Recognition Algorithm of AE Signal of Rock Fracture Based on Multiscale 1DCNN-BLSTM

Table 1

Multiscale 1DCNN-BLSTM network model parameter.

Layer	Type	Output size	Convolution kernel size/step size	Number of neurons	Convolution kernel number	Dropout	Activation function	Pooling layer
1	Conv1	16  16	32  1/1	—	16	0.6	SELU	2
2	Conv2	64  64	128  1/1	—	64	0.6	SELU	2
3	Conv3	128  128	256  1/1		128	0.6	SELU	2
4	ECA	—	—	—	—	—	—	—
5	MCNN	16  64	2  1/1	—	64	0.6	SELU	1
6	BLSTM	40  1	—	(40, 40)	—	—	Tanh	—
7	Dense1	100  1	—	100	—	0.8	Softmax	—
8	GAP	100  1	100		1	0.8	Softmax	—