Skip to main content

Table 5 Comparison with the classification results from the literature on the bearing dataset

From: Connected Components-based Colour Image Representations of Vibrations for a Two-stage Fault Diagnosis of Roller Bearings Using Convolutional Neural Networks

Reference No. Method Testing data size Classification accuracy (%)
[85] Raw Vibration data with entropic features. 10-fold cross-validation 98.9 ± 1.2
Compressed measurements of 0.5 sampling rate followed by signals recovery. 92.4 ± 0.5
Compressed sampled 0.25 sampling rate followed by signals recovery. 84.6 ± 3.4
[86] CS   98.6 ± 0.3
CS-PCA 40% 98.5 ± 0.4
CS-LDA   89.8 ± 3.5
[87] FMM-RF (SampEn + PS) 5-fold cross-validation 99.81 ± 0.41
[88] SVM 10-fold cross-validation 93.5 ± 0.50
MLP 93.7 ± 0.21
[89] GP generated feature sets (un-normalised) 50%  
ANN 96.5
SVM 97.1
This paper RGBVI-CNN (Our proposed method) 50% 99.8% ± 0.1
40% 99.9% ± 0.1
30% 100% ± 0.0