Federated Meta-Learning for Few-Shot Fault Diagnosis with Representation Encoding

TL;DR

This paper introduces REFML, a federated meta-learning framework that enhances fault diagnosis accuracy in scenarios with limited data and domain discrepancies by leveraging representation encoding and adaptive model interpolation.

Contribution

The paper proposes a novel federated meta-learning approach with representation encoding and adaptive interpolation to improve few-shot fault diagnosis across diverse equipment and conditions.

Findings

Achieves 2.17%-6.50% higher accuracy on unseen conditions.

Achieves 13.44%-18.33% higher accuracy on unseen equipment types.

Effectively handles data scarcity and domain discrepancy in federated settings.

Abstract

Deep learning-based fault diagnosis (FD) approaches require a large amount of training data, which are difficult to obtain since they are located across different entities. Federated learning (FL) enables multiple clients to collaboratively train a shared model with data privacy guaranteed. However, the domain discrepancy and data scarcity problems among clients deteriorate the performance of the global FL model. To tackle these issues, we propose a novel framework called representation encoding-based federated meta-learning (REFML) for few-shot FD. First, a novel training strategy based on representation encoding and meta-learning is developed. It harnesses the inherent heterogeneity among training clients, effectively transforming it into an advantage for out-of-distribution generalization on unseen working conditions or equipment types. Additionally, an adaptive interpolation method that calculates the optimal combination of local and global models as the initialization of local training is proposed. This helps to further utilize local information to mitigate the negative effects of domain discrepancy. As a result, high diagnostic accuracy can be achieved on unseen working conditions or equipment types with limited training data. Compared with the state-of-the-art methods, such as FedProx, the proposed REFML framework achieves an increase in accuracy by 2.17%-6.50% when tested on unseen working conditions of the same equipment type and 13.44%-18.33% when tested on totally unseen equipment types, respectively.