阅读说明：本技术 一种由三维荧光光谱预测岩屑萃取液中原油浓度的方法 (Method for predicting crude oil concentration in rock debris extraction liquid by three-dimensional fluorescence spectrum ) 是由 陈瀑 许育鹏 李敬岩 褚小立 于 2018-06-29 设计创作，主要内容包括：本发明涉及一种由三维荧光光谱预测岩屑萃取液中原油浓度的方法,本发明方法采集不同浓度的原油溶液标样,测定三维荧光光谱,再选取三维荧光光谱特征区间中的二维发射光谱首尾相连,得到一组荧光发射强度二维谱,将荧光发射强度二维谱进行归一化并添加随机光谱噪声消除多重共线性后组成校正集,然后将该校正集通过偏最小二乘法建立分析模型,采用该分析模型能够预测岩屑萃取液中原油浓度。(The invention relates to a method for predicting crude oil concentration in a rock debris extraction liquid by using a three-dimensional fluorescence spectrum, which comprises the steps of collecting crude oil solution standard samples with different concentrations, measuring the three-dimensional fluorescence spectrum, selecting two-dimensional emission spectra in a characteristic interval of the three-dimensional fluorescence spectrum, connecting the two-dimensional emission spectra end to obtain a group of fluorescence emission intensity two-dimensional spectra, normalizing the fluorescence emission intensity two-dimensional spectra, adding random spectral noise to eliminate multiple collinearity to form a correction set, establishing an analysis model for the correction set by using a partial least square method, and predicting the crude oil concentration in the rock debris extraction liquid by using the analysis model.)

1. A method for predicting crude oil concentration in a rock debris extract from three-dimensional fluorescence spectroscopy, the method comprising:

(1) preparing a plurality of crude oil solution standard samples to ensure that the crude oil concentration is different; wherein the crude oil concentration in the standard sample does not exceed the range in which the three-dimensional fluorescence spectrum and the crude oil concentration are linearly related;

(2) collecting a three-dimensional fluorescence spectrum of each standard sample, wherein three-dimensional coordinates of the three-dimensional fluorescence spectrum are respectively emission wavelength, excitation wavelength and fluorescence emission intensity;

(3) selecting a characteristic interval of emission wavelength in the three-dimensional fluorescence spectrum of each standard sample, and acquiring the fluorescence emission intensity of the characteristic interval of the emission wavelength under each excitation wavelength to obtain a series of two-dimensional emission spectra under different excitation wavelengths; according to the ascending sequence of the excitation wavelengths, connecting the two-dimensional emission spectrums corresponding to each excitation wavelength end to obtain a group of fluorescence emission intensity two-dimensional spectrums; wherein the characteristic interval covers the strongest characteristic peak of the crude oil in each two-dimensional emission spectrum;

(4) dividing the fluorescence emission intensity two-dimensional spectrum of each standard sample by the crude oil concentration of the standard sample, then adding, and taking the sum and averaging to obtain a concentration element two-dimensional spectrum of the crude oil solution;

(5) multiplying the fluorescence emission intensity of the concentration element two-dimensional spectrum by a coefficient k respectively, wherein k is the concentration value of the crude oil solution to obtain derivative fluorescence emission intensity two-dimensional spectra of the crude oil solutions with different concentrations, and the maximum value of k is not more than the linear correlation range of the concentration of the crude oil solution measured by the three-dimensional fluorescence spectrum;

(6) adding random spectral noise to the derived fluorescence emission intensity two-dimensional spectrum of each group of crude oil solutions with different concentrations to obtain a simulated fluorescence emission intensity two-dimensional spectrum, and forming a correction set by the simulated fluorescence emission intensity two-dimensional spectrum and the corresponding crude oil solution concentration;

(7) correlating the simulated fluorescence emission intensity two-dimensional spectrums of the crude oil solutions with different concentrations in the correction set with the crude oil solution concentration corresponding to the two-dimensional spectrums by a Partial Least Squares (PLS) method to establish an analysis model;

(8) and (3) obtaining a two-dimensional spectrum of the fluorescence emission intensity of the rock debris extract liquid sample to be detected according to the methods in the steps (2) to (3), and substituting the two-dimensional spectrum into the analysis model established in the step (7) to obtain the concentration of the crude oil in the rock debris extract liquid sample to be detected.

