Fusion of Quadratic Time-Frequency Analysis and Convolutional Neural Networks to Diagnose Bearing Faults Under Time-Varying Speeds
Mohammad Al-Sa'd, Tuomas Jalonen, Serkan Kiranyaz, Moncef Gabbouj
TL;DR
The paper addresses bearing fault diagnosis under time-varying rotational speeds, a non-stationary regime where traditional DL methods struggle. It introduces a time-frequency CNN (TF-CNN) that fuses quadratic time-frequency representations, notably the compact kernel distribution (CKD), with deep learning to extract joint time-frequency fault signatures. The signal model describes $x(t) = s(t) + \alpha \eta(t)$ with $s(t) = h(t) + f(t)$ and a time-varying fault frequency coupled to motor speed $f_r(t)$; CKD uses a Doppler-lag kernel $g(\nu,\tau)$ controlled by parameters $c$, $D$, $E$ to suppress cross-terms. On KAIST data, the TF-CNN achieves near-ideal performance on clean data and substantial robustness under noise, with accuracy gains up to about 15 percentage points over recent baselines, corroborated by t-SNE and Grad-CAM analyses. The work demonstrates a practical path for robust bearing health monitoring in real-world, non-stationary environments and suggests future enhancements such as attention mechanisms and post-processing.
Abstract
Diagnosis of bearing faults is paramount to reducing maintenance costs and operational breakdowns. Bearing faults are primary contributors to machine vibrations, and analyzing their signal morphology offers insights into their health status. Unfortunately, existing approaches are optimized for controlled environments, neglecting realistic conditions such as time-varying rotational speeds and the vibration's non-stationary nature. This paper presents a fusion of time-frequency analysis and deep learning techniques to diagnose bearing faults under time-varying speeds and varying noise levels. First, we formulate the bearing fault-induced vibrations and discuss the link between their non-stationarity and the bearing's inherent and operational parameters. We also elucidate quadratic time-frequency distributions and validate their effectiveness in resolving distinctive dynamic patterns associated with different bearing faults. Based on this, we design a time-frequency convolutional neural network (TF-CNN) to diagnose various faults in rolling-element bearings. Our experimental findings undeniably demonstrate the superior performance of TF-CNN in comparison to recently developed techniques. They also assert its versatility in capturing fault-relevant non-stationary features that couple with speed changes and show its exceptional resilience to noise, consistently surpassing competing methods across various signal-to-noise ratios and performance metrics. Altogether, the TF-CNN achieves substantial accuracy improvements up to 15%, in severe noise conditions.