Rapid Determination of Cetane Value of Diesel Oil by Raman Spectroscopy Combined with Partial Least Square Method

A method for the determination of cetane value of diesel oil by Raman spectroscopy combined with partial least square (PLS) was established. 124 Raman spectra of diesel oil samples with different cetane values were collected. The amount of calibration set and prediction set, the spectral pretreatment method, the number of principal factor and the spectral region were optimized to accomplish PLS model. The optimized modeling conditions were as follows. The quantity of calibration set and prediction set was 103 and 21, respectively. The spectra was preprocessed by smoothing convolution and standard normal variable change.The number of principal factor was 7. The spectral regions were 700-1 700 cm^-1 and 2 700-3 100 cm^-1. The results showed that the root mean square error of correction of the PLS model was 0.913, the correlation coefficient was 0.954 3, the root mean square error of prediction of model was 0.911 and the prediction coefficient was 0.924 6. This model was used to analyze diesel oil, and the results of PLS model were consistent with the results of standard engine method, with absolute error in the range of -2.9-2.9.