Research On Quantitative Spectral Detection Method For Oil Content Of Settling Water Mixture In Oil Storage Tank

With the development of industrialization,the use of oil storage tanks is becoming increasingly widespread.Oil products are usually stored in storage tanks,which are then exported during use.The settling water mixture in the oil storage tank contains a large amount of oil substances,and causes pollution to the environment.Based on the investigation of the detection method of the oil content in the settling water mixture,in view of the low accuracy and precision of the traditional manual dehydration of crude oil,the quantitative analysis model of oil content has been successfully established by combining the spectral analysis method and the principle of chemometrics.This model can effectively solve the problems in the traditional crude oil dehydration process and provide a theoretical basis for online detection of oil content in oilfield reinjection water.The specific content is as follows:Firstly,the experimental preparation of the mixed solution of precipitation was completed,and UV and infrared spectra were scanned,collected,and analyzed.Infrared spectra with an optical path of 0.1mm and UV visible spectra with an optical path of 20mm were selected.Then,different sample set partitioning methods were compared and SPXY was selected as the partitioning method for the two datasets.Secondly,based on infrared spectroscopy,a prediction model for the oil content of the settling water mixture is established.The spectrum of 1250-4000cm^-1 was selected as the original spectrum and the oil content prediction model of precipitation mixture was established after preprocessing the spectral data and selecting the characteristic wavelength.The results showed that the infrared spectral modeling results of the oil content in the settling water mixture treated with Discrete Wavelet Transformation are the best.The root mean square error（RMSEP）of the test set and the correlation coefficient（R²）of the test set are 1.3411mg/L and0.9905,respectively.Through variable selection method,the optimal modeling combination was RF-PLSR（random frog partial least squares regression）RMSEP was 1.3745mg/L,R² was0.9910,which could accurately predict the oil content of the settling water mixture.Finally,based on UV spectrum,a prediction model for the oil content of the settling water mixture is established.The oil content prediction model of precipitation mixture was established after preprocessing the spectral data and selecting the characteristic wavelength.The results show that the infrared spectrum modeling results of Standard Normal Variate Transformation pretreatment of settling water mixture are the best.The root mean square error（RMSEP）of the test set and the correlation coefficient（R²）of the test set are 0.5383mg/L and0.9987,respectively.Through variable selection method,the optimal modeling combination was RF-PLSR（random frog partial least squares regression）RMSEP was 0.2566mg/L,R² was0.9997,which could accurately predict the oil content of the settling water mixture.By comparing the two spectral modeling methods,it is found that the prediction performance of infrared spectrum is better when the oil content is low（<10mg/L）,and the modeling effect of ultraviolet spectrum is better when the oil content is high.The experimental results show that the method can accurately detect the oil content in the mixed liquid of settling water of storage tank.The research results also lay a theoretical and technical foundation for the research and development of the dehydration control equipment of settling water of storage tank online,and provide a new technical means for the prevention and control of environmental pollution.