基于近红外高光谱成像快速预测牛肉中猪肉掺入量  

Fast Prediction of Pork in Beef Based on Near-infrared Hyperspectral Imaging

作　　者：陈亮亮 朱亚东[1] 李梦姣 李柯杰 程达 李旭卓 何鸿举[1] CHEN Liangliang;ZHU Yadong;LI Mengjiao;LI Kejie;CHENG Da;LI Xuzhuo;HE Hongju(School of Food Science,Henan Institute of Science and Technology,Xinxiang 453003,China)

机构地区：[1]河南科技学院食品学院,河南新乡453003

出　　处：《海南师范大学学报（自然科学版）》2022年第4期402-406,共5页Journal of Hainan Normal University（Natural Science）

基　　金：河南科技学院大学生创新创业计划项目(2021CX027);河南省科技攻关项目(212102310491)。

摘　　要：基于900~1700 nm波段范围内的近红外高光谱成像技术结合线性算法快速预测牛肉中猪肉掺入量。使用BC、SGS和SNV3种方法预处理光谱信息,结果显示原始光谱信息更适合于构建PLSR模型,预测牛肉掺假效果良好(R2P=0.95,RMSEP=6.82%,RPD=4.89)。经Stepwise算法从全波段486个波长中筛选出8个最优波长(918、926、935、1053、1055、1086、1088、1382 nm)简化全波段预测模型,结果显示,基于这8个最优波长构建的MLR模型预测效果良好(R2P=0.96,RMSEP=5.70%,RPD=5.56),可代替全波段光谱信息实现牛肉掺假的快速预测。A fast method for the prediction of pork content in beef was conducted based on near-infrared hyperspectral imaging combined with linear algorithm in the range of 900~1700 nm.Three methods including BC,SGS and SNV were used to preprocess spectral information,and the results showed that the original raw spectral information was more suitable for the construction of PLSR model,and the performance of adulteration prediction was good(R2P=0.95,RMSEP=6.82%,RPD=4.89).The Stepwise algorithm was used to select 8 optimal wavelengths from 486 wavelengths of the whole bands such as 918,926,935,1053,1055,1086,1088,1382 nm to simplify the full-band prediction model,and the results showed that the prediction performance of MLR model based on the 8 optimal wavelengths was also good(R2P=0.96,RMSEP=5.70%,RPD=5.56),which can replace the full-band spectral information to realize the rapid prediction of beef adulteration.

关 键 词：猪肉 牛肉 掺假 光谱 变量 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程]