Fatigue criterion and data-driven integrated method for gear fatigue pitting life prediction

By integrating fatigue criteria and data-driven methods, and using literature data and GAN model-generated data, combined with GCN model, the problem of insufficient data in gear fatigue pitting life prediction was solved, and high-precision prediction results were achieved.

WO2026138053A1PCT designated stage Publication Date: 2026-07-02CHONGQING UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
CHONGQING UNIV
Filing Date
2025-09-29
Publication Date
2026-07-02

AI Technical Summary

Technical Problem

Existing technologies lack sufficient high-quality data for predicting the fatigue pitting life of gears, resulting in insufficient accuracy of machine learning methods, and the acquisition of data is costly and time-consuming.

Method used

An integrated fatigue criterion and data-driven approach was adopted. The original dataset was constructed by literature meta-analysis, and data was generated by combining multi-axis fatigue criteria and adversarial network GAN model to expand the training set. Finally, the graph convolutional network GCN model was used for prediction to establish an accurate gear fatigue pitting life prediction model with a small sample set.

Benefits of technology

In situations where data is scarce, high-precision prediction of gear fatigue pitting life is achieved, supporting failure analysis of gear transmission systems and reducing data acquisition costs and time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2025125143_02072026_PF_FP_ABST
    Figure CN2025125143_02072026_PF_FP_ABST
Patent Text Reader

Abstract

Provided in the present invention is a fatigue criterion and data-driven integrated method for gear fatigue pitting life prediction. The method comprises: collecting gear contact fatigue data of different materials from published literature, so as to construct an original data set; on the basis of gear parameters and engagement conditions in the literature, using ABAQUS simulation software to establish a two-dimensional finite element model of a local spur gear; using precise stress and strain results obtained by means of simulation to perform life calculation and verification in combination with an FS method; calculating pitting life within an experimental contact stress range by means of changing loads and using the FS method, so as to expand the original data set; applying a generative adversarial network model to further expand the data set; and finally, on the basis of the recombined data set, establishing a graph convolutional network model to perform fatigue pitting life prediction. The present method exhibits higher accuracy in prediction results. The implementation of the method provides an effective solution for small-sample failure analysis of a gear transmission system, thereby demonstrating engineering application value.
Need to check novelty before this filing date? Find Prior Art