Open AccessNEURAL NETWORK INFORMATION-MEASURING SYSTEM IN TASKS OF RESOURCE PREDICTION
Author(s) -
Mari. Dubyago,
N.K. Poluyanovich
Publication year - 2021
Language(s) - English
Resource type - Conference proceedings
DOI - 10.30987/conferencearticle_61c997ee5142f0.21613507
Subject(s) - artificial neural network , computer science , activation function , function (biology) , power (physics) , layer (electronics) , resource (disambiguation) , thermal , mean squared prediction error , artificial intelligence , machine learning , computer network , materials science , quantum mechanics , evolutionary biology , meteorology , composite material , biology , physics
It was established that methods based on artificial neural networks (HC) find the most widespread in predicting thermal processes in power cable networks. Analysis of influence of various functions of HC activation on forecast error of thermoflux processes in power cable networks was carried out. It is established that the minimum error of thermal processes prediction in power cable networks is HC with function of logsig activation in hidden layer and pureline in output layer.