节点文献

应用近红外光谱技术快速识别山楂汁品种的研究

Application of Near-Infrared Spectroscopy to quickly identify species of hawthorn juice research

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 赵华民张淑娟张海红杨福存

【Author】 Zhao Huamin, Zhang Shujuan, Zhang Haihong, Yang Fucun (College of Engineering and Technology,Shanxi Agricultural University,Shanxi,Taigu,030801)

【机构】 山西农业大学工程技术学院

【摘要】 为了实现山楂汁品种的快速无损鉴别,提出了一种采用近红外光谱分析技术快速无损鉴别山楂汁品牌的新方法。首先用主成分分析法对山楂汁品牌进行聚类分析,再结合人工神经网络技术进行品种鉴别。采用美国ASD公司的FieldSpec3光谱仪对三种不同品种的山楂汁进行光谱分析,各获取30个样本数据。采用平均平滑法和多元散射校正(MSC)方法对样本数据进行预处理,再用主成分分析法对光谱数据进行聚类分析并获得各主成分数据。将样本随机分成40个建模样本和十个预测样本,将建模样本的10个主成分数据作为BP网络的输入变量,山楂汁品种作为输出变量,建立三层人工神经网络鉴别模型,并用该模型对十个预测样本进行预测。结果表明,在阈值设定为±0.15的情况下,该模型对预测集样本品种鉴别准确率达到了100%。所以应用近红外光谱技术结合主成分分析和神经网络算法能够快速准确的判定山楂汁的品种。

【Abstract】 In order to realize the hawthorn juice fast lossless distinction, a technique using near-infrared spectroscopy to identify non-destructive rapid new method for species of hawthorn juice. First of all, using principal component analysis carried out on cluster analysis of hawthorn juice varieties, combined with the artificial neural network technology for species identification. American ASD spectrometer FieldSpec3 company on three different species of hawthorn juice for spectral analysis, all 30 samples to obtain data. The use of the average smoothing method and the multiple scatter correction (MSC) method of sample pre-treatment data, and then principal component analysis of spectral data for cluster analysis and principal component of the data. Samples were randomly divided into 40 samples and 10 prediction modeling samples, 10 samples will be modeling the data as a principal component of the BP network input variables, hawthorn juice varieties as output variables, the establishment of three artificial neural network identification model, and the model prediction of 10 samples for prediction. The results showed that the threshold is set to ± 0.15, the set of the model to predict the accuracy of the sample species identification reached 100%. Therefore, the application of near-infrared spectroscopy combined with principal component analysis and neural network algorithm can rapidly and accurately determine the species of hawthorn juice.

【基金】 山西省科技攻关项目(项目编号:2007031109-2)
  • 【会议录名称】 纪念中国农业工程学会成立30周年暨中国农业工程学会2009年学术年会(CSAE 2009)论文集
  • 【会议名称】纪念中国农业工程学会成立三十周年暨中国农业工程学会2009年学术年会(CSAE 2009)
  • 【会议时间】2009-08-22
  • 【会议地点】中国山西太谷
  • 【分类号】TS255.44
  • 【主办单位】中国农业工程学会
节点文献中: