Smart Aggregate-Based Concrete Stress Monitoring via 1D CNN Deep Learning of Raw Impedance Signals

Table 5

Specifications of the four 1D CNN architectures (M1 − M4).

Layers	Type	Depth	Filter	Stride
Model M1
1	Conv	4	1 × 6	1
2	ReLU	—	—	—
3	Maxpool	—	1 × 2	2
4	Conv	4	1 × 4	1
5	ReLU	—	—	—
6	Maxpool	—	1 × 2	2
7	Conv	8	1 × 5	1
8	ReLU	—	—	—
9	Maxpool	—	1 × 2	2
10	Fc1	48	—	—
11	Fc2	32	—	—
12	Fc3	1	—	—
13	Regression	—	—	—
Model M2 (see Table 1)
Model M3
1	Conv	4	1 × 6	1
2	ReLU	—	—	—
3	Maxpool	—	1 × 2	2
4	Conv	4	1 × 4	1
5	ReLU	—	—	—
6	Maxpool	—	1 × 2	2
7	Conv	8	1 × 5	1
8	ReLU	—	—	—
9	Maxpool	—	1 × 2	2
10	Conv	8	1 × 5	1
11	ReLU	—	—	—
12	Maxpool	—	1 × 2	2
13	Conv	8	1 × 5	1
14	ReLU	—	—	—
15	Maxpool	—	1 × 2	2
16	Fc1	48	—	—
17	Fc2	32	—	—
18	Fc3	1	—	—
19	Regression		—	—
Model M4
1	Conv	4	1 × 6	1
2	ReLU	—	—	—
3	Maxpool	—	1 × 2	2
4	Conv	4	1 × 4	1
5	ReLU	—	—	—
6	Maxpool	—	1 × 2	2
7	Conv	8	1 × 5	1
8	ReLU	—	—	—
9	Maxpool	—	1 × 2	2
10	Conv	8	1 × 5	1
11	ReLU	—	—	—
12	Maxpool	—	1 × 2	2
13	Conv	8	1 × 5	1
14	ReLU	—	—	—
15	Maxpool	—	1 × 2	2
16	Conv	8	1 × 5	1
17	ReLU	—	—	—
18	Maxpool	—	1 × 2	2
19	Fc1	48	—	—
20	Fc2	32	—	—
21	Fc3	1	—
22	Regression		—	—