摘要
为了探讨不同冷藏时间草莓鲜榨汁的品质变化规律,利用电子舌对4℃冷藏草莓鲜榨汁进行逐天的品质检测并进行定性和定量分析。结果显示:通过主成分分析和线性判别分析,电子舌能够很好的反映出不同冷藏时间草莓的鲜榨汁品质的变化趋势,在冷藏最初期草莓鲜榨汁品质下降迅速,其后鲜榨汁品质下降缓慢。采用BP神经网络对草莓鲜榨汁冷藏时间进行预测,训练集的整体识别率达到94.90%,预测集的整体识别率为85.71%。采用偏最小二乘法与多元线性回归对草莓鲜榨汁的主要成分预测可知,偏最小二乘法回归模型的预测效果更好,模型的预测值和实际值的相关系数大于0.9,能够为草莓鲜榨汁品质的预测提供参考。
In order to discuss the change rules of fresh juice quality of strawberry with different cold storage time, the electronic tongue was used to evaluate the physicochemical properties and qualities of fresh juice of strawberries quantitatively and qualitatively. Based on principal component analysis （PCA） and linear discriminant analysis （LDA）, it was concluded that fresh juice of strawberries stored for different time could be described by electronic tongue well. In early storage of strawberries, quality of the fresh juice decreased rapidly. With middle storage period of strawberries, quality of the fresh juice decreased slowly. Back propagation neural network （BPNN） was also employed to predict the storage time of the samples. The result show that BPNN can get 94.9% training accuracy and 85.71% prediction accuracy. Partial least-square （PLS） regression and diverse line regression were employed to predict the component of fresh juice. PLS regression was more capable of building models with good prediction accuracy （R^2〉0.9） than multiple line regression, which showed the good predictive validity on the quality of fresh juice of strawberries.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2012年第23期250-256,共7页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金(31201368)
国家科学部支撑计划(/2012BAD29B02-4)资助
关键词
主成分分析
偏最小二乘法
模型
电子舌
草莓鲜榨汁
贮藏时间
principal component analysis, least squares approximation, models, electronic tongue, fresh strawberry juice, storage period
