基于独立组分分析和BP神经网络的可见/近红外光谱稻谷年份的鉴别  

DISCRIMINATION YEARS OF ROUGH RICE BY USING VISIBLE/NEAR INFRARED SPECTROSCOPY BASED ON INDEPENDENT COMPONENT ANALYSIS AND BP NEURAL NETWORK

作　　者：邵咏妮[1] 曹芳[1] 何勇[1] 

机构地区：[1]浙江大学生物系统工程与食品科学学院,浙江杭州310029

出　　处：《红外与毫米波学报》2007年第6期433-436,共4页Journal of Infrared and Millimeter Waves

基　　金：国家自然科学基金(30671213);国家十一五科技支撑计划(2006BAD10A07);高等学校博士学科点专项科研基金(20040335034)资助项目

摘　　要：建立了一种基于独立组分分析的可见/近红外光谱反射技术快速鉴别稻谷年份的新方法.首先用独立组分分析方法获取不同年份稻谷的可见/近红外光谱载荷图,将载荷图中相关性最大的波段(特征波段)作为人工神经网络的输入建立稻谷年份的鉴别模型.每个年份40个样本,3个年份共120个样本用来建立BP神经网络模型,剩余的3个年份各20个样本用于预测.预测的结果表明,准确率达到100%.同时通过独立组分分析,得到了稻谷主要成分对应的敏感波段.说明本文提出的基于独立组分分析的方法具有很好的鉴别效果,为稻谷的年份鉴别提供了一种新方法.A new method for discrimination years of rough rice based on independent component analysis was developed by using visible/near infrared spectroscopy （Vis/NIRS）. First, the Vis/NIR loading weight of rough rice with different years was got by using independent component analysis （ICA） and setting the wavelengths corresponding to the maximal correlation as the inputs of artificial neural network （ANN）, then the discrimination model was build. 120 samples （40 with each year） from three years were selected randomly as a calibration set; the left 60 samples （20 with each year） were as the prediction set. The discrimination rate of 100% was achieved. Synchronously, the sensitive wavelengths corresponding to the main components in rough rice were obtained with ICA. It indicates that the result for discrimination years of rough rice is very good based on ICA method, and it offers a new approach to the fast discrimination years of rough rice.

关 键 词：可见/近红外光谱 稻谷 独立组分分析 BP神经网络 

分 类 号：S511.22[农业科学—作物学] S511.33