Open AccessSingle Information Extraction Algorithm of Mechanical Equipment Usage Information Recording Based on Deep Learning
Author(s) -
Yibin Luo,
Wenyuan Wu,
Pengcheng Lyu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1865/4/042032
Subject(s) - fault (geology) , computer science , set (abstract data type) , identification (biology) , information extraction , data mining , preprocessor , gas compressor , data set , data pre processing , artificial intelligence , engineering , mechanical engineering , botany , seismology , biology , programming language , geology
The compressor work order record document records the compressor fault information and the corresponding solution. This paper attempts to use natural language processing technology to analyze the compressor work order record document and automatically identify the equipment entity and fault description information. Firstly, the equipment information and fault information are separated from the work order record document, and the equipment entity data set and fault description data set are constructed. Then, based on the BERT preprocessing model, the sequence labeling model is fine-tuned, and the automatic identification models of compressor equipment name and fault description are constructed respectively. The experimental results on the equipment entity data set and fault description data set show that the automatic identification F1 value of the above model for compressor equipment entity and fault description reaches 95.05% and 74.44% respectively, which exceeds 9.71% and 16.85% of the BiLSTM+CRF model commonly used in the industry, which verifies the effectiveness of the method.