基于知识图谱嵌入和补全的电梯故障预测

Elevator failure prediction based on embedding and completion of knowledge graph

为及早预测电梯发生的常见故障,提高电梯设备的维保质量和效率,提出基于规则推理、知识图谱嵌入技术和知识图谱补全技术实现电梯故障预测的方法,在构建电梯故障知识图谱后,通过改进的组合模型将三元组中的实体和关系训练为连续的低维向量空间,实现三元组对于故障预测相关运算的兼容,通过组合模型实现电梯实体、关系和故障实体三元组的预测。电梯故障预测实验经知识图谱的构建、TransE模型向量化,以及ProjE模型训练及测试,验证了提出的电梯故障预测方法的有效性。

To predict the common failures of elevators early and improve the maintenance quality and efficiency,a method based on rule-based reasoning,knowledge graph embedding technology and knowledge graph complement technology was proposed.After constructing the knowledge graph of elevator failure,the entities and relationships in the triples were trained into a conti-nuous low-dimensional vector space through the improved combination model,which realized the compatibility of the triples with the related operations of fault prediction.The combination model realized the prediction of the triplet of elevator entity,relationship and fault entity.The elevator failure prediction experiment verifies the effectiveness of the proposed elevator failure prediction method through the construction of the knowledge graph,the vectorization of the TransE model,and the training and testing of the ProjE model.