Feasibility Study of the GST-SVD in Extracting the Fault Feature of Rolling Bearing under Variable Conditions

Chinese Journal of Mechanical Engineering

Table 4 Classification result comparison among different classification algorithms

Different feature extraction method	Model	Classification accuracy Training data	Testing data	Running time (s) Training data	Testing data
Standard S transform	PSO-SVM	1	0.9444	7.752	5.606
Standard S transform	BP	0.9854	0.9000	8.351	6.936
LMD and SVD	PSO-SVM	1	0.9861	6.608	4.652
LMD and SVD	BP	0.9896	0.9667	7.021	5.654
Wavelet approximation entropy	PSO-SVM	1	0.9722	6.608	4.652
Wavelet approximation entropy	BP	0.9875	0.9639	8.021	6.654
GST-SVD-PCA	PSO-SVM	1	1	5.687	4.553
GST-SVD-PCA	BP	0.9958	0.9694	6.008	5.583