Abstract
The use of artificial neural networks for simulation of the control of an anthropomorphic mechanism when its mass-inertia characteristics are changing is considered in this paper. It is shown in theory that the change in mass-inertia characteristics of the model does not have a linear relationship with the output parameter but a spurt-like change in the pattern of the output function after a certain threshold. The data of the model experiment served as a basis for a full-scale test with discus throwers who had different height-weight parameters during a year of study. Specific traits in the technique of standing discus throwing are shown, and their dependence on the changes of anthropometric characteristics of sportsmen.
