摘要
针对变电主设备缺陷文本较多、分类性能较差的问题,提出基于NLP的变电主设备缺陷文本自动化挖掘算法。进行变电主设备文本分词,提取变电主设备缺陷文本主要特征,并利用Word2vec模型表示特征信息,调整LSTM的结构以改进文本分类决策,并在NLP文本分类模型中添加注意力机制,通过相似性与归一化计算获取对应的权重系数,增加了内容权重值,有针对性地获得变电主设备缺陷文本的自动化挖掘结果。
Based on the problem of much defect text and poor classification performance of substation main equipment,an automatic mining algorithm of substation main equipment defect text based on NLP is proposed.The text segmentation is carried out in substation main equipment,and the main features of substation main equipment defect text are extracted.The Word2vec modelis used to represent the feature information,the structure of LSTM is adjusted to improve the text classification decision to add attention mechanism to NLP text classification model.Then,the corresponding weight coefficient is obtained through similarity and normalization calculation,and the content weight value is increased.The automatic mining results of defect text of main substation equipment are obtained.
作者
麦晓庆
张天湖
王亮
胡长武
王仙
史渊源
MAI Xiao-qing;ZHANG Tian-hu;WANG Liang;HU Chang-wu;WANG Xian;SHI Yuan-yuan(Zhongwei Power Supply Company of State Grid Ningxia Electric Power Co.,Ltd.,Zhongwei 755000,Ningxia Hui Autonomous Region,China;State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan 750000,China)
出处
《信息技术》
2022年第9期151-156,共6页
Information Technology
关键词
自然语言处理
变电主设备
缺陷
文本挖掘
natural language processing
main equipment of substation
defects
text mining
