Least square support vector machine (LSSVM) combined with successive projection algorithm (SPA) method was applied for near-infrared (NIR) quantitative determination of the octane number in fuel petrol . T he NIR spectr a of 87 fuel petrol samples were scanned for model establishment and optimization. First order derivative Savitzky-Golay smoother (1 st - d’SG ) was utilized to improve the NIR predictive ability . Its pretreatment effect was compared with the raw data . SPA w as applied for the extraction of informative wavelengths . Considering the linear and non-linear training mechanism , LSSVM regression was employed to establish calibration models. The correlation coefficient (R) and root mean square error (RMSE) were used as the model evaluation indices; accordingly the octane number in fuel petrol was quantitatively determined with the prospective predictive indices. Results showed that after pretreated by 1 st - d’ SG , 8 SPA-selected wavelengths was generated as the inputs of LSSVM, so that the calibration models were optimized in the way of combining the SPA-LSSVM regression with the SG smoother. The prediction results were quite satisfactory, with t he calibrating correlation coefficient of 0.951 and the RMSE of 3.282 . An independent testing sample set was used to evaluate the optimal model, the testing correlation coefficient was 0.9 0 3 and the RMSE was 4.128 . We conclude that NIR spectro metr y is feasible to determine the octane in fuel petrol by establishing SPA-LSSVM models. The 1st- d’ SG pretreatment ha s the advantage of selecting wavelength s containing the implicit information . The combination of 1st- d’ SG pretreatment and SPA-LSSVM show its applicable potential to predict the octane number in fuel petrol.
Near Infrared (NIR) spectroscopy is a physical, rapid, non-destructive method to measure the combination and overtone vibration of chemical bonds in molecules, requiring minimal or no sample preparation and, in contrast with traditional chemical analysis, does not require reagents, nor produces wastes [
Petrol (or gasoline) is a petroleum-derived, transparent liquid primarily used as a fuel in internal combustion engines. It contains over 300 different chemical compounds, mainly hydrocarbons (compounds that contain only carbon and hydrogen) [
NIR spectroscopy can be used for quantitative determination of the octane in the fuel petrol because the spectral information of main structural components of fuel petrol is distinctly reflected in near-infrared region, and the information is quite stable. Brouillette and his coworkers proposed a method for detecting 22 properties (including the octane) in diesel, petrol, and jet fuels. They calculated the octane by NIR models according to the intensity of C=O peak [
Fuel petrol is a complex multi-component material. Its near-infrared spectrum contains information of all components as well as measurement noise. The prospective precise of calibration model will be difficult to improve if the full-range NIR spectrum is used. Wavelength selection is quite necessary for model optimization [
Calibration model establishment requires a set of spectra with reference concentrations of the target component. According to Beer-Lambert law, the NIR spectrum theoretically is the linear combination of the pure absorbance of every single component. However, the target is always one of all components, so that nonlinear methods should be employed if the linear model cannot meet the relationship between the spectra and the concentration. The least squares support vector machine (LSSVM) regression method was proved owning the capability to handle ill-posed problems and lead to unique global models [
Our study is concern with the wavelength selection for NIR quantitative determination of fuel petrol samples. For the modeling (calibration-validation) & testing sample division system, we randomly selected out a certain number of independent samples to be the test set, and the calibration-validation partitioning for the modeling set is achieved by the method of sample set partitioning based on joint x-y distances (SPXY) [
All the computational works in this study, such as the data pretreatment, model establishment and optimization, etc., were archived by MATLAB R2014a in a PC with a 8-core CPU and a 16 GB memory.
A total of 87 fuel petrol samples were collected. An oil filter (TY-II/1000) was dedicated to evaporate samples until fully dewatering. The dewatering samples were considered as pure petrol. The octane number of the pure petrol samples can be detected by using SKY2102-VII Gasoline Octane Number Tester (Shanghai Shenkai Petroleum Instrument Co. Ltd.). The 87 values range from 83.4 to 89.6. The arranged pure petrol samples were placed in an ultrasonic cleaning machine for oscillation beforehand.
NIR spectra of the 87 pure petrol samples were recorded on the NIR-Quest 256 spectrometer (Ocean Optics, USA) fitted with In GaAs detector and quartz-halogen light source. The scanning spectral range was set 800 - 2498 nm with 4 nm resolution, so that we have 850 wavelengths per spectrum. The experiment temperature was controlled at 25˚C ± 1 ˚C and the relative humidity was at 47% ± 1% RH throughout the scanning process. Each sample was measured thrice and the mean value was calculated for model establishment.
