Rapid fertilizer nutrient analysis enabled by handheld LIBS (hLIBS) and an intelligent Euclidean distance prediction algorithm.

Conventional methods for detecting fertilizer nutrient content are time-consuming and labor-intensive, which limits rapid on-site analysis. This study presents a rapid method for the rapid detection of nitrogen (N), phosphorus (P), and potassium (K) in compound fertilizers using laser-induced breakdown spectroscopy (LIBS). Within the 190-950 nm wavelength range, the LIBS spectra exhibited intense, characteristic emission lines for N, P, and K with high signal-to-background ratios. A sorted feature matrix was constructed from the spectral data using a Euclidean distance algorithm and subsequently used to build a partial least squares regression (PLSR) model. The optimized PLSR model achieved rapid and accurate predictions, and its performance was evaluated using the coefficient of determination (R2), root mean square error (RMSE), and residual predictive deviation (RPD). The results were as follows: N (R2 = 0.9828, RMSE = 0.44 g kg-1, RPD = 7.46), P (R2 = 0.9541, RMSE = 1.02 g kg-1, RPD = 4.64), and K (R2 = 0.9610, RMSE = 0.90 g kg-1, RPD = 5.06), with all prediction errors below 6 %. This method provides an efficient and practical solution for real-time, on-site nutrient detection, demonstrating significant potential for application by fertilizer producers, regulators, and agricultural agencies to enhance nutrient use efficiency and support sustainable agricultural practices.