摘要
针对米糠蛋白碱法生产中存在的产品得率低、色泽深等普遍问题,本文在单因素实验确定的几个蛋白提取关键影响因素基础上,采用中心组合实验的方法探讨了因素之间交互作用的问题,最后利用广义回归神经网络技术(general regression neural network,GRNN)对蛋白提取率和提取液色泽进行了多目标优化问题研究。GRNN模型结果显示:在温度36.5℃,液料比11.5:1(v/w),pH10.9,辅助剂用量0.56%的条件下,米糠蛋白提取率理论值为61.0%,色度值为53.3,与实际值的误差分别为3.6%和4.7%,且相比较目前工业生产普遍使用的生产条件,蛋白提取率与色度值分别提高了31.4%与43.3%,采用GRNN方法优化的米糠蛋白提取条件具有很好的实用价值。
According to the general problems including the low yield and the deep colour of product during the alkali-extraction process of rice bran protein,the cross-action of different factors was discussed through central composite experiments on the basis of the key factors determined by single factor experiments.Furthermore,the optimization of interaction-effect including protein yield and color were performed by generalized regression neural network（GRNN）method.The GRNN model results showed that 61.0% protein yield and 53.3 color value were obtained under the conditions of temperature 36.5℃,ratio of solve to material 11.5：1（v/w）,pH value 10.9,and auxiliary agent amount 0.56%,which existed a deviation of 3.6% and 4.7% with actual values,respectively.Compared with the conditions for industrial production,the protein extraction yield and colour value raised 31.4% and 43.3%,respectively,under the optimized conditions.It was proved that the optimized extraction conditions of rice bran protein by GRNN had great pragmatic value for industrial production.
出处
《食品工业科技》
CAS
CSCD
北大核心
2012年第15期187-190,共4页
Science and Technology of Food Industry
基金
国家自然科学基金(31101348
31000780)
上海市晨光计划项目(2008CG55
09CG50)
