Asymmetric Adaptation-based Real-time Fault Diagnosis Under Transitional Operating Conditions

2026-05-23

AI Summary

This work addresses the challenge of significant distribution shifts between static offline models and dynamic online industrial data streams, which often contain unseen transitional operating conditions that degrade fault diagnosis accuracy. To bridge this gap, the authors propose a novel approach integrating offline domain generalization with online test-time adaptation. The method constructs normalized fault prototypes as semantic anchors and introduces a periodic prototype reprojection mechanism during inference to dynamically update prototype positions. Additionally, an asymmetric learning rate strategy and a geometry-aware classifier update scheme based on geometric distribution are designed to enable efficient online adaptation while preserving offline generalization capabilities. Experimental results demonstrate that the proposed framework substantially enhances diagnostic robustness and real-time accuracy under non-stationary operating conditions.

Abstract

Data streams in real-world industrial scenarios often contain transitional operating conditions that are uncovered during offline training, leading to significant distribution shifts. To bridge the gap between static offline models and dynamic online data, a novel asymmetric adaptation-based fault diagnosis method is proposed in this paper. Specifically, in the offline stage, we employ domain generalization techniques to extract domain-invariant features from multiple stable conditions and construct robust normalized fault prototypes as reference anchors. Subsequently, during online inference, we design an online test-time adaptation method based on a periodic prototype re-projection mechanism to dynamically update prototype positions. Furthermore, we utilize the geometric distribution derived from anchors to guide the updates of classifiers and adopt an asymmetric learning rate strategy for the feature extractor and classifier. The proposed approach ensures rapid adaptation to new transitional conditions while preserving the discriminative power inherited from the offline domain generalization initialization. Experimental results demonstrate that this mechanism effectively leverages offline generalized knowledge to guide online inference, significantly improving robustness in non-stationary environments.

Problem

• fault diagnosis
• transitional operating conditions
• distribution shift
• offline-online gap
• non-stationary environments

Innovation

• asymmetric adaptation
• domain generalization
• test-time adaptation
• fault diagnosis
• prototype re-projection