Quantitative Analysis of Nitrogen in Compound Fertilizers Using Laser-Induced Breakdown Spectroscopy Coupled with Multivariate Regression.

Quantitative analyses of the nitrogen concentration in compound fertilizer were performed using laser-induced breakdown spectroscopy (LIBS). Thirty-two samples were used as a calibration set, and eight samples were used as a validation set. To eliminate the matrix effect, partial least-squares regression (PLSR) and least-squares support vector regression (LS-SVR) were used to establish models. For the partial least-squares regression model, the correlation coefficients for the calibration and prediction sets were 0.837 and 0.794, respectively. While using the LS-SVR method, the correlation coefficients for the calibration and prediction sets were improved to 0.994 and 0.993, respectively. Therefore, the LS-SVR method improved the analysis accuracy. The prediction mean absolute error was 0.023%. The results indicate that LIBS coupled with LS-SVR is a reliable and accurate method for determining the nitrogen concentration in compound fertilizer. [ABSTRACT FROM AUTHOR]