| 초록 |
The goal of optimization is to maximize productivity and efficiency. Response surface methodology (RSM) is widely used to optimize process parameters. RSM shows low accuracy because it is poor predicting performance when out of range. In order to improve of RSM performance, research to optimize the artificial neural network (ANN) and genetic algorithm (GA) by combining with RSM is being conducted. In this study, vacuum frying (VF) technology was applied to the production process to produce healthy sweet potato chips. However, vacuum fried chips do not taste better than deep fried chips and have a lower brownness. In this study, the variables, osmotic dehydration (OD) concentrations, OD temperatures and VF temperatures were designed to optimization yield (%), oil content (%) and BI index. The optimal conditions were investigated using RSM and RSM-ANN-GA. From the coefficient of determination, root mean squared error, and mean absolute error were indicated that RSM-ANN-GA provided greater accuracy than the RSM. This research could be utilized in the commercial production of vacuum frying. |
