Rolling bearing fault diagnosis method based on multi-modal knowledge graph

• Main Authors: YUAN XINPAN, PENG CHENG, ZHOU XIAOHONG, SHENG YANYAN, LI CHANGYUN
• Format: Patent
• Language: Chinese; English
• Published: 22.12.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; MEASURING; PHYSICS; TESTING; TESTING STATIC OR DYNAMIC BALANCE OF MACHINES ORSTRUCTURES; TESTING STRUCTURES OR APPARATUS NOT OTHERWISE PROVIDED FOR

The invention discloses a rolling bearing fault diagnosis method based on a multi-modal knowledge graph, and the method comprises the steps: carrying out the sliding window processing of a vibration signal, obtaining a signal entity node, carrying out the preprocessing, and obtaining the multi-modal data of the signal entity node; fusing the multi-modal features by adopting a low-rank multi-modal fusion method to obtain representation of signal entity nodes, extracting a relation and attribute value nodes based on a description text, establishing connection between the nodes according to a semantic rule, and preliminarily constructing a signal-based multi-modal knowledge graph; according to the method, attribute value nodes and relations are embedded and coded by using a relation cascade attention network, then a two-dimensional convolution knowledge graph is embedded as a decoder to obtain a score sequence of candidate triads, link prediction is carried out, and fault diagnosis of the rolling bearing is real