Intermediary bearing fault diagnosis method based on digital twinning and multi-modal incremental learning

By generating multi-fidelity simulation data through digital twin models and employing multimodal incremental learning methods, the problems of data scarcity and complex operating conditions in intermediate bearing fault diagnosis have been solved, achieving efficient and accurate fault diagnosis and improving the operational reliability of aero-engines.

CN122242151APending Publication Date: 2026-06-19NANJING UNIV OF INFORMATION SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANJING UNIV OF INFORMATION SCI & TECH
Filing Date
2026-03-26
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing fault diagnosis technologies for intermediate bearings in aero-engines face challenges such as data scarcity, limited modal information, insufficient generalization, and incremental learning deficiencies. These limitations result in insufficient diagnostic accuracy and robustness under complex operating conditions, failing to meet the reliability requirements of engineering applications.

Method used

By constructing a digital twin model to generate multi-fidelity simulation data, and combining multimodal incremental learning methods, including meta-learning, cross-layer transfer and hybrid fine-tuning, a global diagnostic model is constructed. The model is then optimized using scaling strategies and memory constraints to achieve fault feature extraction and real-time diagnosis.

Benefits of technology

It significantly improves the data acquisition efficiency and diagnostic accuracy of intermediate bearing fault diagnosis, enhances the model's generalization ability and robustness under varying speed and load conditions, and meets the real-time diagnostic needs of aero-engines.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122242151A_ABST
    Figure CN122242151A_ABST
Patent Text Reader

Abstract

This invention discloses a fault diagnosis method for intermediate bearings based on digital twins and multimodal incremental learning, belonging to the field of aero-engine fault diagnosis technology. The method includes: constructing a digital twin model and generating vibration simulation data from it; pre-training an initial diagnostic model using the vibration simulation data to obtain a first diagnostic model; constructing incremental learning samples to improve the structure of the first diagnostic model and training the improved model; optimizing a second diagnostic model using a combination of scaling strategies and memory constraints; and constructing a model integration mechanism on the server side to obtain a final global diagnostic model by weighted fusion of parameters from a third diagnostic model across multiple rounds of tasks, thereby achieving real-time fault diagnosis. This invention effectively solves the diagnostic difficulties caused by the scarcity of real-world fault data and the complex and variable operating conditions of intermediate bearings, significantly improving the model's generalization ability and robustness in incremental scenarios.
Need to check novelty before this filing date? Find Prior Art