Application of a neural network mathematical model in the development of hot air roasting process technology for Chironji (Buchanania lanzan) kernels

The hot air roasting of Chironji kernels was performed at 25 different combinations of temperature and time. The influence of roasting parameters on bulk density (BD), hardness, water absorption capacity (WAC), water activity (WA), browning index (BI), total phenol content (TPC), and antioxidant activity (AA) were assessed. A decrease in BD, hardness, and WA, while an increase in WAC, BI, TPC, and AA was observed in roasted kernels. An Artificial Neural Network (ANN) model combined with the Genetic Algorithm (GA) was used for obtaining the optimum values of roasting parameters. The values of responses measured at the optimum conditions of 146°C temperature and 34 min were BD of 537 ± 1.21 kg/m³, hardness of 15.01 ± 0.31 N, WAC of 1.33 ± 0.11 g/g, WA of 0.0059 ± 0.007, BI of 74.62 ± 0.33, TPC of 17.68 ± 0.49 mg GAE/g dry wt, and AA of 74.09 ± 0.19%. Practical applications: Value addition of minor forest produce like Chironji kernels can improve the livelihood of the rural populace, and roasted Chironji kernels can be a popular value-added product in the market. The results of this study can help the food processors for roasting of Chironji kernels while retaining its quality characteristics.