基于BP神经网络的微波真空膨化浆果脆片工艺参数优化

Optimization of Process Parameters for Microwave Puffing Berry Snacks Based on BP Neural Network

为了优化微波真空膨化浆果脆片的工艺参数,在单因素试验基础上做四因素五水平中心组和响应面试验。将响应面试验数据作为神经网络的样本,采用神经网络的方法优化全局最优的工艺条件,优化结果:在微波强度27.92 W/g、初始含水率20%、膨化时间90 s和真空压强36 k Pa条件下,膨化度为3.80。在此条件下做验证试验,神经网络预测值与验证试验数据的相对误差为4.29%,响应面优化数值与验证试验数据的相对误差为6.58%。神经网络方法与遗传算法结合优化食品加工工艺参数是可行的,将其与响应曲面法结合起来,为食品加工工艺参数的优化提供一个有效的方法。

Response surface center combination experiment with four factors and five levels were employed to optimize the process parameters for microwave puffing berry snacks based on single factor experiment before. Experiment data was taken as sample of neural network. Back-Propagation neural network method was used to optimize global optimum process conditions. Microwave intensity is 27.92 W/g, initial moisture content is 20%, puffing time is 90 S and the vacuum pressure is 36 kPa. Puffing degree is 3.80. Verification test was carried out on the optimal process conditions. The relative error of Back-Propagation neural network and experimental data is 4.29% while that of response surface method is 6.58%. It＇s feasible that neural network method combined with genetic algorithm was used to optimize food process parameters. Neural network method combined with response surface method is effective for food process parameters optimization.