Physical enhancement federal continuous fault diagnosis method for complex electromechanical system
By employing a physical-enhanced federated continuous fault diagnosis method, the generator synthesizes vibration signals that match the current operating conditions. Combined with physical similarity filtering of historical generators, this method solves the problems of data privacy, catastrophic forgetting, and lack of physical information in fault diagnosis of complex electromechanical systems, achieving fault diagnosis with high accuracy and low forgetting rate.
Patent Information
- Application Number
- CN202511943959.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-22
- Publication Date
- 2026-04-07
AI Technical Summary
Existing fault diagnosis methods for complex electromechanical systems suffer from problems such as the conflict between data privacy and distributed training, catastrophic forgetting caused by dynamic changes in operating conditions, and poor generalization due to the lack of physical information.
A physical-enhanced federated continuous fault diagnosis method is adopted. The generator synthesizes vibration signals that match the current working conditions, and combines physical similarity to screen historical generators. Knowledge distillation and optimization are carried out to achieve client iteration and server aggregation, ensuring the model's continuous diagnostic capability in dynamic task scenarios.
While protecting data privacy, it achieves high accuracy and low forgetting rate in fault diagnosis across operating conditions and incremental category scenarios, resolves the contradiction between privacy protection and distributed training, and enhances the model's robustness to changes in operating conditions.