2. The method of claim 1, wherein in step (1), the solvent in the crude oil solution standard is n-hexane or n-heptane.

3. The method according to claim 1, wherein in the step (2), a fluorescence spectrophotometer is adopted to collect the three-dimensional fluorescence spectrum, the excitation wavelength range is 250-550 nm, and the emission wavelength range is 250-550 nm.

4. The method according to claim 1, wherein in the step (3), the characteristic interval of the emission wavelengths is an emission wavelength range in which the emission wavelength λ at the strongest characteristic peak of the two-dimensional emission spectrum at a certain excitation wavelength extends to at least 60-85nm towards both sides of the emission spectrum.

5. The method according to claim 1, wherein in the step (3), the obtained two-dimensional spectrum of fluorescence emission intensity is smoothed.

6. The method according to claim 1, wherein in the step (5), the coefficient k is obtained by increasing the concentration value of crude oil in the extraction liquid by unit concentration, and the unit concentration is 1 mg/L.

7. The method of claim 1, wherein in (6), the method of adding random spectral noise comprises: and (3) repeatedly acquiring a three-dimensional fluorescence spectrum of a certain crude oil extract standard sample in the step (1), acquiring two fluorescence emission intensity two-dimensional spectrums according to the method in the step (3), subtracting the two fluorescence emission intensity two-dimensional spectrums to obtain a difference spectrum F, wherein the random spectrum noise Z is F multiplied by M, M is a random numerical value in 0.5-1.2, and adding the simulated fluorescence emission intensity two-dimensional spectrum of each crude oil with different crude oil concentration and the random spectrum noise to obtain the simulated fluorescence emission intensity two-dimensional spectrum of the correction set sample.

8. The method according to claim 1, wherein in step (7), the spectra in the calibration set are subjected to second order differential processing and mean centering, and in step (8), the two-dimensional fluorescence emission intensity spectrum of the obtained sample to be tested is subjected to second order differential processing and mean centering.

Technical Field

The invention relates to a method for rapidly predicting the concentration of crude oil in a rock debris extraction liquid by using a three-dimensional fluorescence spectrum, in particular to a method for predicting the concentration of crude oil in the rock debris extraction liquid by using a three-dimensional quantitative fluorescence spectrum.

Background

Aromatic compounds are important components of crude oil, generally account for 10% -45% of total hydrocarbons, are relatively stable, and are not susceptible to biodegradation. The fluorescence analysis method is one of the common methods for aromatic hydrocarbon analysis, the concentration, the weight and other properties of the crude oil can be known by analyzing the components and the compositions of aromatic hydrocarbon in the crude oil by using fluorescence spectroscopy, and the generated fluorescence logging method has the advantages of high sensitivity and high speed and is very suitable for finding trace oil gas hidden in drilling cuttings.

The traditional darkroom UV irradiation method is easy to leak light oil layer, and chloroform or carbon tetrachloride solvent is not ideal fluorescent reagent because of toxicity or quenching effect. The subsequent one-dimensional and two-dimensional quantitative fluorescence logging technology can determine the oil content of the rock debris extracting solution by measuring with a standard crude oil sample to make a correction curve, thereby realizing the conversion from qualitative interpretation to quantitative interpretation of the fluorescence logging. However, due to the limited amount of spectral data presented, various oil compositions cannot be optimally identified, and the quantitative data is limited and the accuracy is not high.

The fluorescent stereogram and the fingerprint provided by the three-dimensional quantitative fluorescence analysis technology can reflect the full appearance of the fluorescent substance comprehensively, and further derive more evaluation indexes to evaluate various properties of the crude oil more comprehensively, so that the oil-gas identification is more accurate and reliable. However, the crude oil quantitative means in the current three-dimensional fluorescence technology is similar to that in the past, the quantitative analysis is still carried out through the single-point intensity of the crude oil characteristic peak, all information of the three-dimensional fluorescence spectrum is not utilized, and the accuracy is not high.

CN201010580466.5 provides a three-dimensional quantitative fluorescence spectrum total volume integration method capable of measuring the concentration of a single-component fluorescent substance and quantitatively explaining and distinguishing the content of various hydrocarbon substances in an oil-gas layer. According to the Monte-Carlo principle, the method randomly generates random numbers distributed in a three-dimensional space by a computer, further calculates the volume integral value of a sample spectrum, and finally calculates the concentration of the hydrocarbon sample according to the value.

