基于文本分类的输变电设备缺陷分析

为提高电力设备缺陷文本利用效率，构建缺陷文本分类模型。首先分析了中文文本分类的流程，然后结合电力设备缺陷文本的特点，构建了4层卷积神经网络的电力缺陷文本分类模型；最后以某市近10年来的电力设备缺陷文本记录作为数据来源，结合上述的模型，对数据进行训练和分类，并与部分传统的机器学习分类模型进行比较。结果表明，提出的分类模型的错误率为2.86%，远低于传统的6.99%，具有明显的优势。

Defects Analysis of Transmission and Distribution Equipment Based on Text Classification

To improve the efficiency of defects text utilization of power equipment, a defects text classification model is constructed. Firstly, the process of Chinese text categorization is analyzed, and then a four-layer convolutional neural network model for text categorization of power equipment defects is constructed based on the characteristics of the defects text of power equipment. Finally, the data are trained and categorized by using the defects text records of power equipment in a city in recent 10 years as the data source. The model is compared with some traditional machine-learning classification models. The result shows that the error rate of the proposed classification model is 2.86%, which is much lower than the traditional 6.99%, and thus it has obvious advantages.