Real-time adaptive learning bearing fault classification method and system

By using prototype alignment and parameter-wise adaptive learning rate, the problem of model drift in adaptive methods during testing is solved, achieving stable adaptation and efficient fault classification in the target domain, which is suitable for dynamic changes in industrial environments.

CN121765477APending Publication Date: 2026-03-31GUANGDONG UNIV OF PETROCHEMICAL TECH +1
0 Cites 0 Cited by

Patent Information

Application Number
CN202610049961.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-15
Publication Date
2026-03-31

AI Technical Summary

Technical Problem

Existing test-time adaptive methods are prone to model drift due to catastrophic forgetting and error accumulation when learning online outside the source domain. This causes the model performance to decline rather than improve during continuous adaptation, which seriously restricts the reliability of its practical application.

Method used

We employ a prototype alignment and parameter-wise adaptive learning rate approach. We generate feature prototypes through a source domain pre-training module, smoothly update teacher parameters using the EMA Teacher module, and combine a target domain adaptive training module and a parameter-wise adaptive learning rate module to achieve continuous and stable adaptation of the model in the target domain, thereby suppressing feature space drift and model forgetting.

Benefits of technology

Without accessing source domain data, the model can achieve continuous and stable adaptation on the target domain, effectively suppressing feature space drift and model forgetting, improving the accuracy and robustness of fault classification, and adapting to dynamic changes in the industrial environment.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention belongs to the technical field of vibration data identification, and discloses a real-time adaptive learning bearing fault classification method and system based on prototype alignment and a parameter-by-parameter adaptive learning rate, and the method comprises the steps: inputting a source domain sample into a source domain pre-training module, obtaining a basic classification model through supervised training, and carrying out the training of a source domain pre-training module; extracting a feature mean value of each category to construct a source domain feature prototype set; in a target domain test stage, generating a relatively stable pseudo tag for a target domain sample by using an EMA Teamer module; then, an online self-adaptive updating process is executed on a target domain sample by using a pseudo tag, and feature alignment is performed in the process by combining symmetric cross entropy loss and prototype-based comparison loss; meanwhile, model parameters are stably updated by adopting a parameter-by-parameter adaptive learning rate strategy; and finally, outputting an accurate prediction label for a target domain sample by the model after adaptive training.
Need to check novelty before this filing date? Find Prior Art