Fluorescence Spectroscopy Combined With Feature Extraction And Decomposition Algorithm For Detection Of Peteroleum Type Oil

Petroleum is an important fossil resource,which provides important energy and power for the development of human society.However,there are leakage and pollution in the process of oil exploitation and use,resulting in property loss,waste of resources,environmental pollution and so on.Therefore,the accurate detection and analysis of petroleum substances is of great significance for the protection of the environment,the recovery and utilization of resources.This paper is based on the fluorescence spectrum analysis technology,and combined with machine learning method and multi-dimensional decomposition algorithm to study petroleum substances.The specific research content is divided into the following points:Firstly,explain the fluorescence generation mechanism of substances,study the luminescence process of fluorescent materials,and the principle and reason of scattering.In order to reveal the advantages and disadvantages of the two fluorescence spectroscopy technologies,the similarities and differences between three-dimensional fluorescence spectroscopy technology and total synchronous fluorescence spectroscopy technology are compared.Explain the system composition,working principle and working process of the experimental instrument,which can provide a basis for obtaining fluorescence spectrum data.Secondly,taking diesel,gasoline,jet fuel and lubricating oil as the research objects,the identification and detection model of pure oil has been established.Combining threefluorescence spectroscopy with machine learning method,an oil detection system is established to realize the purpose of different types of oil discrimination.Firstly,the spectral data are collected by experiments and are preprocessed.Then,the spectral features are extracted by feature extraction method.Finally,the petroleum samples were classified.The experimental results show that by extracting the characteristics of the sample spectral data,the purpose of distinguishing different types of oil can be realized,which provides a good method and idea for the detection technology of oil.Then,the three-dimensional fluorescence spectra of jet fuel and lubricating oil mixed oil were analyzed.By introducing different salinity as a dimension factor,a mixed oil detection model based on three-dimensional fluorescence spectroscopy was established.The main steps are: design a mixed oil spectrum experiment,preprocess the obtained oil spectral data,pre-estimate the composition of the mixed oil solution,analysis the spectrum data.The experimental results show that this method can analyze the mixed oil samples with the participation of environmental factors,and realize the purpose of qualitative analysis and quantitative detection of mixed oil samples.At the same time,it provides a new reference method for the detection of mixed oil solution under complex conditions.Finally,In order to characterize the spectral information of oil more comprehensively,aviation kerosene and lubricating oil are taken as the research object,and the spectrum of oil is explored from the angle of different spectral techniques,and the mixed oil detection model under different spectral techniques is established.The three-dimensional fluorescence spectrum data and the total synchronous fluorescence spectrum data of petroleum samples are analyzed by multivariate curve resolution-alternate least square algorithm,and the results are compared.The experimental results show that the analytical results obtained by total synchronous fluorescence spectrum are better than that of threedimensional fluorescence spectrum,which accurately predict the concentration of each component in the mixed oil sample,and analyze the components in the mixed oil.This method expands the detection method of mixed oil and provides a new technique and method for oil detection.