摘要
研发了基于机器学习算法的GIS设备运行状态分析模型,根据其PD的严重程度做出GIS设备状态的评估。该模型主要完成了ISODATA算法和模糊KNN(Fuzzy K-Nearest Neighbor)算法开发。其中,ISODATA算法对训练样本数据集进行训练,得到具有若干个聚类的新样本数据集;在此基础之上,FKNN算法对新样本进行设备状态分类。实验结果显示,对设备状态评估多分类问题,研发的模型可以保持95%以上的准确率,而且相较于原始KNN算法能减少90%以上时间开销,具有良好的应用前景。
A GIS equipment operation status analysis model based on machine learning algorithm was developed,and the GIS equipment status was evaluated based on the severity of its PD.This model mainly completed the development of ISODATA algorithm and Fuzzy K-Nearest Neighbor algorithm.Among them,the ISODATA algorithm trains the training sample data set to obtain a new sample data set with several clusters.Based on this,the FKNN algorithm classifies the new sample for device state.The experimental results show that the model developed by this paper can not only maintain the accuracy of more than 95%,but also reduce the time overhead by more than 90%compared with the original KNN algorithm,and has a good application prospect.
作者
方钦
陈建峡
张晓星
金淼
郑建
李秀卫
FANG Qin;CHEN Jianxia;ZHANG Xiaoxing;JIN Miao;ZHENG Jian;LI Xiuwei(School of Computer Science,Hubei Univ.of Tech.,Wuhan 430068,China;School of Electrical Engin.,Wuhan Univ.,Wuhan 430072,China;State Grid Shandong Electric Power Company Electric Power Research Institute,Jinan 250002,China)
出处
《湖北工业大学学报》
2018年第2期62-66,共5页
Journal of Hubei University of Technology
基金
国家高技术研究发展计划(2015AA050204)
关键词
GIS设备
PD
模糊理论
FKNN
gas insulated switchgear
partial discharge
Fuzzy theory
FKNN
