Table 6.

Comparison between the proposed technique and other techniques presented in the literature.

Work (years)	Applied Method/Approach	Operating condition	Validation	Load %	Level defect
2015 [17]	Empirical Mode Decomposition (EMD) and MCSA	Steady-state	Experimental	50 and 75	Incipient BRB (Half BRB)
2016 [18]	MUSIC and MCSA	Steady-state	Experimental	Light load	Incipient BRB (Half BRB)
2017 [19]	HT and estimation of statistical parameters	Startup-transient	Experimental	0 and 100	Incipient BRB (Half BRB)
2017 [20]	Homogeneity	Startup-transient	Simulation	Quarter load	Incipient BRB (Half BRB)
2017 [21]	-MUSIC for frequency location -FFT for severity estimation	Steady-state	Experimental	50 and 100	Incipient BRB (9 mm)
2018 [22]	MUSIC and MCSA	Steady-state	Experimental	5,10, 20, 30, 50 and 75	Incipient BRB (Half BRB)
2018 [24]	random forest (RF) classifier	Steady-state	Experimental	Medium and high	Incipient BRB (4.2 mm and 9.4 mm)
2019 [44]	-MCSA is used for fault extraction. -Neural Networks (NN) is used for classification	Steady-state	Experimental	50 and 75	Incipient BRB (5 mm and 10 mm)
2020 [45]	-Short Time Fourier Transform (STFT) is employed to transform the measured signals to images. -A Convolutional Neural Networks (CNN) is used as features classifier	Startup-transient	Experimental	_	Incipient BRB (5 mm)
2021 [46]	-The combined ETSA/MCSA method is used for fault extraction -FLA is used for classification	Steady-state	Experimental	0, 50 and 100	Three levels of Incipient BRB
Proposed work	-The combined DWT/ETSA method is used for fault extraction -FLA is used for classification	Steady-state	Experimental	0, 50 and 100	Incipient BRB fault (2 mm)