Optimization of goose breast meat tenderness by rapid ultrasound treatment using response surface methodology and artificial neural network.

Abstract

The aim of this study was to develop a prediction model on tenderization of goose breast meat by response surface methodology (RSM) and artificial neural network (ANN). The experiments were operated on the basis of a three-level, three-variable (ultrasound power, ultrasound time, and storage time) Box-Behnken experimental design. Under RSM and ANN optimum conditions, experimental Meullenet-Owens razor shear (MORS) of meat (1862.6g and 1869.9g) was in reasonable agreement with predicted one. Nevertheless, better prediction capability of ANN was proved by higher R2 (0.996) and lower absolute average deviation=4.257) compared to those for RSM (0.852 and 16.534), respectively. These results revealed that ANN was more accurate and much better than RSM model for the optimization of tenderness of meat. The optimum conditions of ultrasound power, ultrasound time, and storage time given by ANN were 812W, 24.5min and 25.7hr, respectively. Under the optimized condition, the cooking loss of meat significantly decreased by ultrasound treatment compared with untreated meat. Lower cooking loss and MORS at the optimal condition were beneficial to meet the satisfaction of consumer and producers for meat factory.