Research on Question Answering for Knowledge graph of Aircraft PHM Fault

A question recognition method based on BERT-BiLSTM-ATT-CRF is proposed to solve the problem of entity recognition difficulties faced by Question answering in the field of aircraft PHM. First, the BERT model is used for preprocessing, and then the entity information in the question is further processed by the BiLSTM-ATT-CRF. Then, the Bayesian classifier is used to obtain the question classification. Finally, the Knowledge graph is used to realize intelligent question answering in the field of aircraft PHM. After experimental comparison, the proposed BERT-BiLSTM-ATT-CRF model has better recognition ability in the field of aircraft PHM, with accuracy, recall, and F1 values of 72.236%, 73.563%, and 72.8995% on the aircraft PHM corpus, respectively. The implementation of intelligent Q&A in the field of aircraft PHM is of great significance for the intelligent development of aircraft PHM.