Fault Text Classification Based on Convolutional Neural Network

The fault text records various fault information of the power system operation, and it is an important data source for analyzing the power system operation. The text management of power faults is becoming more and more intelligent, and the task of classification of fault texts has gradually changed from manual operation to automatic classification of the system. In order to realize automatic classification and improve the classification efficiency and accuracy of power fault texts, in view of the characteristics of power fault short texts, this paper proposes a Convolutional Neural Networks (CNN) short text based on a mixture of word vectors and character vectors. Classification model, which inputs the processed data set information to this classification model to classify short texts of power failures. The experimental results show that the accuracy rate of the proposed model on the power fault classification dataset can reach 88.35%. Compared with other classification models, the feature extraction ability is stronger and the classification effect is better.