On-line Analysis And Control For Blended Gasoline Octane Number Based On Raman Spectral Decomposition

Gasoline pipeline blending technology has been widely used in the refining process.The online analysis and control of blended gasoline is directly related to the quality of the gasoline and the economic benefits of the refinery.As a new molecular spectrum,online Raman analysis technology has the advantages of short detection period,perfect characteristics,and low on-site maintenance cost.However,due to the complex composition of gasoline,its Raman spectra are seriously overlapping.Therefore,this research studies the signal separation of Raman spectrum of gasoline.The main research contents include:(1)Combined with the similarity of the online detection spectrum of blended gasoline,a fast Raman spectroscopy algorithm is designed,which greatly reduces the running time of the algorithm without losing analytical accuracy;the correlation coefficient is used as a similar indicator,Raman feature analysis is performed on the oil sample set,and Raman feature components are classified.(2)To improve the accuracy of the quantitative analysis of research octane number(RON),Raman feature analysis is combined with the physical meaning of the spectral components to select the feature components.A hybrid model is designed combined with multivariate linear regression(MLR)and partial least squares(PLS).The number of principal components required for the hybrid model is small;while the capacity of the sample set changes,the performance of the hybrid model is more stable,and the accuracy is also better than the conventional PLS model.The on-site commissioning results show that the online Raman analyzer embedded in the above-mentioned hybrid model is operating well.Compared with the manual analysis data,the mean square error of the hybrid model is 0.26.(3)To improve the performance of the control system of RON,the work proposes a control system based on a nonlinear predictor.The nonlinear predictor is designed based on the Smith predictor,according to the gasoline blending process mechanism,and the selection of the parameter of the nonlinear predictor is discussed in detail.Simulation results show that the performance of our nonlinear predictor is better than that of the classical PID control system.