A Novel Method for Fault Diagnosis of the Two-Input Two-Output Nonlinear Mass-Spring-Damper System Based on NOFRF and MBPCA

Table 3

The training time.

Kernel function	Training time
	Traditional LSSVM model (s)	LSSVM simplified model (s)
Linear kernel	0.63	0.56
Polynomial kernel	0.67	0.62
GRB kernel	0.78	0.66
ERB kernel	0.90	0.69
MLP kernel	1.28	0.98