基于熵权TOPSIS法和遗传算法-反向传播神经网络模型的酶-盐联合嫩化牛肉工艺参数优化OA

To investigate the optimal process parameters for enzyme-salt combined tenderization of fresh beef hind legs,this study evaluated tenderization effectiveness using shear force,water-holding capacity,cooking loss rate,myofibril fragmentation index(MFI),and hardness as indicators.The entropy-weighted technique for order preference by similarity to ideal solution(TOPSIS)were employed to comprehensively evaluate multiple indicators and screen the optimal tenderization solution combination.Subsequently,using shear force as the key indicator,and based on single-factor experiments results,the genetic algorithm-backpropagation neural network(GA-BPNN)and response surface methodology were employed to optimize the main process parameters for beef tenderization,including tenderization time,temperature,and solution concentration.The effects of these factors on the sensory score of cooked beef were also investigated.The results demonstrated that the entropy-weighted TOPSIS method identified bromelain-papain-sodium tripolyphosphate(BRO-PAP-STPP)as the optimal tenderization solution,exhibited significantly higher MFI and water-holding capacity than other groups(P＜0.05),with the highest comprehensive evaluation index(Ci=0.944 2).Compared to response surface methodology,the GA-BPNN demonstrated superior global optimization capabilities,yielding predicted values closer to response experimental measurements,resulting in optimized parameters:tenderization time of 64 min,temperature of 51℃,and solution concentration of 7.8 mg·mL-1.The optimized process parameters significantly enhanced beef tenderness while maintaining favorable sensory quality.This study provides critical data support for developing tenderization solutions and optimizing industrial-scale beef tenderization processes.