基于近红外光谱与PLS-DA的红枣品种识别研究  

Jujube species identification based on near infrared spectroscopy and PLS-DA

作　　者：张景川[1,2] 张晓[3] 白铁成 石鲁珍[3] 

机构地区：[1]塔里木大学机械电气化工程学院,新疆阿拉尔843300 [2]新疆维吾尔自治区普通高等学校现代农业工程重点实验室,新疆阿拉尔843300 [3]塔里木大学信息工程学院,新疆阿拉尔843300 [4]南疆农业信息化研究中心,新疆阿拉尔843300

出　　处：《食品工业科技》2017年第8期68-71,76,共5页Science and Technology of Food Industry

基　　金：国家自然科学基金项目(61501314;41561088;61462074);应急管理项目(61640413)

摘　　要：新疆红枣品种繁多,采后红枣在加工过程中需要将其他品种的红枣挑选出,本研究应用近红外光谱分析技术结合偏最小二乘判别分析(PLS-DA)法对新疆红枣品种进行判别。结果表明,采用一阶导数对原始光谱进行预处理,并使用方差分析法选择波长变量结合PLS-DA方法对校正样本建立判别分析模型,其验证集预测结果与实际分类结果的相关系数(RP)均大于0.92,预测标准偏差(RMSEP)都小于0.27,最后模型对验证集中的骏枣、灰枣和冬枣3个品种的识别率都为100%。该结果为新疆红枣品种快速识别提供理论依据。There are a variety of jujubes in Xinjang.It＇ s necessary to pick out other varieties of jujubes in jujube processing. This research was conducted to attempt to discriminate jujube varieties by the method that near-infrared spectroscopy combined with partial least squares discriminant analysis （ PLS- DA ） method. The discriminatory analysis models of jujube varieties was established by using first derivative （ FD ）, the wavelength selection method of deviation analysis and partial least square discriminant analysis （PLS- DA）.The experimental results showed that the correlation coefficient （Rp）of validation set predictions with the actual classification was greater than 0.92, the standard error of prediction （RMSE）was less than 0.27. Finally, the model recognition rate for three varieties of Jun-jujube, Hui-jujube and Dong-jujube was 100%.The result offered theory evidences for the varieties identification of jujube in Xinjang.

关 键 词：近红外光谱 PLS-DA 品种识别 红枣 方差分析 

分 类 号：TS207.3[轻工技术与工程—食品科学]