Artificial neural networks for prediction of local thermal insulation of clothing protecting against cold

Purpose The purpose of this paper is to develop artificial neural networks (ANNs) allowing us to simulate the local thermal insulation of clothing protecting against cold on a basis of the characteristics of materials and design solutions used. Design/methodology/approach For this purpose, laboratory tests of thermal insulation of clothing protecting against cold as well as thermal resistance of textile systems used in the clothing were performed. These tests were conducted with a use of thermal manikin and so-called skin model, respectively. On a basis of results gathered, 12 ANNs were developed that correspond to each thermal manikin's segment besides hands and feet which are not covered by protective clothing. Findings In order to obtain high level of simulations, optimization measures for the developed ANNs were introduced. Finally, conducted validation indicated a very high correlation (above 0.95) between theoretical and experimental results, as well as a low error of the simulations (max 8 percent). Originality/value The literature reports addressing the problem of modeling thermal insulation of clothing focus mainly on the impact of the degree of fit and the velocity of air movement on thermal insulation properties, whereas reports dedicated to modeling the impact of the construction of clothing protecting against cold as well as of diverse material systems used within one design of clothing on its thermal insulation are scarce.