Application of artificial neural network for identification of parameters of a constitutive law for soils

A common problem of excavation machinery based on mechanical actions is the unknown interaction of the cutting tools with geological settings. This interaction determines for different soils a different wear and consequently different economical costs for the excavation. We apply a strategy for soil modelling which is based on discretization of the continuum with rigid disks and suitable contact models and concentrate at contact level the real mechanical behaviour of the soil. In order to carry out the proposed strategy a “macro” and a “micro” level are established. In this paper an application of Artificial Neural Network (ANN) for identification of the parameters of the contact constitutive law is shown. The ANN is first trained using the theoretical results obtained from the developed numerical model. Results of some numerical tests concerning the choice of the proper topology of ANN, the best training set and the sensitivity of the identified parameters are shown.