基于近红外光谱技术构建民勤羊肉养分含量检测模型  

Model construction for nutrient content detection of Minqin lamb based on near-infrared spectroscopy

作　　者：梁静 赵祥民 李璐璐 康景 唐德富[2] 刘佳[1] 郭怡文 杨涛 郝生燕 窦晓利 宋淑珍[1] LIANG Jing;ZHAO Xiangmin;LI Lulu;KANG Jing;TANG Defu;LIU Jia;GUO Yiwen;YANG Tao;HAO Shengyan;DOU Xiaoli;SONG Shuzhen(Animal Husbandry,Pasture and Green Agriculture,Gansu Academy of Agricultural Science,Lanzhou 730070,China;College of Animal Science and Technology,Gansu Agricultural University,Lanzhou 730070,China;Agricultural and Livestock Products Quality and Safety Inspection and Testing Center,Maqu County,Linxia 747399,China;Agricultural and Rural Bureau,Maqu County,Linxia 747399,China)

机构地区：[1]甘肃省农业科学院畜草与绿色农业研究所,甘肃兰州730070 [2]甘肃农业大学动物科学技术学院,甘肃兰州730070 [3]玛曲县农畜产品质量安全检测检验中心,甘肃临夏747399 [4]玛曲县农业农村局,甘肃临夏747399

出　　处：《食品与发酵工业》2023年第14期272-279,共8页Food and Fermentation Industries

基　　金：甘肃省青年科技基金计划(20JR5RA102);甘肃省农业科学院中青年基金(2020GAAS36);甘肃省农业科学院科技创新专项重点研发计划(2019GAAS18);甘肃省科技计划重点研发项目(20YF3 NA024)。

摘　　要：应用近红外光谱技术结合化学计量学法快速预测民勤羊肉(地理标志产品)营养成分含量,旨在为定量检测民勤羊肉营养成分含量提供理论依据以及为民勤羊肉的营养特性和产品价值提供技术支持。采集民勤90头羊的背最长肌肉、前腿肉、后腿肉共269份肉样,并采集近红外光谱。采用多种预处理方法建立羊肉肌肉脂肪、蛋白质和脂肪酸含量的主成分回归法(principal component regression,PCR)、偏最小二乘法(partial least squared PLS)、改良偏最小二乘法(modified partial least squared,MPLS)的定标模型。除硬脂酸最佳建模方法为PLS外,其余指标模型最佳建模方法均为MPLS。其中,脂肪、蛋白、单不饱和脂肪酸模型的决定系数(RSQ)均在0.9以上,分别为0.936、0.916、0.911;棕榈酸、油酸、亚油酸、饱和脂肪酸、多不饱和脂肪酸模型RSQ值均大于0.8,小于0.9,分别为0.806、0.843、0.883、0.852、0.895;肉豆蔻酸和硬脂酸模型RSQ值均小于0.7,分别为0.323和0.561。说明应用近红外技术结合化学计量学法可以有效预测羊肉肌肉营养成分含量。This study applied near-infrared spectroscopy combined with chemometric methods to rapidly predict the nutrient content of Minqin lamb(a geographical indication product),aiming to provide a theoretical basis for the quantitative detection of the nutrient content of Minqin lamb and technical support for the nutritional properties and product value of Minqin lamb.A total of 269 meat samples were collected from the longest back muscle,foreleg meat,and hind leg meat of 90 sheep in Minqin,and near-infrared spectra were collected.The calibration models of lamb muscle fat,protein,and fatty acid content were established by using various pretreatment methods,including principal component regression(PCR),partial least squares(PLS),and modified partial least squares(MPLS).Except that the best modeling method of stearic acid was partial least squares,the best modeling method of other indicator models was improved partial least squares.Among them,the RSQ of fat,protein,and monounsaturated fatty acid models were higher than 0.9,which were 0.936,0.916,and 0.911,respectively.The RSQ values of palmitic,oleic,linoleic,saturated,and polyunsaturated fatty acid models were greater than 0.8 and less than 0.9,which were 0.806,0.843,0.883,0.852,and 0.895,respectively,while the RSQ values of myristic acid and stearic acid models were less than 0.7,which were 0.323 and 0.561,respectively.The application of the near-infrared technique combined with chemometric methods can effectively predict the nutritional composition of lamb muscle.

关 键 词：近红外光谱技术 民勤羊肉 定量检测 营养成分 化学计量学法 

分 类 号：TS251.53[轻工技术与工程—农产品加工及贮藏工程] O657.33[轻工技术与工程—食品科学与工程]