汽车工程 ›› 2023, Vol. 45 ›› Issue (1): 52-60.doi: 10.19562/j.chinasae.qcgc.2023.01.006

所属专题: 智能网联汽车技术专题-感知&HMI&测评2023年

• • 上一篇    下一篇

基于深度学习的汽车故障知识图谱构建

胡杰1,2,3(),李源洁1,2,3,耿號1,2,3,耿黄政1,2,3,郭雄4,易红卫4   

  1. 1.武汉理工大学,现代汽车零部件技术湖北省重点实验室,武汉  430070
    2.武汉理工大学,现代零部件技术湖北省协同创新中心,武汉  430070
    3.新能源与智能网联车湖北工程技术研究中心,武汉  430070
    4.上汽通用五菱汽车股份有限公司,柳州  545000
  • 收稿日期:2022-08-08 出版日期:2023-01-25 发布日期:2023-01-18
  • 通讯作者: 胡杰 E-mail:auto_hj@163.com

Construction of Vehicle Fault Knowledge Graph Based on Deep Learning

Jie Hu1,2,3(),Yuanjie Li1,2,3,Hao Geng1,2,3,Huangzheng Geng1,2,3,Xiong Guo4,Hongwei Yi4   

  1. 1.Wuhan University of Technology,Hubei Key Laboratory of Modern Auto Parts Technology,Wuhan  430070
    2.Wuhan University of Technology,Auto Parts Technology Hubei Collaborative Innovation Center,Wuhan  430070
    3.Hubei Technology Research Center of New Energy and Intelligent Connected Vehicle Engineering,Wuhan  430070
    4.SAIC-GM-Wuling Automobile Company Limited,Liuzhou  545000
  • Received:2022-08-08 Online:2023-01-25 Published:2023-01-18
  • Contact: Jie Hu E-mail:auto_hj@163.com

摘要:

本文将知识图谱引入汽车故障诊断领域,以某公司的售后业务数据为来源,根据文本特点,设计了一种知识图谱构建流程:在传统构建流程的基础上,加入文本预分类和实体重组流程。基于DPCNN模型的文本预分类用于处理目标字段存在信息冗余的问题;基于BERT-BiLSTM-MUL-CRF模型的实体抽取与基于语法规则的实体重组的组合形式可以有效解决文本中的嵌套实体问题和非连续实体问题;采用结合术语相似度和结构相似度的方法完成知识融合;最后,选用Neo4j图数据库进行存储,完成汽车故障知识图谱的构建。

关键词: 知识图谱, 深度学习, 实体抽取

Abstract:

This paper introduces knowledge graph into the field of automobile fault diagnosis. Taking the after-sales business data of a company as the source, a knowledge graph construction process is designed according to the characteristics of the text, which adds text pre-classification and entity reorganization process based on the traditional construction process. Text pre-classification based on DPCNN model is used to deal with the problem of information redundancy in target fields. The combination of entity extraction based on BERT-BiLSTM-MUL-CRF model and entity reorganization based on grammar rules can effectively solve the problems of nested entities and discontinuous entity problems in text. The method combining term similarity and structural similarity is used to complete knowledge fusion. Finally, Neo4j graph database is used for storage so as to complete the construction of vehicle fault knowledge graph.

Key words: knowledge graph, deep learning, entity extraction