[Application of PARAFAC method and 3-D fluorescence spectra in petroleum pollutant measurement and analysis].

A method for identification and concentration measurement of petroleum pollutant by combining three-dimensional (3-D) fluorescence spectra with parallel factor analysis (PARAFAC) was proposed. The main emphasis of research was the measurement of coexisting different kinds of petroleum. The CCl4 solutions of a 0# diesel sample, a 97# gasoline sample, and a kerosene sample were used as measurement objects. The condition of multiple petroleum coexistence was simulated by petroleum solutions with different mixed ratios. The character of PARAFAC in complex mixture coexisting system analysis was studied. The spectra of three kinds of solutions and the spectra of gasoline-diesel mixed samples, diesel-kerosene mixed samples, and gas oline-diesel mixed with small counts of kerosene interference samples were analyzed respectively. The core consistency diagnostic method and residual sum of squares method were applied to calculate the number of factors in PARAFAC. In gasoline-diesel experiment, gasoline or diesel can be identified and measured as a whole respectively by 2-factors parallel factors analysis. In diesel-kerosene experiment, 2-factors parallel factors analysis can only obtain the characters of diesel, and the 3rd factor is needed to separate the kerosene spectral character from the mixture spectrum. When small counts of kerosene exist in gasoline-diesel solution, gasoline and diesel still can be identified and measured as principal components by a 2-factors parallel factor analysis, and the effect of interference on qualitative analysis is not significant. The experiment verified that the PARAFAC method can obtain characteristic spectrum of each kind of petroleum, and the concentration of petroleum in solutions can be predicted simultaneously, with recoveries shown in the paper. The results showed the possibility of petroleum pollutant identification and concentration measurement based on the 3-D fluorescence spectra and PARAFAC.