Fast and simultaneous prediction of animal feed nutritive values using near infrared reflectance spectroscopy

International Conference on Agriculture, Environment, and Food Security

农业科学;生态环境科学

Samadi^1 ; Wajizah, S.^1 ; Munawar, A.A.^2

Department of Animal Science, Universitas Syiah Kuala, Banda Aceh, Indonesia^1

Department of Agricultural Engineering, Universitas Syiah Kuala, Banda Aceh, Indonesia^2

关键词: Animal production;    Coefficient correlations;    Leave-one-out cross-validation (LOOCV);    Near infrared reflectance spectroscopy;    Prediction performance;    Principal component regression;    Standard normal variates;    Wavelength ranges;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/122/1/012112/pdf
DOI  :  10.1088/1755-1315/122/1/012112

来源: IOP

【 摘 要 】

Feed plays an important factor in animal production. The purpose of this study is to apply NIRS method in determining feed values. NIRS spectra data were acquired for feed samples in wavelength range of 1000 - 2500 nm with 32 scans and 0.2 nm wavelength. Spectral data were corrected by de-trending (DT) and standard normal variate (SNV) methods. Prediction of in vitro dry matter digestibility (IVDMD) and in vitro organic matter digestibility (IVOMD) were established as model by using principal component regression (PCR) and validated using leave one out cross validation (LOOCV). Prediction performance was quantified using coefficient correlation (r) and residual predictive deviation (RPD) index. The results showed that IVDMD and IVOMD can be predicted by using SNV spectra data with r and RPD index: 0.93 and 2.78 for IVDMD ; 0.90 and 2.35 for IVOMD respectively. In conclusion, NIRS technique appears feasible to predict animal feed nutritive values.