YOLO-based Bearing Fault Diagnosis With Continuous Wavelet Transform

This letter presents a locality-aware bearing fault diagnosis framework that operates on time-frequency representations and enables spatially interpretable decision-making. One-dimensional vibration signals are first mapped to two-dimensional time-frequency spectrograms using the continuous wavelet transform (CWT) with Morlet wavelets to enhance transient fault signatures. The diagnosis task is then formulated as object detection on the time-frequency plane, where YOLOv9, YOLOv10, and YOLOv11 are employed to localize fault-relevant regions and classify fault types simultaneously. Experiments on three public benchmarks, including Case Western Reserve University (CWRU), Paderborn University (PU), and Intelligent Maintenance System (IMS), demonstrate strong cross-dataset generalization compared with a representative MCNN-LSTM baseline. In particular, YOLOv11 achieves mAP@0.5 of 99.0% (CWRU), 97.8% (PU), and 99.5% (IMS), while providing region-aware visualization of fault patterns in the time-frequency domain. These results suggest that detection-based inference on CWT spectrograms provides an effective and interpretable complementary approach to conventional global classification for rotating machinery condition monitoring.

💡 Research Summary

This paper introduces a novel bearing fault diagnosis framework that leverages continuous wavelet transform (CWT) to convert one‑dimensional vibration signals into two‑dimensional time‑frequency spectrograms, and then applies state‑of‑the‑art YOLO object detection models (YOLOv9, YOLOv10, YOLOv11) to simultaneously localize fault‑relevant regions and classify fault types. The authors argue that traditional deep‑learning approaches for rotating‑machinery condition monitoring typically perform global classification on whole signal segments, which obscures the spatial origin of fault signatures and limits interpretability. By treating the diagnosis task as an object detection problem on the CWT‑generated spectrograms, the proposed method provides region‑aware visual explanations and can be more robust to variations in operating conditions because decisions are based on localized evidence.

The methodological pipeline consists of three stages. First, raw vibration data from three public bearing datasets—Case Western Reserve University (CWRU), Paderborn University (PU), and Intelligent Maintenance System (IMS)—are segmented into 2048‑sample windows with 50 % overlap. Each segment is transformed using a real‑valued Morlet wavelet, which offers balanced time‑frequency localization and resilience to high‑frequency noise. The resulting scalograms are logarithmically compressed, normalized to