电气工程学报 ›› 2019, Vol. 14 ›› Issue (1): 83-88.doi: 10.11985/2019.01.015

• • 上一篇    下一篇

基于机器学习的换流站故障分析报告智能分类方法

张彦龙,翟登辉,许丹,张子彪   

  1. 许继电气股份有限公司 许昌 461000
  • 收稿日期:2018-09-19 出版日期:2019-03-25 发布日期:2019-11-01
  • 作者简介:张彦龙,男,1993年生,工程师,主要从事智能电网、电力人工智能应用方面的研究|翟登辉,男,1984年生,高级工程师,主要从事电力人工智能、分布式电源与微电网方面的研究。E-mail:1746395882@qq.com

Intelligent Classify Methods Based on Machine Learning for Convertor Station Failure Analysis Report

ZHANG Yanlong,ZHAI Denghui,XU Dan,ZHANG Zibiao   

  1. XJ Electric Co.,Ltd.,Xuchang 461000 China
  • Received:2018-09-19 Online:2019-03-25 Published:2019-11-01

摘要:

针对换流站故障分析报告大量堆积得不到充分利用的情况,结合机器学习算法对故障分析报告进行智能分类。首先对故障分析报告进行文本分词,并针对分词结果进行建模、聚类分析。第一种方法是利用朴素贝叶斯理论构建模型,提取故障类别与特征词对应关系,当新的故障分析报告产生时,通过贝叶斯计算得到其所属故障类别;第二种方法是利用K-means聚类,根据分词结果将故障分析报告聚成故障簇,新的故障分析报告产生时,根据该故障报告与已有故障簇的相似度对故障分析报告分类。

关键词: 故障分析报告, 贝叶斯理论, 文本分词, 聚类, 相似度

Abstract:

The convertor station failure analysis reports accumulate massively and can’t be fully utilized, for that the convertor station failure analysis reports are classified intelligently based on machine learning. Firstly text segmentation is done for failure analysis report. According to segmentation result, Na?ve bayes theory is used to build model and the relationship between the failure and key words is extralted. Another method is that using cluster analysis and similarity analysis to classify failure analysis report according to segmentation result.

Key words: Failure analysis report, Bayes theory, text segmentation, cluster, similarity

中图分类号: