Skip to Main Content
Near infrared spectroscopy combined with chemometrics was investigated to determine the total amino acids (TAA) in oilseed rape leaves. The samples in calibration, validation and prediction set were 80, 40 and 30, respectively. Different spectral preprocessing were compared, and three calibration methods were employed including partial least squares (PLS), multiple linear regression (MLR) and least squares-support vector machine (LS-SVM). The performance evaluation standards were determination coefficients (R2) and root mean square error (RMSE). Successive projections algorithm (SPA) was applied as variable selection method. The optimal model was achieved by SPA-LS-SVM using 13 relevant wavelengths with R2 = 0.9830 and RMSEP = 0.3964. The LS-SVM outperformed PLS and SPA-MLR models. The results indicated that near infrared spectroscopy was successfully applied for the determination of TAA in oilseed rape leaves. This detection method could be used for the on field monitoring of growing status and other physiological parameters of oilseed rape.