面向民航飞机故障安全诊断的知识图谱构建方法  

Construction method of knowledge graph for fault safety diagnosis of civil aviation aircrafts

作　　者：朱江[1] 谢涛[1] ZHU Jiang;XIE Tao(Faculty of Civil Aviation and Aeronautics,Kunming University of Science and Technology,Yunnan Kunming 650500,China)

机构地区：[1]昆明理工大学民航与航空学院,云南昆明650500

出　　处：《中国安全生产科学技术》2025年第3期186-194,共9页Journal of Safety Science and Technology

基　　金：国家自然科学基金项目(62163021);云南省科技厅科技计划项目基础研究专项项目(202301AT070420)。

摘　　要：为更好地管理和利用民航飞机设备故障维修知识,提高飞机故障安全诊断的决策效率,提出融合数据增强和多尺度注意力机制的飞机设备故障知识图谱构建方法。首先,创建基于语义相似性的实体集构建模式,结合余弦相似度计算扩充数据样本。其次,采用多尺度注意力对BERT-BiLSTM-CRF模型进行优化改进,以提升知识抽取时局部和全局信息的关注度。最后,利用Neo4j图数据库搭建飞机设备故障知识图谱,并辅助开发智能问答系统用于决策推荐。研究结果表明:所提方法有效解决模型在小样本数据上的局限性,且故障文本知识抽取性能较基准模型显著提升,实体识别精确率、召回率和F 1分别达到92.59%,94.68%和93.62%,为搭建知识图谱提供可靠信息。研究结果可为实现飞机故障的高效诊断和预防飞机事故风险提供参考。In order to better manage and utilize the knowledge of fault maintenance for civil aviation aircraft equipment,and improve the decision-making efficiency of aircraft fault safety diagnosis,a construction method of knowledge graph for the aircraft equipment fault integrating the data enhancement and multi-scale attention mechanism was proposed.Firstly,a construction mode of entity set based on semantic similarity was created,and the data samples were expanded combining with the cosine similarity calculation.Secondly,the BERT-BiLSTM-CRF model was optimized and improved using the multiscale attention to enhance the attention of local and global information during knowledge extraction.Finally,the Neo4j graph database was utilized to build the knowledge graph of aircraft equipment faults,and an intelligent question answering system was developed for decision-making recommendation.The results show that the proposed method effectively solves the limitation of the model on small sample data,and significantly improves the performance of fault text knowledge extraction compared with the baseline model.The entity recognition precision,recall rate and F1 reach 92.59%,94.68%and 93.62%respectively,which provided reliable information for the construction of the knowledge graph.The research results can provide a reference for the efficient diagnosis of aircraft faults and the prevention of aircraft accident risk.

关 键 词：飞机设备 故障诊断 数据增强 多尺度注意力 知识图谱 智能问答 

分 类 号：X949[环境科学与工程—安全科学]