Open set domain generalization method for bogie fault diagnosis

By constructing an open set-domain generalization (EP-OSDG) model based on ellipsoid prototype, the problem of poor generalization ability when identifying unknown faults under unknown operating conditions is solved, and higher intelligence and practicality are achieved.

CN119714946AActive Publication Date: 2025-03-28GUANGZHOU YIZHI INTELLECTUAL PROPERTY OPERATION CO LTD
12 Cites 0 Cited by

Patent Information

Application Number
CN202411796490.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-09
Publication Date
2025-03-28
Estimated Expiration
2044-12-09

AI Technical Summary

Technical Problem

Traditional deep learning-based fault diagnosis methods show poor generalization capabilities when facing unknown fault types or different operating conditions, and it is difficult to cope with diversified needs in actual production.

Method used

An open-group domain generalization method for bogie fault diagnosis is proposed. The acceleration sensor is used to collect vibration signals, obtain the spectrum through fast Fourier transform, and an open-group domain generalization (EP-OSDG) model based on ellipsoid prototype is constructed, including feature extraction module, domain obfuscation classification module, fault classification module and ellipsoid prototype measurement module. Through adversarial training and Marbanian distance measurement, an asymmetric ellipsoid decision boundary is constructed.

Benefits of technology

It realizes accurate identification of unknown faults under unknown working conditions, effectively improves the intelligence and practicality of the diagnostic method, adapts to the different scales and correlations of various features in the data, and has higher flexibility and applicability.

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

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a bogie fault diagnosis-oriented open set domain generalization method, which comprises the following steps of: acquiring an original vibration signal of an axle box bearing in a bogie, and performing data preprocessing and fast Fourier transform to obtain a frequency spectrum of the vibration signal; dividing the data set into a source domain and a target domain; and constructing an open set domain generalization model based on the ellipsoid prototype, wherein the model comprises a feature extraction module, a domain confusion classification module, a fault classification module and an ellipsoid prototype measurement module. Inputting the source domain data into an open set domain generalization model, and performing iterative training by using a back propagation algorithm to obtain model parameters and a prototype, a covariance matrix and a boundary threshold value of each category; then, determining a fault type of a target domain sample according to a known ellipsoid boundary threshold value; according to the method, the ellipsoid decision boundary is constructed for each category through mahalanobis distance measurement, and a more flexible and universal method is provided for identifying unknown faults under unknown working conditions.
Need to check novelty before this filing date? Find Prior Art