Development of an information system for the approximate finding of indicators of an athlete thrower using mathematical modeling of nucleus pushing and the use of neural network technologies

Abstract

Melnykov O. Yu., Kadatsky M. A. Development of an information system for the approximate finding of indicators of an athlete thrower using mathematical modeling of nucleus pushing and the use of neural network technologies // Herald of the DSEA. – 2019. – № 2 (46). – P. 145–149.

The thesis describes the main factors affecting the range of the nucleus. A formula is given for calculating the range, from which it follows that the smaller the force acting on the core, the greater should be the direction angle of this force, and at a certain angle for an angle, an optimal combination of all quantities occurs, which leads to a maximum range of the projectile. The task of designing a system is formulated - an application capable of calculating the main indicators and the result of an athlete in order to use them in the training process to achieve maximum communication between the force and speed of the thrower. An information model of such a system was developed in the form of a set of UML diagrams (use-case diagram, class diagram). Software implementation of the model is described. The results of the software product, showing how fast, with what angle of release and with what force an athlete must push the core to achieve the maximum range of flight, are given. It is concluded that the description of sports equipment solely by the equations of mechanics may not take into account a number of factors which, being of minor significance for the absolute values of the results, can have a serious impact on the relative indicators. The possibility of using modern methods to solve the problem of forecasting is substantiated. The data on the characteristics of eight athletes (age, height, body weight, the preferred method of throwing), as well as their athletic performance (initial speed of the nucleus, throwing angle, height of free hand and distance of flight) are given. Two forecasting tasks were formulated: based on the available data on the age, height, and body mass of the athlete, as well as the characteristics of the core flight, determine the range of this flight; according to the available data on the age, height, body mass of the athlete, as well as the range of the kernel’s flight, determine the optimal combination of flight characteristics – initial velocity, angle and separation height. A method of artificial neural networks with a two-layer perceptron architecture, an activation function by sigmoid, and an error back-propagation algorithm for network training is proposed. Examples of calculation in the environment of Deductor Studio Lite are given.