：为了更好适应智能电网高维数据异常识别,提出了一种加权kNN数据异常值检测识别方法,该方法使用Z阶曲线来识别kNN。利用Z阶曲线,提出了一种加权kNN异常数据检测方法。用信息熵衡量所有属性的重要性,用Z阶曲线对高维数据进行编码并映射为Z值。实验结果表明,智能电网集群计算节点的数量越多,算法的运行速度就越短。发电数据异常检测准确率达到最高99.2%,较随机森林算法提高8.165%。且kNN算法的运行时间均优于随机森林算法运行时间,最小算法运行时间为4 s,进一步表明kNN算法可有效检测智能电网5G海量接入数据。In order to better adapt to the identification of high-dimensional data anomalies in smart grids,a weighted kNN data anomaly detection and recognition method was proposed,which used Z-order curves to identify kNN.A weighted kNN anomaly data detection method was proposed using Z-order curves.The importance of all attributes was measured with information entropy,and high-dimensional data was encoded and mapped to Z-values with Z-order curves.The experimental results showed that the more computing nodes there were in the smart grid cluster,the shorter the running speed of the algorithm.The accuracy of anomaly detection in power generation data reached a maximum of 99.2%,which was 8.165%higher than the random forest algorithm.Moreover,the running time of the kNN algorithm was better than that of the random forest algorithm,with a minimum algorithm running time of 4 seconds,further indicating that the kNN algorithm can effectively detect massive 5G access data in the smart grid.

关 键 词：KNN算法 智能电网 5G 数据异常 检测 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]