Position: Home > Articles > The Calibration Model Optimization of Gross Energy in Feed by Near Infrared Reflectance Spectroscopy(NIRS)
Journal of Agricultural Mechanization Research
2008
(8)
24-27
饲料总能近红外反射光谱定量分析模型的优化
作 者：
牛智有;韩鲁佳
单 位：
华中农业大学工程技术学院;中国农业大学工学院
关键词：
饲料总能;近红外反射光谱(NIRS);偏最小二乘(PLS);模型优化
摘 要：
利用近红外反射分析技术,采用偏最小二乘(PLS)回归方法,分别对光谱进行附加散射校正、变量标准化、一阶导数和二阶导数处理,建立了饲料中总能的预测模型。通过比较,附加散射校正和一阶导数处理定标效果最优。定标集化学分析值与预测值之间的决定系数R2和标准差RMSEC分别为0.9060和0.153,相对分析误差为3.36;验证集化学分析值与预测值之间的决定系数r2和标准差RMSEP分别为0.8924和0.156,相对分析误差为3.22。结果表明,利用近红外光谱反射(NIRS)分析技术可以定量检测充料总能的含量。
译 名：
The Calibration Model Optimization of Gross Energy in Feed by Near Infrared Reflectance Spectroscopy(NIRS)
作 者：
NIU Zhi-you1 , HAN Lu-jia2 (1.College of Engineering and Technology, Huazhong Agricultural University, Wuhan 430070,China; 2. Engineering College, China Agricultural University, Beijing 100083,China)
关键词：
feed gross energy; near infrared reflectance spectroscopy(NIRS); partial least squares(PLS); model optimization
摘 要：
The use of near Infrared reflectance spectroscopy(NIRS) was investigated as an alternative method for predicting gross energy in feed. The intial spectrum was pretreated by multiplicative scantter correction(MSC),standard norma lized variate(SNV),first derivative and second derivative ,respectively.The NIRS calibration models were developed using the partial least squares(PLS) technique. The methods of standard normalized variate(SNV) and first derivative were chosen as the best pretreatment. The coefficient of determination in calibration (R2 ) and standard error(RMSEC) of chemical analysis value and NIRS prediction value of calibration set were 0.9060 and 0.153.The coefficient of determination in calibration (r2 ) and standard error(RMSEP) of validation set were 0.892 4 and 0.156,and the RPD were all over 3. It was concluded that NIRS can be used as a method to monitor the chemocal composition and gross energy of feed for ruminant, and a good measuring precision could be expected.
导出引用
计量
文章访问数: 14
HTML全文浏览量: 0
PDF下载量: 0
所属期刊
