摘要
目的：探索近红外光谱定性分析技术鉴别苹果产地的可行性。方法：对天津、陕西和北京3个产地富士苹果品质进行了分析,比较了波段范围、导数处理和散射及标准化处理建立主成分分析结合最小二乘法产地鉴别模型。结果：在1100~1848 nm波段范围内采用一阶导数结合趋势变化法散射处理的光谱预处理方法最优,校正集的鉴别正确率为100%,校正交互验证误差（SECV）0.1245,交互验证相关系数（Rcv）0.9641,预测集的鉴别正确率为98.33%。结论：应用近红外光谱定性分析技术可以准确地、快速地追溯苹果产地的溯源。
Objective： This study aimed to explore the feasibility of the discrimination method of different origin apples by near infrared qualitative spectrochemical analysis technique. Method： Quality analysis of fuji apples produced in Tianjin, Shaanxi and Beijing, and effect of spectral range, derivative, different scattering and standard treatments on origin discrimination model with principal component analysis（PCA） and partial least squares（PLS） were studied. Result： The PCA-PLS model with respect to first derivative combined with detrend only in the wavelength range of 1100～1848 nm, provided better prediction performance for different origin apples, with correct classification of calibration of 100%, with standard error of cross-validation of 0.1245, with correlation coefficient of 0.9641, with correct classification of prediction of 98.33%. Conclusion： Different origin apples could be retrospected by near infrared qualitative spectrochemical analysis technique accurately and rapidly.
出处
《食品科技》
CAS
北大核心
2014年第11期305-309,共5页
Food Science and Technology
基金
国家科技支撑项目(2012BAD38B01)
天津市农业科学院院长基金项目(12004)
(11010)
关键词
苹果
近红外光谱
产地
鉴别
apple
near infrared spectrum
origin
discrimination
