摘要
在考虑托辊的运行机理以及带式输送机各结构之间复杂联锁关系的情况下,提出一种结合知识图谱(Knowledge Graph,KG)和贝叶斯网络(Bayesian Neural Network,BNN)的托辊故障诊断方法。首先,结合已知结构化数据和专家知识利用七步法构建托辊故障本体;其次,利用Bert-BiLSTMCRF模型从数据集中抽取故障知识;然后,结合本体模型中预定义的实体关系和结构化故障知识构成知识三元组形式,并利用Noe4j图数据库实现知识存储及可视化;随后,为解决贝叶斯网络推理过程中故障数据样本量过少且计算量大的问题,在模型中引入了noisy-OR方法;最后,通过案例应用验证该研究方法的可行性,为故障维修人员提供了便利。
Considering the operation mechanism of rollers and the complex interlocking relationship between the structures of belt conveyors,a fault diagnosis method for rollers based on knowledge graph and Bayesian neural network was proposed.Firstly,the seven-step method was used to construct the roller fault ontology by combining the known structured data and the expert knowledge.Secondly,the Bert-BiLSTM-CRF model was used to extract the fault knowledge from the dataset.Then,the predefined entity relationship and the structured fault knowledge in the ontology model were combined to form a knowledge triplet,and the Noe4j graph database was used to realize knowledge storage and visualization.In order to solve the problem that the fault data sample size was too small and the calculation amount was large in the Bayesian network inference process,the noisy-OR method was introduced into the model.Finally,the feasibility of the research method was verified by the case application,which provided convenience for the fault maintenance personnel.
作者
吴胜
WU Sheng(Guizhou Panjiang Clean Coal Co.,Ltd.,Liupanshui 553539,Guizhou,China)
出处
《矿山机械》
2025年第6期21-28,共8页
Mining & Processing Equipment
关键词
托辊
故障诊断
知识图谱
贝叶斯推理
roller
fault diagnosis
knowledge graph
Bayesian inference
作者简介
吴胜,男,1970年生,高级工程师,主要从事煤矿选煤生产管理工作。
