Fault Text Classification of Rotating Machine Based BERT
Chen Ling, Liu Yimin, Ji Lianlian · 2021
Accurate classification of equipment fault is one of the effective ways to improve the efficiency of fault analysis and maintenance. How to establish an effective equipment fault classification model has become a hot research topic. Therefore, a fault text classification method for rotating machine based on BERT (Bidirectional encoder representation from transformers) is proposed, which uses fine-tune on the BERT pre-training language model to make it suitable for downstream tasks and takes bidirectional transformers for feature extraction to get understanding of text semantics. The experimental results show that the$F1$of the model on the test set is up to 97.4%, which is 18.4% higher than the$F1$of the THUCTC (THU Chinese Text Classification) classification tool on average; the model based on BERT not only has a higher accuracy rate, but also has a certain reference significance for similar downstream tasks of natural language processing.