一种油田原油含水率的近红外光谱检测与分析方法

石油作为重要战略资源，对其组分进行实时分析检测在石油化工领域有着重要意义。随着石油资源的不断开发，在已长时间开采油井的生产过程中以及新油井开采前，需要对井下原油组分进行分析检测，以判定开采的必要性。原油组分实时检测，在原油开采、生产、储运以及销售过程中都起着关键的作用，针对传统检测方法存在精度低、效率低等问题，近年来在原油组分检测技术的研究方法上引入了在测量领域得到广泛、有效应用的近红外光谱测量技术。以井下原油作为研究对象，利用从大庆油田获得的提纯原油与水按体积比配制了原油占比分别为1%~20%共39个组分的实验样品来模拟井下原油。研究了近红外光谱透射法测量原油组分的基本原理，并利用SW2520型近红外光谱仪与卤素光源以及配套组件集成了原油样品近红外光谱数据测量系统，完成了系统的标准化实验并利用此标准化后的系统采集了39个组分原油样品的近红外吸收光谱数据。利用移动窗口平滑法、Savitzky-Golay卷积平滑法以及Savitzky-Golay卷积求导法对原油近红外光谱数据进行预处理以消除噪声，利用偏最小二乘法和支持向量机回归方法SVR两种分析方法对预处理后的原油样品近红外光谱进行建模，分别建立了原油组分分析预测模型。研究结果表明，利用偏最小二乘法模型预测均方根误差为0.003 755 14，决定系数R2为0.999 999，预测精度优于0.1%，预测效果十分理想。利用近红外光谱技术对配制的不同比例的井下原油模拟测试样品进行测试建模和分析，为井下原油组份检测提供了一种新的思路，该方法可以有效的解决原油含水率的检测问题，为开发油田现场实时原油检测分析装备提供技术支撑。

Detection and Analysis of Water Content of Crude Oil by Near Infrared Spectroscopy

As an important strategic resource,petroleum has an important significance in real-time analysis and detection of its components in the petrochemical industry.With the continuous development of petroleum resources,in the production process of long-term oil wells and before the exploitation of new oil wells,it is necessary to analyze and test the components of underground crude oil to determine the necessity of mining.Real-time detection of crude oil components plays a key role in the process of crude oil exploitation,production,storage,transportation and sales.In view of the low accuracy and low efficiency of traditional detection methods,near-infrared spectroscopy,which has been widely and effectively applied in the field of measurement,has been introduced into the research methods of crude oil components detection technology in recent years.In this paper,the underground crude oil was used as the research object,and the experimental samples of 39 components with crude oil ratio of 1%~20%were prepared by using the purified crude oil and water obtained from Daqing Oilfield to simulate the underground crude oil.In this paper,the basic principle of the measurement of crude oil components by near-infrared spectroscopy is studied.The near-infrared spectroscopy data measurement system of crude oil samples is integrated with the halogen source and supporting components by SW2520 near-infrared spectroscopy,and 39 groups are collected by this measurement after system standardization testing.Moving window smoothing Savitzky-Golay convolution smoothing and Savitzky-Golay convolution derivation method of near infrared spectrum data of crude oil near infrared spectral data is used to remove noise,two analysis methods of near-infrared spectra of crude sample partial least squares and support vector machine(SVM)regression method are used for retreatment modeling,crude oil component analysis prediction model is established respectively.The results show that the RMS predicted by partial least square method is 0.00375514,and the coefficient of determination R 2 was 0.999999.The prediction effect is very good.In this study,the near infrared spectroscopy(NIRS)technology is used to test and analyze the simulated test samples with different proportions,which provides a new idea for the detection of the components of the downhole crude oil.This method effectively solves the detection problem of the water content of the crude oil and provides technical support for the development of the real-time crude oil detection and analysis equipment in the field.