Comparative Study on Deep Learning Models for Bearing Fault Diagnosis

Fault diagnosing of bearings generally depends upon vibrating signals and a group of features will be derived for classifying the faults. A particular feature portrays only a single characteristic of vibration signals. As a result, a lot of works merge more than a single feature to enhance the performances of fault diagnosing. This work mostly aims on proposing a new Bearing Fault Diagnosis model, where, “Empirical Wavelet Transform (EWT), Empirical Mode Decomposition (EMD) and Wavelet Transform (WT) based features” are extracted at first. Further, these extracted features are classified via Quantum Neural Network (QNN), where the weights are tuned optimally by Whale with Modified Update Evaluation (WMUE). At the end, the supreme performance of the adopted model is proven by comparing different deep learning models.