反向传播神经网络耦联遗传算法与响应面设计烤制鸽肉工艺优化  

Process Optimization of Roasted Pigeon Meat by Back Propagation Neural Network Coupled with Genetic Algorithm and Response Surface Methodology Design

作　　者：赵清香 李大军[1] 李亚萍 姜宇纯 李庚 袁永旭 ZHAO Qing-xiang;LI Da-jun;LI Ya-ping;JIANG Yu-chun;LI Geng;YUAN Yong-xu(College of Food Science and Engineering,Jilin Agricultural University,Changchun 130118,China)

机构地区：[1]吉林农业大学食品科学与工程学院,长春130118

出　　处：《中国调味品》2023年第10期128-133,共6页China Condiment

基　　金：吉林省市场监督管理厅项目(BY-FWZB-20200905)。

摘　　要：为研究烤制鸽肉工艺便捷设计方法和确定关键工艺参数,采用烤制工艺制备烤制鸽肉,探究腌制时间、烘烤时间和烘烤温度对产品食用性状的影响,以回复性、弹性、硬度、咀嚼性、胶着性、内聚性及感官评分为食用性状评价指标,采用信息熵法确定各指标的熵权系数,计算出综合值,利用响应面法(response surface methodology,RSM)和反向传播(back propagation,BP)人工神经网络耦联遗传算法(genetic algorithm,GA)(BP-GA)进行参数优化及预测。结果表明,RSM得到最优工艺参数组合为腌制时间15.4 h、烘烤时间10 min、烘烤温度205℃,该法相对误差为2.54%;BP-GA神经网络得到最优工艺参数组合为腌制时间14 h、烘烤时间10 min、烘烤温度240℃,该模型相对误差为0.17%。BP-GA神经网络的相对误差小,拟合与优化性能好,且试验次数少,可容纳参数多。依托BP-GA神经网络优化的参数制作的烤制鸽肉综合值与RSM一致。该研究可为鸽肉加工工艺优化提供一定的参考。To study the convenient design method of roasted pigeon meat process and determine the key process parameters,roasted pigeon meat is prepared by roasting process,and the effects of curing time,baking time and baking temperature on the edible characters of the product are investigated.With resilience,elasticity,hardness,chewiness,adhesiveness,cohesiveness and sensory score as the edible character evaluation indexes,the entropy weight coefficient of each index is determined by information entropy method and the comprehensive value is calculated.Response surface methodology(RSM)and back propagation(BP)artificial neural network coupled with genetic algorithm(GA)(BP-GA)are used to optimize and predict parameters.The results show that the optimal combination of process parameters obtained by RSM are curing time of 15.4 h,baking time of 10 min and baking temperature of 205℃,and the relative error of this method is 2.54%.The optimal combination of process parameters obtained by BP-GA neural network is curing time of 14 h,baking time of 10 min and baking temperature of 240℃,and the relative error of this model is 0.17%.The relative error of BP-GA neural network is small,the fitting and optimization performance is good,the number of tests is few,and BP-GA neural network can accommodate many parameters.The comprehensive value of roasted pigeon meat optimized by BP-GA neural network is consistent with that of RSM.This study can provide some references for the optimization of pigeon meat processing technology.

关 键 词：烤制鸽肉 神经网络 遗传算法 响应面法 工艺参数 

分 类 号：TS251.55[轻工技术与工程—农产品加工及贮藏工程]