NIR analytical process requires a modeling-testing division for samples. And modeling optimization process should have the samples partitioned for calibration and validation. Firstly, 22 samples were randomly selected for testing, which were not subjected to the modeling process. The remaining 65 samples were used for modeling. We are planning to have 45 samples for calibration and 20 for validation. To obtain representative sample sets, the calibration-validation partition should be carried out based on both the absorbance values and the measured values from SKY2102-VII, so the SPXY method [
The NIR spectra of 87 pure petrol samples were showed in
According to the physical properties of the NIR absorption band of H-O and C=O, we utilized the 1st-derivative Savitzky-Golay smoother (1st-d’SG) for data pretreatment, and the pretreated spectra were showed in
The NIR spectral data after 1st-d’SG pretreatment were used for quantitative determination for the octane number in petrol samples. Based on the spectral in full scanning range, wavelength selection models were established for the calibration and validation samples by SPA method, so that the informative wavelengths can be selected. The selected wavelengths will be taken as the input variables of LSSVM regression and the NIR analytical models can be further established for optimization.
SPA was used to select wavelengths for NIR quantitative analysis of fuel petrol. The NIR data were beforehand pretreated by first derivative SG smoother (1st-d’SG). According to Balabin’s reports [
Next, we reset the number of variables changed from 4 to 9 and repeat wavelength selection by SPA. Then we have the best 8 variables (the most informative wavelengths) for the NIR analysis of the octane in fuel petrol. The specific 8 wavelengths were 1040, 1316, 1392, 1476, 1800, 1856, 1904, 2298 (the 8 circles in
NIR calibration models were established by using LSSVM regression with kernel function. Theradial basis function (RBF) kernel is common used in chemometrics [
The results in
For model evaluation, the 22 independent testing samples, not subjected to the modeling process, were used to examine the optimal SPA-LSSVM models with or without 1st-d’SG pretreatment. The predictive values of the 22 testing samples can be calculated respectively for raw data and the pretreated data, and the correlation charts were showed in
γ | σ | Calibration set | Validation set | |||
---|---|---|---|---|---|---|
R | RMSE | R | RMSE | |||
Raw spectra | 87.51 | 28.16 | 0.946 | 3.746 | 0.913 | 4.048 |
1st-d’SG | 89.34 | 32.45 | 0.951 | 3.282 | 0.927 | 3.841 |
Combined considering the modeling results showed in
NIR spectrometry was applied to quantitative determination of the octane number in fuel petrol samples. LSSVM models were established with its merit on linear and non-linear training mechanism, and the informative wavelengths were extracted by SPA method. Additionally, we noted that pretreatment is another way to improve the prediction results of calibration models.
The NIR spectral data keep the octane number going through the 1st-d’SG pretreatment. The main feature of octane in fuel petrol was reserved and obviously appeared at the peaks and troughs in the 1st-d’SG pretreated spectra. The noise interference was reduced and the pretreated data become much smooth. Additionally, SPA wavelength selection method was discussed respectively for the raw spectra and the 1st-d’SG data. Results showed that the SPA-selected wavelengths contained the main information of the full spectrum, and also the data dimension was effectively reduced. The 8 SPA-selected wavelengths all distribute at the peak or trough locations of the 1st-d’SG spectrum. It meant that the 1st-derivative peaks physically reflect the spectral information of octane.
LSSVM models were established for determination of the octane number. Results show that SPA-LSSVM modeling based on the 1st-d’SG pretreated data gave out high correlation coefficient and relatively low RMSE for validation samples. And, for further testing, the 22 independent samples evenly distributed on both sides of the regression line and the distance to the regression line is relatively closer. It meant that the best testing results was obtained by SPA-LSSVM model based on 1st-d’SG data.
The research was funded by the National Natural Scientific Foundation of China (No. 61505037), the Natural Scientific Foundation of Guangxi (No. 2015GXNSFBA139259, No. 2016GXNSFBA380077) and the Scientific Research Project of Guangxi Education Office (No. KY2015LX538).
The authors declare no conflicts of interest regarding the publication of this paper.
Xu, L.L., Gu, J., Chen, H.Z., Wen, J.B. and Xu, G.L. (2018) LSSVM Combined with SPA Applied to Near-Infrared Quantitative Determination of the Octane in Fuel Petrol Samples. Open Journal of Applied Sciences, 8, 422-430. https://doi.org/10.4236/ojapps.2018.89032