Sieć neuronowa w optymalizacji zjawisk fizycznych zachodzących w trakcie przejazdu pojazdu szynowego przez tor

In the article, the authors present the concept of using neural network in analysis of data obtained from simulation processes presenting physical phenomena occurring at the moment of passing a rail vehicle through the track. The main parameters in which attention is drawn are the normal, longitudinal, transverse and vertical forces in the contact area resulting from the contact of the wheel with the rail. After a brief introduction, a mathematical model describing a number of physical phenomena occurring between the various components of the rail vehicle is discussed. In the article, after aging of the neural network's own concept and explanation of its functioning, an algorithm has been described, which is responsible for elimination of values of forces deviating significantly from optimal values. Results from simulation studies are presented in two groups. The first one refers to the graphs obtained during the study using the Uniwersal Mechanism software. In the second part the results of the data obtained in the first stage of the test were passed through the neural network. research article