CN200510002087.7 entitled "method for measuring oil content in core of rock debris in oil logging" provides a new method for quantitative analysis of crude oil by three-dimensional synchronous fluorescence spectrum generated by mathematical interpolation. Firstly, a continuous crude oil three-dimensional concentration synchronous fluorescence spectrum library is established through a standard sample, and the crude oil concentration to be detected can be obtained by comparing the synchronous spectrum of a certain sample to be detected with the spectrum data of the spectrum library by the least square method. The method essentially utilizes simultaneous fluorescence rather than three-dimensional fluorescence.

Disclosure of Invention

The invention aims to provide a method for predicting the concentration of crude oil in a rock debris extraction liquid by using a three-dimensional fluorescence spectrum.

In order to achieve the above object, the present invention provides a method for predicting crude oil concentration in a debris extract from a three-dimensional fluorescence spectrum, the method comprising:

(1) preparing a plurality of crude oil solution standard samples to ensure that the crude oil concentration is different; wherein the crude oil concentration in the standard sample does not exceed the range in which the three-dimensional fluorescence spectrum and the crude oil concentration are linearly related;

(2) collecting a three-dimensional fluorescence spectrum of each standard sample, wherein three-dimensional coordinates of the three-dimensional fluorescence spectrum are respectively emission wavelength, excitation wavelength and fluorescence emission intensity;

(3) selecting a characteristic interval of emission wavelength in the three-dimensional fluorescence spectrum of each standard sample, and acquiring the fluorescence emission intensity of the characteristic interval of the emission wavelength under each excitation wavelength to obtain a series of two-dimensional emission spectra under different excitation wavelengths; according to the ascending sequence of the excitation wavelengths, connecting the two-dimensional emission spectrums corresponding to each excitation wavelength end to obtain a group of fluorescence emission intensity two-dimensional spectrums; wherein the characteristic interval covers the strongest characteristic peak of the crude oil in each two-dimensional emission spectrum;

(4) dividing the fluorescence emission intensity two-dimensional spectrum of each standard sample by the crude oil concentration of the standard sample, then adding, and taking the sum and averaging to obtain a concentration element two-dimensional spectrum of the crude oil solution;

(5) multiplying the fluorescence emission intensity of the concentration element two-dimensional spectrum by a coefficient k respectively, wherein k is the concentration value of the crude oil solution to obtain derivative fluorescence emission intensity two-dimensional spectra of the crude oil solutions with different concentrations, and the maximum value of k is not more than the linear correlation range of the concentration of the crude oil solution measured by the three-dimensional fluorescence spectrum;

(6) adding random spectral noise to the derived fluorescence emission intensity two-dimensional spectrum of each group of crude oil solutions with different concentrations to obtain a simulated fluorescence emission intensity two-dimensional spectrum, and forming a correction set by the simulated fluorescence emission intensity two-dimensional spectrum and the corresponding crude oil solution concentration;

(7) correlating the simulated fluorescence emission intensity two-dimensional spectrums of the crude oil solutions with different concentrations in the correction set with the crude oil solution concentration corresponding to the two-dimensional spectrums by a Partial Least Squares (PLS) method to establish an analysis model;

(8) and (3) obtaining a two-dimensional spectrum of the fluorescence emission intensity of the rock debris extract liquid sample to be detected according to the methods in the steps (2) to (3), and substituting the two-dimensional spectrum into the analysis model established in the step (7) to obtain the concentration of the crude oil in the rock debris extract liquid sample to be detected.

According to the method, almost all useful information in the three-dimensional fluorescence spectrum is utilized by intercepting the three-dimensional fluorescence spectrum interval of the crude oil, the calibration set of crude oil solutions with different concentrations is established after concentration normalization, the crude oil solution concentration and the derived fluorescence emission intensity two-dimensional spectrum are correlated by using a PLS method to establish an analysis model, and the method has high prediction accuracy.

Drawings

FIG. 1 is a three-dimensional fluorescence spectrum (with solvent background subtracted) of a standard solution of crude oil A at a concentration of 10mg/L in example 1 of the present invention, in which the X-axis represents the emission wavelength Em, the Y-axis represents the excitation wavelength Ex, and the Z-axis represents the fluorescence emission intensity; the region (including the maximum characteristic peak lambda) between two lines parallel to the Y-axis in the figure is the selected characteristic interval. The two-dimensional emission spectra under different excitation wavelengths in the interval are connected end to form the fluorescence emission intensity two-dimensional spectrum of the standard sample solution.

