A Deep Semantic Based Text Mining Technique for Power Equipment Defects for Power Equipment Condition Evaluation Technique
Junfei Jiang, Wenxing Sun, Yingting Luo, Jianming Liu, E Shenglong, Yongchao Liang · 2024
A large amount of text data related to equipment health, such as defect records and inspection methods, has been accumulated during the operation and maintenance of power equipment. These unstructured data reflect the equipment health status and potential fault information, and it has become an inevitable trend to combine equipment management with big data to mine information related to the health status of power equipment from defective text, so as to carry out a comprehensive power equipment condition assessment. However, it is difficult to apply directly to the calculation due to the descriptive text belonging to unstructured data as well as the problems of text ambiguity and record irregularity. And the power equipment defect records involve professional knowledge in the field of electric power, the text is difficult to split the word, the composition method of the word and the semantic information has the diversity, but also increases the difficulty of mining power text. Based on this, this paper explores how to extract effective health status information from the text of power equipment defects through in-depth study of natural language processing technology, and then constructs the power equipment health status evaluation model. Combining deep learning techniques such as recurrent neural network (RNN) and convolutional neural network (CNN), the article proposes a classification model for power equipment defective text, and further applies it to the automatic evaluation of power equipment health status. By automatically analysing and processing the historical defective texts in the power system, the manual workload can be greatly reduced and the efficiency and accuracy of equipment health management can be improved.