A processing method and system for cross-working-condition bearing fault classification diagnosis

CN118518356BActive Publication Date: 2026-05-29HEFEI UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
HEFEI UNIV OF TECH
Filing Date
2024-05-24
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing bearing fault classification and diagnosis methods are limited by the receptive field and network depth, making it difficult to simultaneously capture the local details and global patterns of bearing faults, resulting in low accuracy in fault classification and diagnosis of bearings operating under different conditions.

Method used

A hybrid vision model is used to preprocess the bearing signal set to generate a preprocessed time-frequency dataset. Then, through feature extraction, feature enhancement, flattening, and multi-level class-level domain alignment, a target domain aligned feature dataset is generated, which is finally used for fault classification.

Benefits of technology

It improves the accuracy of bearing fault classification and diagnosis across operating conditions, effectively constructs the dependency relationship between local details and global modes, and improves the accuracy of diagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118518356B_ABST
    Figure CN118518356B_ABST
Patent Text Reader

Abstract

The application provides a processing method and system for cross-condition bearing fault classification diagnosis, comprising: acquiring a bearing signal set, the bearing signal set comprising a target domain signal set and a source domain signal set corresponding to the label of the target domain signal set; preprocessing the bearing signal set to generate a preprocessed time-frequency data set; inputting the preprocessed time-frequency data set into a mixed vision model for feature extraction processing to generate an attention feature data set; inputting the attention feature data set into the mixed vision model for feature enhancement processing to generate a target reinforced feature data set; inputting the target reinforced feature data set into the mixed vision model for flattening processing to generate a flattened feature data set; inputting the flattened feature data set into the mixed vision model for multi-layer class-level domain alignment processing to generate a target domain alignment feature data set, and performing fault classification processing on the target domain alignment feature data set, thereby improving the fault diagnosis classification precision of the cross-condition bearing.
Need to check novelty before this filing date? Find Prior Art