Semi-Supervised Co-Training of Time and Time-Frequency Models: Application to Bearing Fault Diagnosis

2025-03-14

AI Summary

To address the scarcity of labeled data in industrial bearing fault diagnosis, this paper proposes a semi-supervised co-training framework that jointly optimizes a lightweight time-domain network and a time-frequency domain model (based on STFT/CWT), enhancing robustness and accuracy without increasing inference overhead. The key contributions are: (i) the first introduction of a dual-view co-training mechanism integrating time-domain and time-frequency domain representations; and (ii) support for cloud-edge collaborative deployment, enabling real-time edge-side diagnosis and continuous cloud-based model refinement. Experimental results demonstrate that, under multi-noise and few-shot conditions, the proposed method achieves diagnostic accuracy improvements of 10.6%–33.9% over self-training baselines, significantly boosting reliability in industrial fault identification.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Computer Vision: Multi-modal Vision

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Neural networks require massive amounts of annotated data to train intelligent solutions. Acquiring many labeled data in industrial applications is often difficult; therefore, semi-supervised approaches are preferred. We propose a new semi-supervised co-training method, which combines time and time-frequency (TF) machine learning models to improve performance and reliability. The developed framework collaboratively co-trains fast time-domain models by utilizing high-performing TF techniques without increasing the inference complexity. Besides, it operates in cloud-edge networks and offers holistic support for many applications covering edge-real-time monitoring and cloud-based updates and corrections. Experimental results on bearing fault diagnosis verify the superiority of our technique compared to a competing self-training method. The results from two case studies show that our method outperforms self-training for different noise levels and amounts of available data with accuracy gains reaching from 10.6% to 33.9%. They demonstrate that fusing time-domain and TF-based models offers opportunities for developing high-performance industrial solutions.

Problem

• Develops semi-supervised co-training for time and time-frequency models
• Improves bearing fault diagnosis with reduced labeled data dependency
• Enhances accuracy in noisy environments and limited data scenarios

Innovation

• Semi-supervised co-training of time and time-frequency models
• Cloud-edge network operation for real-time monitoring
• Fusion of time-domain and TF-based models for accuracy