DOMINANT TECHNOLOGIES IN “INDUSTRY 4.0”
Application of artificial neural networks for assessing the Psi-factor of thermal bridges under various geometries and materials
Industry 4.0, Vol. 9 (2024), Issue 5, pg(s) 173-177
The publication examines the use of artificial neural networks to calculate the linear thermal conductivity (Psi-factor) of thermal bridges given various parameters, such as geometrical data and the thermal resistance R of the thermal bridge components. The neural network is trained on examples of IF, IW, and B thermal bridges, considering the straightforward task of determining Psi using given parameters. The neural network training results show high accuracy in calculations – RMSE is 1.132% on training data and 1.1423% on test data, and the correlation coefficient (R²) is around 0.9997 for both data sets. The applicability of the approach to seismic conditions in the Balkans is assessed.