Analysis of the Relationship between the Load-Displacement Curve and Characteristics of Fracture of Low-Alloy Steel by Neural Networks

The relationship between the parameters of the load-displacement curves and shear fracture area (SFA) of samples during impact bending tests has been studied. Using neural networks, a correspondence has been established between the form of the load-displacement curve and the fracture structure in a series of different microstructures. The applicability of the trained neural network to low-alloy steel with a microstructure that is different from the microstructures of the samples used in training has been tested. The minimal set of parameters of the load-displacement curve describing the shear fracture area of the sample is established. The nonlinearity of the relationship between the parameters of the dynamic curve and the shear fracture area of the sample is shown.