A CBA-KELM-Based Recognition Method for Fault Diagnosis of Wind Turbines with Time-Domain Analysis and Multisensor Data Fusion

Table 2

Comparisons with peer-to-peer classification.

Methods	Data type	Average training accuracy (%)	Average testing accuracy (%)
The proposed CBA-KELM method	Fusion data	100	96.25
	Raw data	53.47	44.22
The proposed BA-KELM method	Fusion data	100	95.66
	Raw data	50.22	42.63
The proposed FA-KELM method	Fusion data	100	95.52
	Raw data	50.05	41.34
The KELM method	Fusion data	90.99	87.90
	Raw data	40.95	37.79
The ELM method	Fusion data	69.06	68.46
	Raw data	39.20	33.99
The BPNN method	Fusion data	77.78	68.97
	Raw data	28.12	26.47