基于近红外光谱的猪肉蛋白质及脂肪含量检测  

Detection of meat protein and fat percentage composition based on near-infrared spectrum

作　　者：林岩[1] 郭培源[1] 王昕琨[1] 

机构地区：[1]北京工商大学计算机与信息工程学院,北京100048

出　　处：《食品科技》2014年第2期262-266,共5页Food Science and Technology

基　　金：北京市自然科学基金项目(4122020)

摘　　要：蛋白质及脂肪是猪肉的重要营养成分。随着人们对饮食健康的要求越来越高,对猪肉蛋白质及脂肪含量快速检测也成为必然。通过近红外光谱技术对猪肉进行光谱数据采集,将光谱数据分为校正集样本和预测集样本,然后,在MATLAB中利用多元散射校正(MSC)与均值中心化相结合的方法进行光谱预处理并采用联合区间偏最小二乘方法(SiPLS)获得猪肉蛋白质及脂肪含量与光谱数据特征之间的对应关系,从而定量分析猪肉蛋白质及脂肪含量。实验结果表明,建立的SiPLS检测脂肪及蛋白质含量预测模型的最优组合分别为划分为20个光谱区间并联合4个子区间和9个主成分因子,和划分19个光谱区间并联合4个子区间和10个主成分因子。其预测集的相关系数分别为0.9798、0.9788,交互验证均方根误差分别为0.228,0.241。研究结果表明,利用近红外光谱结合SiPLS算法可以快速准确检测猪肉蛋白质与脂肪含量。Protein and fat are important nutrients of meat. With the development of people requirement of healthy diet, it becomes necessary to rapid detect protein and fat percentage composition. Spectroscopy spectrum data collected by near infrared spectroscopy are divided into calibration sample set and forecast sample set. Then, using MSC in MATLAB and average centralized methods for spectra pretreatment. The corresponding relation between the spectral data characteristics and meat protein and fat content which can quantitative analysis of the meat protein and fat content is got by SiPLS. The experimental results show that the optimal combination of the forecasting model established SiPLS detection of fat and protein content are divided into 20 respectively spectrum and joint 4 subsystems interval and 9 principal component factor, and divided into 19 spectral range and joint 4 subsystems interval and 10 principal component factors. The correlation coefficient of prediction set are 0.9798, 0.9788, RMSECV are 0.228,0.241. The research results show that using near infrared spectroscopy combined with SiPLS algorithm can accurately nondestructively test meat protein and fat content.

关 键 词：近红外光谱 联合区间偏最小二乘法 蛋白质及脂肪含量 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程]