Out-of-distribution fault detection method and system based on energy propagation and graph learning

• Main Authors: LI MENG, LI XINMING, YAO JIACHI, GAO ZHIKANG, WANG YANXUE, LI SHANSHAN
• Format: Patent
• Language: Chinese; English
• Published: 19.11.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to the technical field of building machinery external distribution fault intelligent detection, and discloses an external distribution fault detection method and system based on energy propagation and graph learning, and the method comprises the steps: collecting a vibration acceleration signal in a typical fault state, carrying out the similarity calculation, obtaining an adjacent matrix composed of maximum mutual information coefficients, and carrying out the calculation of the similarity; and taking as input in the graph neural network. And performing feature extraction on the adjacent matrix by adopting a GraphSage graph convolution method to generate each node representation. And calculating an energy score of each node, and distinguishing data inside and outside the distribution. Enhancing the confidence estimation of the out-of-distribution data of each node, and carrying out the recognition and detection of the out-of-distribution data of the rolling bearing under different worki