Modelling approaches for predicting moisture transfer during baking of chhana podo (milk cake) incorporated with tikhur ( Curcuma angustifolia ) starch

Chhana podo, a popular dairy product of India, is prepared by baking the dough of chhana (heat-acid coagulated milk solids), semolina and sugar. Heat induced moisture loss during baking influences the chain of physico-chemical changes, which in turn determine the quality of the final product. Thus, in this work, modelling approaches for predicting moisture transfer during baking of chhana podo prepared by replacing semolina with tikhur ( Curcuma angustifolia ) starch were developed. Moisture loss, moisture ratio, specific volume, oven spring, volumetric expansion and other physico-chemical quality changes of the dough were determined during baking, and were used in the development of moisture transfer models. The effective moisture diffusivity of chhana podo increased from 8.10 × 10 –9 to 37.05 × 10 –9 m 2 /s as baking temperature increased from 110 to 150 °C. The artificial neural networks (ANN) model based on multilayer feed-forward (MLFF) algorithm and radial basis function (RBF) and adaptive neuro-fuzzy inference system (ANFIS) models based on Sugeno-type fuzzy inference system were developed to predict moisture ratio, and their predictive performance was compared with classical multiple linear regression and empirical models. Amongst the four prediction approaches, the MLFF neural network with baking temperature and time as input factors, and with 12 neurons in the hidden layer, produced the best performance in predicting moisture ratio with R 2 as high as 0.9946 (99.46% accuracy) and RMSE of 0.0188 for the entire range of baking temperature studied. Sensitivity analysis showed that baking temperature was the most influential parameter (strength ratio was 0.89) deciding moisture transfer. MLFF neural network was observed to be a simple and efficient technique for predicting the complex moisture transfer phenomenon during baking, which could be used by the industry to optimize the baking conditions.