FIG. 2 is a two-dimensional spectrum (emission wavelength interval of 301-450 nm) of crude oil solution concentration elements obtained in example 1 of the present invention, wherein the abscissa represents the number of emission wavelength points, and since the emission wavelength step diameter is 1nm, the interval of 301-450 nm contains 150 emission wavelength points, i.e., the emission spectrum at each excitation wavelength Ex contains 150 wavelength points, and the number of excitation wavelength points is 31, so that the graph contains 31 × 150 to 4650 emission wavelength points in total. The ordinate is the sum and average fluorescence emission intensity.

Detailed Description

The method comprises the steps of collecting crude oil solution standard samples with different concentrations, measuring a three-dimensional fluorescence spectrum, selecting fluorescence emission intensity of a characteristic interval of the emission spectrum under each excitation spectrum of the three-dimensional fluorescence spectrum to obtain a series of two-dimensional emission spectra, connecting the two-dimensional emission spectra end to obtain a group of two-dimensional fluorescence emission intensity spectra, and adding and averaging the two-dimensional fluorescence emission intensity spectra of the solutions with different crude oil concentrations to obtain a concentration element two-dimensional spectrum of the crude oil solution. Multiplying the elementary two-dimensional spectrum by different coefficients k to derive crude oil solution spectra with different concentrations, adding random spectral noise to eliminate multiple collinearity to form a correction set, correlating the spectra of the correction set with the corresponding crude oil solution concentration by a partial least square method, and establishing an analysis model which can accurately predict the crude oil concentration in the rock debris extraction liquid.

(1) In the step, the solvent in the crude oil solution standard sample can be n-hexane or n-heptane, preferably n-heptane, the number of the standard samples can be multiple, preferably 3-6, and the concentration of the standard samples can be freely prepared in a range in which the three-dimensional fluorescence spectrum and the crude oil concentration are linearly related, preferably in a gradient manner.

(2) In the step, a fluorescence spectrophotometer is adopted to collect the three-dimensional fluorescence spectrum, the excitation wavelength range is preferably 250-550 nm, and the emission wavelength range is preferably 250-550 nm.

(3) In the step, the emission wavelength characteristic interval is preferably an emission wavelength range in which the emission wavelength λ at the strongest characteristic peak of the two-dimensional emission spectrum at a certain excitation wavelength extends by 60 to 85nm to both sides of the emission spectrum.

(3) In the step (4), the obtained fluorescence emission intensity two-dimensional spectrum is preferably smoothed, and then the concentration elementary two-dimensional spectrum of the crude oil solution is obtained by calculation, wherein the smoothing method can adopt segmented Gaussian smoothing.

(5) In the step, the value of the coefficient k is preferably obtained by increasing the concentration value of the crude oil solution by unit concentration, and the unit concentration (c) is preferably 1 mg/L. Under the condition that c is 1mg/L, the first group of derived fluorescence emission intensity two-dimensional spectrum variables S₁(S x 1) corresponding to a concentration of 1mg/L, S is a concentration-element two-dimensional spectrum, and a second set of derived fluorescence emission intensity two-dimensional spectrum variables S₂The concentration upper limit does not exceed the linear correlation range of the concentration of the crude oil solution measured by the three-dimensional fluorescence spectrum, if the concentration upper limit of the linear correlation range is 100mg/L, the final group of derived fluorescence emission intensity two-dimensional spectrum variables S₁₀₀A total of 100 sets of derived fluorescence emission intensity two-dimensional spectral variations were modeled at corresponding concentrations of 100 mg/L-sx 100.

(6) In step (ii), the step of adding random spectral noise may include: and (3) repeatedly acquiring the three-dimensional fluorescence spectrum of a certain crude oil extraction liquid standard sample in the step (1), acquiring two groups of fluorescence emission intensity two-dimensional spectrums according to the method in the step (3), subtracting the two to obtain a difference spectrum F, wherein the random spectrum noise Z is F multiplied by M, M is a random numerical value in 0.5-1.2, and adding the simulated fluorescence emission intensity two-dimensional spectrum of each crude oil with different crude oil concentrations and the random spectrum noise to obtain the simulated fluorescence emission intensity two-dimensional spectrum of the correction set sample. For example, the two-dimensional spectrum variation of the simulated fluorescence emission intensity after adding the spectral noise is as follows: s₁＝S×1+F×M₁，S₂＝S×2+F×M₂，…，S_m＝S×m+F×M₁₀₀Wherein M is₁～M_mThe random number is 0.5-1.2, F is a difference spectrum, and m is the number of simulated fluorescence emission intensity two-dimensional spectra. And finally, forming a correction set by the m simulated fluorescence emission intensity two-dimensional spectrums after random spectral noise is added and the crude oil solution concentration corresponding to the m simulated fluorescence emission intensity two-dimensional spectrums.

