Table 1.

The settings of the network structure.

Floor number	Name	Output dimension	Other settings
0	Input layer	(batch, 30, 6)	Time step: 30
1	LSTM layer	(batch, 30, 32)	Recurrent dropout = 0.25 Dropout = 0.2
2	Reshape layer	(batch, 30, 32, 1)
3	BN layer	(batch, 30, 32, 1)
4	Convolution layer	(batch, 30, 32, 32)	Convolution kernel size = 3 ∗ 3
5	Maximum pool layer	(batch, 15, 16, 32)
6	Lost layer	(batch, 7680)
7	Full connection layer	(batch, 2)
8	Output layer	(batch, 30, 32, 1)