Prediction model of thermal conductivities of agricultural products postharvest

Knowledge of the thermal conductivity of agricultural products postharvest was fundamentally important in mathematical modeling studies for the design and optimization of agricultural products processing operations involving heat and mass transfer. The objective of this work was to develop an appropriate measurement system to obtain thermal conductivity values for yellow millet with various water contents at different temperature. A BP artificial neural network technique worked satisfactorily for the prediction of thermal conductivities of yellow millet as a function of water content and temperature. The optimal model consisted 2 hidden layer with 4 neurons per layer was obtained by comparison and analysis of the errors. The result shows that the optimal model was able to predict thermal conductivity with a mean relative error of 2.25%, a mean absolute error of 0.0029 W·m^-1 K^-1. As a result, the method can predict the thermal conductivities values of yellow millet at an acceptable level for engineering applications.