(7) In step (a), the spectrum in the calibration set is preferably subjected to second order differential processing and mean centering processing before being used for establishing an analysis model (also referred to as a calibration model). The mean-centering process is to subtract the mean value of the column of the sample from the data of each element constituting the spectral vector of the sample, and here, the method for performing the mean-centering process on each sample in the correction set is as follows: and subtracting the mean simulated fluorescence emission intensity two-dimensional spectrum of the calibration set from the simulated fluorescence emission intensity two-dimensional spectrum of the sample.

(8) And in the step, the obtained fluorescence emission intensity two-dimensional spectrum of the sample to be detected is subjected to second-order differential processing and mean centering processing, wherein the mean centering processing mode is that the fluorescence emission intensity two-dimensional spectrum of the sample to be detected is subtracted by the mean simulated fluorescence emission intensity two-dimensional spectrum of the correction set.

The process of the present invention for establishing a calibration model using partial least squares is described below:

when the PLS method is adopted to establish the correction model, the modeling is based on the Law of Lambert-beer:

Y＝XB+E，

in the formula (I), the compound is shown in the specification,

y-a matrix (m x n) consisting of m samples, the fluorescence emission intensity at n emission wavelength points;

x-a concentration vector (m.times.1) consisting of m samples, one component (crude oil solution concentration);

b-a (1 × n) sensitivity vector consisting of one component, n number of wavelength points;

e-m × n residual matrix.

The number of the emission wavelength points is the number of information recorded by the spectrometer in a set spectrum interval.

The general algorithm for establishing the correction model is as follows:

1. and (3) carrying out mean centering treatment on each element of the fluorescence emission intensity matrix Y (obtained by carrying out second-order differential treatment on the fluorescence emission intensity in the characteristic spectrum region) and the concentration vector X, namely subtracting the mean value of the row of each element data forming the sample spectrum vector from the data of each element.

2. And (3) decomposing the main components of the processed fluorescence emission intensity matrix Y and the processed concentration vector X according to the formulas (1) and (2):

wherein:

t_k(m × 1) -factor score, y-score, which is the fluorescence emission intensity matrix;

v_k(1 xn) -is the factor load, y-loading, of the fluorescence emission intensity matrix;

r_k(m × 1) -factor score for concentration vector, x-score;

q_k(1 × 1) — is a number, the factor load of the concentration vector, x-loading;

p-number of major factors;

E_Y-fluorescence emission intensity matrix residual error;

E_X-concentration vector residual.

In order to ensure that T derived from Y has a good linear relationship with R derived from X, information about R may be introduced when Y is decomposed into T, or information about T may be introduced when X is decomposed into R, which may be achieved by iteratively exchanging the iteration variables, i.e. combining the two decomposition processes into one, namely:

r_k＝b_kt_k (3)

b_k(1×1)—r_kand t_kThe regression coefficient of (2).

3. Solving the feature vector and the number p of main factors

Ignoring the residual matrix E, if p is 1 according to equations (1) and (2):

Y＝tv^t

left multiplication by t^tObtaining: t is t^tY＝t^ttv^tI.e. v^t＝t^tY/t^tt

Right-multiplying v to obtain: yv ═ tv^tv, i.e. t ═ Yv/v^tv (4)

X＝rq

Left multiplying r^tObtaining: r is^tX＝r^trq, i.e. q ═ r^tX/r^tr，

And q is divided on both sides: and r is X/q.

