A TextCNN Based Approach for Multi-label Text Classification of Power Fault Data
Qiang Zhang, Rongrong Zheng, Ziyan Zhao, Bo Chai, Jiangui Li · 2020
With the increasing number of power terminal devices in power systems, fault diagnosis based on textual fault data has become a challenging task. Classification technology plays an important role in power system fault diagnosis. In this paper, we propose a TextCNN based approach to multi-label text classification for analyzing fault text data of power terminal devices. First, the main category labels of faults and failure were constructed. The dataset was preprocessed for feature representation. Next, we adopted the TextCNN based model for multi-label classification of power fault texts. The experimental results show that the proposed method is better than the typical multi-label classification models for power fault text multi-label classification. The constructed label categories were also experimentally evaluated, and these labels reflect the main types of power faults and failure.