Real-time fault diagnosis method based on adaptive Bi-LSTM

By combining the adaptive Bi-LSTM method with time-frequency domain feature extraction and feature dimensionality reduction, the problem of feature redundancy and modal aliasing in gearbox fault diagnosis is solved, and efficient and accurate identification of gearbox compound faults is achieved.

CN120744467AActive Publication Date: 2025-10-03NANCHANG INST OF TECH
5 Cites 0 Cited by

Patent Information

Application Number
CN202511245511.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-09-02
Publication Date
2025-10-03
Estimated Expiration
2045-09-02

AI Technical Summary

Technical Problem

Existing gearbox fault diagnosis methods are difficult to achieve robust feature extraction under complex coupling conditions, and there are problems of feature redundancy and modal aliasing, which affect the diagnostic accuracy.

Method used

A real-time fault diagnosis method based on adaptive Bi-LSTM is adopted, which combines time domain and frequency domain feature extraction, uses elastic network to optimize Lasso regression for feature dimensionality reduction, and constructs a Bi-LSTM neural network improved by sparse attention mechanism for fault diagnosis.

Benefits of technology

It effectively reduces the feature dimension, reduces redundant feature interference, improves the accuracy and real-time performance of fault diagnosis, and realizes the accurate identification of gearbox compound faults.

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

Abstract

The invention records a real-time fault diagnosis method based on adaptive Bi-LSTM, and the method comprises the steps: S1, collecting original fault signals of a gear box, including various fault types, and setting a certain sample size for each fault type to form a data set; key features of the original fault signals are extracted, and feature vectors are constructed; s2, performing feature dimension reduction on the feature vector based on elastic network optimization Lasso regression; and S3, constructing an adaptive Bi-LSTM neural network improved based on a sparse attention mechanism, and performing fault diagnosis on the feature vectors after feature dimension reduction. According to the method, the fault features are extracted from the two aspects of the time domain and the frequency domain, and the essential information of the fault signals can be fully mined. On the basis of an elastic network improved Lasso method, dimensionality reduction is carried out on high-dimensional features, the data dimensionality can be effectively reduced, and redundant feature interference can be reduced. And based on the adaptive Bi-LSTM neural network, fault diagnosis is carried out on the screened features, and accurate and real-time identification of the composite fault of the gearbox is realized.
Need to check novelty before this filing date? Find Prior Art