Determination of Properties of Materials by the SphericalIndentor: A Neural Network Approach

Studies spherical indentation, with which experiments have widely been performed for the determination of mechanical properties of materials. Characters of feed forward backpropagation neural networks, that has the steepest descent with momentum are then analyzed. The relationship between the applied load P and the depth of penetration h is discussed. Based on the experimental formulation, the neural network has been established. On training the established networks with the data from spherical indentator with a repition of 5 000 times. Analysis and comparison of results from neural network and from spherical indentation are carried out. As a result, using this kind of neural network the mechanical properties of materials can well be modelled by using spherical indentation experiments.