Defect Analysis of Power Equipment Based on Information Extraction Model

Inhalt:
In operation and maintenance, State Grid has accumulated a large amount of historical equipment defect text data. For problems such as difficulty in text data mining. This paper established a text information extraction model at first. Extract important information such as equipment defect phenomena and defect locations from historical defect data. Based on it, establishing convolutional neural network (CNN) to achieve defect level classification. By comparing classification effect before and after information extraction, results indicate that text information extraction model can improve accuracy of defect level classification model.