Detection of components of aromatics hydrocarbons unit based on Raman spectrometer

In order to enhance the real-time and improve the accuracy of on-line Raman spectrometer during the testing of composition of aromatic hydrocarbon unit, the prediction model was created based on partial least square (PLS) algorithm and particle swarm optimization (PSO) algorithm. Some samples of aromatic hydrocarbons were tested. Firstly, Raman spectroscopy of aromatic composition was gotten by spectroscopy. Then, the main factors of Raman data were extracted by means of PLS algorithm in order to reduce the redundancy between data. The quick search of composition content of aromatics hydrocarbons were made by PSO algorithm to find the optimal solution. Finally, the correlation of actual values and predictive values of samples was analyzed. The results show that, compared with the old method, the new created model (Raman spectrum with PSO and PLS) has high precision and quick analysis speed. It provides a new method for detection of components of aromatic hydrocarbons unit.