Artificial neural network approaches for the sorption isotherms, enthalpy and entropy of heat sorption of two types block rubber products

Knowledge of the temperature and relative humidity (or water activity) dependence of moisture sorption phenomenaof agro-industrial products provides valuable information about changes related to the thermodynamics of the system. Thusthe moisture sorption characteristics in terms of equilibrium moisture contents (EMC), enthalpy and entropy of heat sorptionof natural rubber (NR) for producing STR 20 and skim block rubber were investigated. Simulation modeling of water sorptionisotherms was performed using the 10 non-linear regression EMC models and the multilayer artificial neural network (ANN)approach. The results showed that the predicted EMC results using the modified Oswin model was the best fitting model forboth NR samples. However, the predicted values of ANN model were more accurate than those predicted results using nonlinear regression method. Finally, enthalpy and entropy of heat sorption for both NR samples were evaluated by applying theClausius-Clapeyron equation showing as the negative exponential function of moisture content.