Evaluating deep learning models for fault diagnosis of a rotating machinery with epistemic and aleatoric uncertainty

2024-12-25

AI Summary

This study addresses uncertainty-aware deep learning for rotating machinery fault diagnosis, systematically evaluating models’ joint modeling capability for epistemic uncertainty (unknown faults) and aleatoric uncertainty (noise-induced perturbations). We propose a dual-threshold uncertainty discrimination mechanism and a novel adaptive thresholding strategy. For the first time, we comprehensively benchmark mainstream approaches—including Monte Carlo Dropout, Bayesian neural networks, and deep ensembles—under controlled uncertainty conditions. Experimental results demonstrate that deep ensembles achieve the highest out-of-distribution (OOD) detection rate, fastest inference speed, and strongest robustness against noise-induced performance degradation across diverse uncertainty scenarios. Furthermore, we uncover a nonlinear relationship between noise intensity and uncertainty estimation accuracy. The work establishes an interpretable, highly reliable uncertainty quantification framework for robust fault diagnosis, advancing both methodological rigor and practical deployability in industrial predictive maintenance.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Uncertainty-aware deep learning (DL) models recently gained attention in fault diagnosis as a way to promote the reliable detection of faults when out-of-distribution (OOD) data arise from unseen faults (epistemic uncertainty) or the presence of noise (aleatoric uncertainty). In this paper, we present the first comprehensive comparative study of state-of-the-art uncertainty-aware DL architectures for fault diagnosis in rotating machinery, where different scenarios affected by epistemic uncertainty and different types of aleatoric uncertainty are investigated. The selected architectures include sampling by dropout, Bayesian neural networks, and deep ensembles. Moreover, to distinguish between in-distribution and OOD data in the different scenarios two uncertainty thresholds, one of which is introduced in this paper, are alternatively applied. Our empirical findings offer guidance to practitioners and researchers who have to deploy real-world uncertainty-aware fault diagnosis systems. In particular, they reveal that, in the presence of epistemic uncertainty, all DL models are capable of effectively detecting, on average, a substantial portion of OOD data across all the scenarios. However, deep ensemble models show superior performance, independently of the uncertainty threshold used for discrimination. In the presence of aleatoric uncertainty, the noise level plays an important role. Specifically, low noise levels hinder the models' ability to effectively detect OOD data. Even in this case, however, deep ensemble models exhibit a milder degradation in performance, dominating the others. These achievements, combined with their shorter inference time, make deep ensemble architectures the preferred choice.

Problem

• Deep Learning
• Uncertainty Quantification
• Rotating Machine Fault Diagnosis

Innovation

• Uncertainty-aware Deep Learning
• Rotary Machine Fault Diagnosis
• Deep Ensembles