Recognition Of Oil Spill Emulsification And Oil Film State Based On LIF

The Oil spill pollution on the sea is a kind of marine pollution with great harm,which seriously affects the survival of marine organisms and the development of human beings.The key to control oil spill pollution is to accurately determine the type of oil spill pollution.There are two types of spilled oil into the sea area: non emulsification and emulsification.The former exists in the form of oil film,and the latter in the form of emulsion.Based on the technology of laser-induced fluorescence(LIF)to detect oil spills on the sea,this paper takes 0 #diesel oil,aviation kerosene and shell red shell 15w40 lubricating oil as examples to design a recognition method to distinguish the emulsification state and oil film state of different oils.LIF spectrum measurement system was set up in the laboratory to prepare oil film with different thickness and emulsion with different water content of diesel oil,kerosene and lubricating oil.Fluorescence spectra of diesel oil emulsion,diesel oil film,kerosene emulsion,kerosene oil film,lubricating oil emulsion and lubricating oil film were collected.Principal component analysis(PCA)and feature parameter extraction were used to extract the feature of fluorescence spectrum data,and then the least square discriminant analysis(PLS-DA)and BP neural network were used for classification and recognition.In this paper,two recognition methods are used: direct recognition and oil classification + type recognition.In the direct recognition method,PCA +PLS-DA,feature parameter + BP neural network and feature parameter + PLS-DA are used to recognize six kinds of fluorescence spectra.Experiments show that these three recognition methods can recognize six kinds of fluorescence spectra well.In the oil classification + type recognition,PCA + PLS-DA,feature parameter + BP neural network and feature parameter + PLS-DA are used to recognize six kinds of fluorescence spectra The experimental results show that the recognition rate of feature parameter + BP neural network method is higher than that of feature parameter +PLS-DA method,and the recognition effect is the best when selecting the lubricating oil emulsification and lubricating oil film as the training set,and diesel oil emulsification,diesel oil film,kerosene emulsification and kerosene film as the test set.