Few-shot Class-incremental Fault Diagnosis by Preserving Class-Agnostic Knowledge with Dual-Granularity Representations

2025-08-16

arXiv.org

AI Summary

Few-shot class-incremental fault diagnosis (FSC-FD) addresses the challenge of continuously emerging fault classes in industrial systems, where only limited samples are available per new class—posing dual challenges of catastrophic forgetting and overfitting to novel classes. To tackle these issues, we propose a dual-granularity representation framework: (1) a dual-granularity guided network that decouples fine-grained discriminative feature learning from coarse-grained class-agnostic universal feature learning, leveraging coarse-grained knowledge to regularize fine-grained training; (2) a boundary-aware sample replay strategy and a decoupled-balanced random forest classifier to enhance model stability and decision fairness; and (3) dynamic feature fusion via multi-order interactive aggregation and multi-semantic cross-attention. Extensive experiments on the Tennessee Eastman Process (TEP) and real-world MFF datasets demonstrate significant improvements over state-of-the-art methods. Our code is publicly available.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Computer Vision: Representation Learning for Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

Few-Shot Class-Incremental Fault Diagnosis (FSC-FD), which aims to continuously learn from new fault classes with only a few samples without forgetting old ones, is critical for real-world industrial systems. However, this challenging task severely amplifies the issues of catastrophic forgetting of old knowledge and overfitting on scarce new data. To address these challenges, this paper proposes a novel framework built upon Dual-Granularity Representations, termed the Dual-Granularity Guidance Network (DGGN). Our DGGN explicitly decouples feature learning into two parallel streams: 1) a fine-grained representation stream, which utilizes a novel Multi-Order Interaction Aggregation module to capture discriminative, class-specific features from the limited new samples. 2) a coarse-grained representation stream, designed to model and preserve general, class-agnostic knowledge shared across all fault types. These two representations are dynamically fused by a multi-semantic cross-attention mechanism, where the stable coarse-grained knowledge guides the learning of fine-grained features, preventing overfitting and alleviating feature conflicts. To further mitigate catastrophic forgetting, we design a Boundary-Aware Exemplar Prioritization strategy. Moreover, a decoupled Balanced Random Forest classifier is employed to counter the decision boundary bias caused by data imbalance. Extensive experiments on the TEP benchmark and a real-world MFF dataset demonstrate that our proposed DGGN achieves superior diagnostic performance and stability compared to state-of-the-art FSC-FD approaches. Our code is publicly available at https://github.com/MentaY/DGGN

Problem

• Addresses catastrophic forgetting of old fault knowledge in incremental learning
• Solves overfitting on scarce new fault data with limited samples
• Mitigates decision boundary bias caused by data imbalance issues

Innovation

• Dual-granularity representations decouple feature learning
• Multi-semantic cross-attention fuses coarse and fine features
• Boundary-aware exemplar prioritization mitigates catastrophic forgetting