Image-Processing-Based Intelligent Defect Diagnosis of Rolling Element Bearings Using Spectrogram Images

Due to the excellent image recognition characteristics of convolutional neural networks (CNN), they have gained significant attention among researchers for image-processing-based defect diagnosis tasks. The use deep CNN models rolling element bearings’ (REBs’) may be computationally expensive, and therefore not suitable some applications where hardware resources limitations exist. However, instead using as end-to-end classifiers, can also used extract features from images those further input machine learning (ML) In addition extracting models, there are other methods feature extraction vibration characteristic images, such handcrafted histogram oriented gradients (HOG) local binary pattern (LBP) descriptors. These classical ML classification this study, a performance comparison between all these techniques was carried out in terms fault detection accuracy computational expense. Moreover, based upon detailed comparison, hybrid-ensemble method involving decision-level fusion is proposed, which far less expensive compared while them classifiers. case minimal training data availability under slightly different operating conditions ascertain their generalizability ability correctly diagnose despite data. proposed remained outstanding REBs’ well slight variation conditions.