Статті КЕОМ

Permanent URI for this collectionhttp://crust.ust.edu.ua/handle/123456789/696

ENG: Articles

Browse

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image
    Item type:Item,
    Intelligent Computer Network for Railway Transport Using Neural Network for Determining the Optimal Route
    (Sergeieva&Co, Karlsruhe, Germany, 2025) Pakhomova, Victoria M.; Budnikov, Oleksandr
    ENG: At the present stage, the information and telecommunications system of railway transport uses local area networks of Ethernet family technologies and the OSPF protocol, when used in real time, a problem arises due to constant changes in the volume of transmitted data, and for its solution it is advisable to use a neural network tool, which confirms the relevance of the topic. As a mathematical apparatus for solving the problem of determining the optimal route, a neural network of the configuration «56-1-X-56» was taken, where 56 (first position) is the number of input neurons (delays on routers); 1 is the number of hidden layers; X is the number of hidden neurons that require additional research; 56 (last position) is the number of output neurons (signs of the entry of computer network channels into the route). In the program mode of the Deep Learning Toolbox package of the MatLAB environment, a corresponding model “Delay_path” was created, on which the root mean square error and the number of epochs of training of neural networks with different numbers of hidden neurons were studied using different neuron activation functions according to different learning algorithms on samples of different lengths. It was determined that the accuracy of the created neural network tool is 70 % for the considered fragment of the information and telecommunications system of railway transport.