CapNet

Contains machine learning model for capsule fracture in concrete

Requirements

• Python (Python 3.9.7 or newer)
• PyTorch (1.11.0 or newer)

Usage

Please see Example.py which has a full toy example, the key details of which are presented below:

Input Data

The input data consits of five inputs, and should be formatted as:

Interface Stiffness	Interface Strength	Interfact Fracture Energy	Capsule Stiffness	Capsule Strength
1.67E-01	1.00E+00	2.50E-01	3.33E-03	1.67E-01

NB: All inputs are given in relative terms to the matrix data.

Network

The network consitis of input, output, and two hidden layers. Five inputs are expected, for interfacial and capsule parameters, and two outputs are given for the case of fracture and non-fracture of the capsule, being 1 and 0 respectively.

Loading

net = Net()
net = torch.load("NNetModel")
net.eval()

Run data through model

out = net(data)
outFrac = net(dataFrac)

Print output

print(torch.argmax(out).numpy())
print(torch.argmax(outFrac).numpy())