Metasurface-Based Solar Absorption Prediction System Using Artificial Intelligence

Table 2

Workflow of the proposed prediction model.

Name of layers	Description
Input data	Featured metasurface solar absorbance
ConV1	11 × 11 conv/stride 4 and [0 0 0 0] of padding
ReLu1	Activation 1
Pool1	3 × 3 max pooling/stride 2 and [0 0 0 0] of padding
ConV 2	5 × 5 conv/stride 2 and [2 2 2 2] of padding
ReLu 2	Activation 2
Pool 2	3 × 3 max pooling/stride 2 and [0 0 0 0] of padding
ConV 3	3 × 3 conv/stride 1 and [1 1 1 1] of padding
ReLu 3	Activation 3
Pool 3	3 × 3 max pooling/stride 2 and [0 0 0 0] of padding
ConV 4	3 × 3 conv/stride 1 and [1 1 1 1] of padding
ReLu4	Activation 4
Pool 4	3 × 3 max pooling/stride 2 and [0 0 0 0] of padding
ConV 5	3 × 3 conv/stride 1 and [1 1 1 1] of padding
ReLu 5	Activation 5
Pool 5	3 × 3 max pooling/stride 2 and [0 0 0 0] of padding
Fully connected 1	4096FC
Dropout	50%
Fully connected 2	4096FC
Fully connected 3	1000FC
Softmax	Softmax activation
Ouptut	Absorption prediction