[1] The weight vector w of the fluorescence emission intensity matrix is solved,

taking a certain column of the concentration array X as an initial iteration value of r, replacing t with r, and calculating a w equation: y ═ rw^tThe solution of (a) is: t is Yw/w^tw

[2]Normalizing w:

[3]calculating the factor score t of the fluorescence emission intensity matrix, and calculating a t equation according to the normalized w: tw ═ Y^tThe solution of (a) is: t is Yw/w^tw

[4]And (3) solving the weight u value of the concentration vector, and replacing r with t to calculate a u equation: the solution for X ═ tu is: t ═ u^tX/t^tt

[5] And (3) obtaining a factor score r, x-score of the concentration vector, and calculating an r equation by using u: the solution for X ═ ru is: r ═ X/u

Then r replaces t and returns to the [1]]Calculating w from wt, calculating ttnew, and repeating the iteration until t converges (II t)_New-t_{Old age}‖≤10^-6‖t_New|), continue with the next operation, otherwise return to step [1]]。

[6]Solving a load vector v, y-loading equation of the fluorescence emission intensity matrix according to the converged t: tv ═ Y^tThe solution of (a) is: v. of^t＝t^tY/t^tt

[7]Solving the load q value of the concentration vector by r, and obtaining an x-loading equation: the solution for X ═ rq is: q ═ r^tX/r^tr

From this r corresponding to the first main factor is determined₁，q₁，t₁，

Substituting into equation (3) to obtain b₁：

[8] Calculating residual error matrix E by formula (1) and (2)

E_X,1＝X-r₁q₁＝X-b₁t₁q₁ (5)

E_Y,1＝Y-t₁v (6)

[9]With E_X,1In place of X, E_Y,1Instead of Y, return to step [1]]Calculating the next component

r₂，q₂，t₂，

b₂

[10]Calculating E according to equations (5) and (6)_X,2、E_Y,2By analogy, X, Y are determined for all main factors.

[11] The number of major factors was determined by cross-checking.

Through the above processes, the establishment of the correction model is completed.

Performing second-order differential processing on a fluorescence emission intensity two-dimensional spectrum obtained after three-dimensional fluorescence spectrum processing of an unknown crude oil solution sample, and then performing mean centralization processing which is the same as modeling to obtain fluorescence emission intensity Y of a sample to be detected_{To be measured}Predicting the oil concentration X from the calibration pattern_{To be measured}The method comprises the following steps:

[1]from Y_{To be measured}And v stored during correction_kSubstituting (4) to calculate t_{k (to be measured)}

[2]From the found t_{k (to be measured)}And b stored during correction_kCalculation of r by substitution formula (3)_{k (to be measured)}

r_{k (to be measured)}＝b_kt_{k (to be measured)}

[3]From the found r_{k (to be measured)}And q stored during the correction_kCalculating X by using the formula (2)_{To be measured}。

The present invention is further illustrated by the following examples, but the present invention is not limited thereto.

The instrument for measuring the three-dimensional fluorescence spectrum of the sample is an F97 fluorescence spectrophotometer produced by Shanghai prism technology Limited, a sample cell is a quartz cuvette with an optical path of 1cm, and the spectrum collection conditions are as follows: the range of the excitation wavelength is 250-550 nm, and the step diameter is 10 nm; the emission wavelength range is 250-550 nm, and the step diameter is 1 nm; the width of each slit is 10 nm; the scanning speed is 1000 nm/min.

Example 1

(1) Preparing a crude oil solution sample and collecting a three-dimensional fluorescence spectrum.

Collecting a certain crude oil A, dissolving with n-heptane to prepare 5 standard sample solutions with concentrations of 10mg/L, 20mg/L, 40mg/L, 60mg/L and 80mg/L respectively. The three-dimensional fluorescence spectrum of the standard sample was measured, wherein the three-dimensional fluorescence spectrum of the standard sample at a concentration of 10mg/L was shown in FIG. 1.

(2) And solving a two-dimensional spectrum of the concentration primitive according to the three-dimensional fluorescence spectrum of the standard sample.

(a) Obtaining a two-dimensional spectrum of fluorescence emission intensity of a standard sample

Taking fig. 1 as an example, since the excitation wavelength range is 250-550 nm and the step diameter is 10nm, a total of (550-)/10 +1 is 31 emission spectra, an emission wavelength interval (containing 150 data points, λ is 370nm) of 301-450 nm is taken as a characteristic interval, the fluorescence emission intensity data of the 31 two-dimensional emission spectra in the characteristic interval are connected end to end in the order of the excitation wavelength from small to large, and the fluorescence emission intensity two-dimensional spectrum H corresponding to the standard sample is obtained₁₀，H₁₀Is a set of 4650 (31X 150) data points containing variables.

