Multidomain Feature Fusion Network for Fault Diagnosis of Rolling Machinery

Table 1

Parameters of MFFN.

Module	Name	Size/step/number	Parameter size	Output size
1D-CNN	Input_1	—	0	(None,4096,1)
	ResBlock_1	3/1/16	912	(None,4096,16)
	ResBlock_2	3/1/16	1568	(None,4096,16)
	ResBlock_3	3/1/4	441	(None,4096,4)
	Max_Pooling	2/1/-	0	(None,2048,4)
2D-CNN	Input_2	—	0	(None,128,128,3)
	Conv2D_1	30/5/256	691200	(None,20,20,256)
	Conv2D_2	6/2/256	2359552	(None,8,8,256)
	Inception_1	(1,3,5,7)/1/32	688768	(None,8,8,128)
	Reshape_1	—	0	(None,2048,4)
SCE	Global_Average_Pooling_1	—	0	(None,4)
	Global_Average_Pooling_2	—	0	(None,4)
	Global_Average_Pooling_3	—	0	(None,4)
	Concatenate_1	—	0	(None,2048,12)
	Concatenate_2	—	0	(None,12)
	Dense_1	6/-/-	78	(None,6)
	Dense_2	12/-/-	84	(None,12)
	Multiply	—	0	(None,2048,12)
Classifier	Flatten_1	—	0	(None,24576)
	Dense_3	100/-/-	2457700	(None,100)
	Dense_4	10,4,3/-/-	1010,404,303	(None,10),(None,4),(None,3)