基于GA-BP神经网络的蒜香调味粉制备工艺优化

Optimization of Processing Conditions of Garlic Flavoring Powder Based on GA-BP Neural Network

为优化蒜香调味粉的制备条件,研究干燥温度、切片厚度及干燥时间对蒜香调味粉风味特点的影响,采用热风干燥法制备高品质蒜香调味粉。以感官评分为响应值,进行Box-Behnken响应面法设计,并采用反向传播(Back propagation,BP)人工神经网络和遗传算法(Genetic algorithm,GA)相结合的设计方法对响应面法所得结果进行相互验证。结果表明:经神经网络结合遗传算法优化,最优参数为:干燥时间5.9 h,干燥温度61℃,切片厚度2.7 mm。此条件下蒜粉感官评分的试验值为18.70分,GA-BP神经网络模型计算出的预测值为18.5622分,试验值和预测值之间的相对误差为0.74%,模型的拟合度很好,说明应用GA-BP神经网络优化蒜香调味粉制备条件合理可行。

The effects of drying temperature,slice thickness and drying time on the flavor quality of garlic flavoring were studied in order to optimize the preparation conditions of garlic flavoring powder,and high-quality garlic flavoring powder was prepared by using hot air drying method.Box-Behnken design and response surface method were used to experiment with sensory score as response value.The results of the response surface method were mutually verified by the combination of artificial neural network and genetic algorithm.The results indicated that the optimal conditions for the preparation of garlic flavoring powder were as follows:the drying temperature was 61℃,the drying time was 5.9 h,and the slice thickness was 2.7 mm.The sensory score of the product obtained under this condition was 18.70 points and the relative error from the model’s predicted value was 0.74%.The predicted value calculated by the GA-BP artificial neural network model was 18.5622 points.This means that the experimental value fits well with the predicted value of the model.Therefore,artificial neural network combined with genetic algorithm is a reasonable and feasible method for the preparation conditions of flavoring garlic powder.