In the same way, two-dimensional fluorescence emission intensity spectrum H of other 4 standard samples with different concentrations can be obtained₂₀、H₄₀、H₆₀、H₈₀。

(b) Obtaining a two-dimensional spectrum S of concentration elements

After the 5 fluorescence emission intensity two-dimensional spectrum variables are smoothed (segmented Gaussian smoothing, window size is 5 wavelength points), each data point in the spectrum is divided by the corresponding concentration, namely H₁₀/10、H₂₀/20，H₂₀/20，H₄₀/40，H₆₀/60，H₈₀/80, each of which is denoted as P₁₀、P₂₀、P₄₀、P₆₀、P₈₀Adding the 5 two-dimensional spectraAnd average, i.e. S ═ P₁₀+P₂₀+P₄₀+P₆₀+P₈₀) And/5, obtaining a concentration element two-dimensional spectrum S as shown in figure 2.

(3) And (5) obtaining a correction set by the simulation calculation of the concentration primitive two-dimensional spectrum S and establishing an analysis model.

(a) Obtaining a two-dimensional spectrum of derived fluorescence emission intensity

The spectrum of crude oil A and the measured crude oil solution concentration have a linear correlation range with an upper limit of 100mg/L and a crude oil solution concentration of 80mg/L within the linear correlation range. Taking the unit concentration as 1mg/L, calculating the concentration as 1-80 mg/L, namely k is 1,2 … 80, and the concentration gradient difference is 1mg/L, namely S₁＝S×1、S₂＝S×2…S₈₀Obtaining 80 derived fluorescence emission intensity two-dimensional spectrums marked as S₁、S₂…S₈₀。

(b) Obtaining a correction set

Repeating the measurement twice to obtain three-dimensional fluorescence spectrum P of sample with concentration of 40mg/L_40-1And P_40-2The difference spectrum is the noise spectrum F, i.e. F ═ P_40-1-P_40-2And then random spectral noise Z is equal to F multiplied by M, wherein M is a random number between 0.5 and 1.2, and the random spectral noise S is added into the 80 derived fluorescence emission intensity two-dimensional spectrums, namely S₁＝S×1+F×M₁、S₂＝S×2+F×M₂…S₈₀＝S×80+F×M₈₀(M₁～M₈₀Random number of 0.5-1.2) to obtain 80 simulated fluorescence emission intensity two-dimensional spectra, forming a correction set by the simulated fluorescence emission intensity two-dimensional spectra and the corresponding crude oil concentration, performing second order differential and mean centering treatment on the correction set two-dimensional spectra, correlating the simulated fluorescence emission intensity two-dimensional spectra with the crude oil concentration by using a Partial Least Squares (PLS) method, and establishing an analysis model, wherein the number of main factors is 3, and the upper limit of the crude oil concentration of the extraction liquid predicted by the model is 80mg/L theoretically.

(4) Model validation

Preparing crude oil A solution with concentration of 15mg/L, 30mg/L, 50mg/L and 70mg/L with n-heptane, collecting three-dimensional fluorescence spectrum, and obtaining fluorescence thereof by the methodEmission intensity two-dimensional spectrum H₁₅、H₃₀、H₅₀、H₇₀After smoothing, second order differentiation and mean value centralization treatment, the crude oil concentration is predicted by substituting the analysis model established in the step (3), the result is shown in table 1, the root mean square error of the interactive verification is 1.58, and the correlation coefficient of the measured value and the predicted value is 0.998.

Example 2

And (3) testing the concentration of the crude oil solution of the actual sample: extracting the crude oil rock debris containing A with n-heptane, observing a three-dimensional fluorescence spectrogram, diluting the extract with n-heptane by 10 times, collecting the three-dimensional fluorescence spectrum of the diluted extract under the same measurement condition as the standard sample spectrum, and obtaining a two-dimensional fluorescence absorption intensity spectrum H of the diluted extract to be measured in the same processing mode as the standard sample_{To be measured}，H_{To be measured}After second order differentiation and mean centering, the diluted extract was substituted into the analysis model obtained in example 1, and the crude oil concentration of the diluted extract was calculated to be 35.6mg/L, and was multiplied by the dilution factor to obtain the crude oil concentration of 356mg/L in the rock debris extract.

TABLE 1

Sample numbering	The preparation concentration is mg/L	Predicted concentration mg/L
			1	15	14.3
2	30	28.9
			3	50	48.5
4	70	68.1