Neural Network for Predicting the Thermal Conductivity of Steel with the Bayesian Method Using Matlab Software
Yurii V. Shanin, Aleksei S. Bondar, Fedor V. Chmilenko, Qi Zhang · 2021
The article offers a solution of prediction of thermal conductivity for different grades of steel at different temperatures. To solve this problem, the authors consider the possibility of using a neural network based on the Bayesian method. The neural network was built using the Matlab software. The aim of the article is to determine the optimal network structure to ensure better prediction accuracy, both on the training set and on the test set. Much attention is given to analyze the obtained results for different numbers of hidden neurons and compare them with the results of other authors. It was found that the greatest influence on the value of thermal conductivity has the value of temperature and the ratio of alloyed elements. The findings are of direct practical relevance.