Leveraging systems' non-linearity to tackle the scarcity of data in the design of Intelligent Fault Diagnosis Systems

2026-06-18

AI Summary

This study addresses the challenge of extremely limited labeled data in mechanical fault diagnosis by proposing a deep transfer learning–based method for vibration signal analysis. The approach leverages periodic multi-excitation to elicit the system’s inherent nonlinear responses, which are transformed into novel visual representations. Combined with a tailored data augmentation strategy, this framework significantly enhances feature separability under small-sample conditions. Integrating convolutional neural networks with transfer learning, the method achieves high-accuracy fault identification on railway pantograph structures using only a minimal number of labeled samples, thereby substantially alleviating the difficulties associated with data-scarce diagnostic scenarios.

Abstract

Deep Transfer Learning (DTL) allows for the efficient building of Intelligent Fault Diagnosis Systems (IFDS). On the other hand, DTL methods still heavily rely on large amounts of labelled data. Obtaining such an amount of data can be challenging when dealing with machines or structures faults. This document proposes a novel approach to the design of vibration-based IFDS using DTL in condition of strong data scarcity. A periodic multi-excitation level procedure leveraging intrinsic non-linearities of real-world systems is used to produce images that can be conveniently analysed by pre-trained Convolutional Neural Networks (CNNs) to diagnose faults. A new data visualization method and its augmentation technique are proposed in this paper to tackle the typical lack of data encountered during the design of IFDS. Experimental validation on a railway pantograph structure provides effective support for the proposed method.

Problem

• Intelligent Fault Diagnosis Systems
• data scarcity
• vibration-based diagnosis
• labelled data
• fault diagnosis

Innovation

• non-linearity exploitation
• data scarcity mitigation
• vibration-based fault diagnosis
• deep transfer learning
• data visualization augmentation