Determinación cuantitativa de concentraciones de petróleo sumergido mediante espectroscopía de fluorescencia resuelta...

Submerged oil remains suspended in water bodies for extended periods, exhibiting high concealment and significant hazards. Accurate quantification of its concentration is crucial for formulating emergency response strategies. Accordingly, a comparative evaluation was carried out to examine the performance of four multi-way calibration strategies for the quantitative determination of submerged oil concentrations using time-resolved fluorescence spectroscopy (TRFS) data. The investigated approaches included parallel factor analysis/alternating penalty trilinear decomposition combined with partial least squares regression (PARAFAC/APTLD-PLSR), N-way partial least squares (N-PLS), as well as N-PLS combined with residual bilinearization (N-PLS/RBL). The results indicate that the N-PLS/RBL approach effectively leverages the ability of N-PLS to capture multidimensional spectral information while simultaneously benefiting from the second-order advantage introduced by the RBL algorithm. Consequently, this method exhibits superior prediction accuracy and greater robustness compared with the other calibration strategies. Among them, crude oil samples exhibited the most outstanding performance, with a residual predictive deviation (RPD) of 14.90 and a coefficient of determination of 0.99 for the validation set. In addition, the limits of detection (LOD) for the four oil types ranged from 0.15 to 1.03 mg·L^-1, indicating high sensitivity that meets the requirements for rapid on-site detection. Overall, this study provides a reliable methodological foundation and technical pathway for the development of low-cost, highly interference-resistant, and rapid on-site detection technologies for submerged oil pollution.