摘要
研究旨在构建肉羊浓缩饲料粗蛋白质含量的近红外光谱预测模型。研究中采集肉羊浓缩饲料样本100份,结合改进最小二乘回归法(MPLS),构建肉羊浓缩饲料粗蛋白质含量的近红外定量预测模型。模型建成后,通过内部验证和外部验证对模型进行评价。结果表明:肉羊浓缩饲料粗蛋白质含量近红外光谱预测模型的内部交叉验证标准误差(SECV)为0.1412,交叉验证相关系数(1-VR)为0.8953,外部交叉验证相对分析误差(RPDV)为3.01。表明近红外光谱预测模型可以准确预测肉羊浓缩饲料中粗蛋白质含量。
This study aimed to construct a near infrared reflectance spectroscopy(NIRS)prediction model for the crude protein content of mutton sheep concentrated feed.In this study,the prediction model of crude protein content in mutton sheep concentrated feed was established with samples of 100 different sources of mutton sheep concentrated feed by using NIRS and modified partial least-squares(MPLS)regression.The result showed that the standard error of cross validation(SECV)was 0.1412,1 minus the variance ratio(1-VR)was 0.8953,and ratio of performance to deviation of validation(RPDV)was 3.01.The experiments suggested that the crude protein content of mutton sheep concentrated feed was exactly predicted by prediction model of near infrared reflectance spectroscopy.
作者
纳嵘
朱海辉
NA Rong;ZHU Haihui(Ningxia Polytechnic University,Yinchuan 750021,China;Yinchuan Zegda Co.,Ltd.,Yinchuan 750001,China)
出处
《饲料博览》
2025年第2期9-12,共4页
Feed Review
基金
2024年宁夏回族自治区教育厅高等学校科学研究项目(NYG2024304)
2024年度宁夏回族自治区青年拔尖人才培养项目。
