Evaluation of sensory characteristics and modeling of the kinetics of hazelnut color indices during infrared roasting

In this study, the effects of infrared beams in the process of hazelnut roasting in three powers of 700, 1100 and 1550 watts at times of 3, 6, 9 and 15 minutes were investigated. Also, the asymmetric methods (completely randomized factorial and partial least squares regression) and symmetric methods (coefficients of determination and principal components analysis) were used for study of color parameters (L, a and b) and sensory aspects (color, odor, taste, texture and overall acceptability). Increase of power and roasting time lead to decrease of L and increase of a. On the other hand, this experiment showed that b component increased with increasing of roasting time until 6 min and then decreased. The highest scores of sensory characteristics were obtained under 6 min infrared roasting at power of 1550 watts and 12 min at power of 1100 watts. Different models were compared in order to simulation of the behavior of hazelnut color components during roasting process and the Suitable models were selected according to the statistical tests of correlation coefficient, χ2, mean bias error and root mean square error. It could be found from principal components analysis (PCA) that the samples with the highest scores of sensory characteristics were in medium ranges of color indices. Correlation coefficients between sensory color and all of color indices were significant. Among three color indices, b index had the most significant relationships with sensory characteristics. Obtained R2 of partial least squares regression models were above the medium range. It shows the high efficiency of these models in prediction of sensory characteristics of infrared roasted hazelnuts from color analysis indices.