A crude oil blending method directly based on near infrared spectroscopy

Abstract

The present invention discloses a crude oil blending method directly based on near-infrared spectroscopy. This method bypasses the traditional idea of ​​property blending, screens out the characteristic bands of near-infrared spectra related to the properties of crude oil through correlation analysis, and directly blends the target crude oil using the characteristic band spectra of the component crude oils combined with the least squares method. The obtained blended crude oil has properties similar to those of the target crude oil, which fully meets the atmospheric and vacuum processing requirements of refining enterprises. This method avoids the complicated modeling and maintenance process of near-infrared spectroscopy online property detection in the blending process based on crude oil properties, and provides a new feasible solution for crude oil blending in refining enterprises.

Description

Technical Field

[0001] The invention relates to the field of near infrared spectrum online analysis, in particular to a method for crude oil blending directly based on near infrared spectrum. Background Art

[0002] Refining and petrochemical enterprises often use multiple components of crude oil to blend into a mixed crude oil with specific properties in a certain proportion to meet the production and processing needs of atmospheric and vacuum units. In the early days, the main method was to manually test the properties of crude oil and then make the proportion. In recent years, with the rapid development of near-infrared spectroscopy technology, a mathematical model between the spectrum and the properties of crude oil has been established through near-infrared spectroscopy analysis technology, and then the proportion optimization control has been carried out to achieve online blending of crude oil.

[0003] However, even if near-infrared spectroscopy online analysis technology is used, there are still some shortcomings in actual use, such as the difficulty of modeling and the high cost of model maintenance. In fact, for a specific refining enterprise, the requirements of its crude oil processing unit, namely the atmospheric and vacuum unit, on the properties of crude oil are relatively fixed, and the crude oil purchased by the enterprise is also purchased based on the range of crude oil properties that the atmospheric and vacuum unit can adapt to. The key properties of crude oil, such as sulfur content, acid value, naphtha yield and other indicators that are focused on during blending, have a good linear additive relationship with the near-infrared spectrum. If crude oil can be blended directly through near-infrared spectroscopy, this will save the tedious intermediate mathematical modeling process and greatly reduce the maintenance cost of the model, which has important application value for refining enterprises. Summary of the invention

[0004] In view of the above problems, the present invention discloses a crude oil blending method directly based on near-infrared spectroscopy. The method avoids the complex and tedious modeling and analysis process of crude oil properties, screens out spectral bands related to each property through correlation analysis, directly blends the target crude oil using the relevant spectral bands of the component crude oils, and calculates the blending ratio. The specific steps are as follows:

[0005] (1) Collect near-infrared spectra of crude oil of m components X = [x 1 ,x 2 ,…,x m ], collect n properties of the crude oil component, and construct the property matrix Y = [y 1 ,y 2 ,…,y n ];

[0006] (2) Preprocess X to obtain spectrum X p , preprocessing includes vector normalization and Savitzky-Golay first-order convolution derivative, for X p The correlation analysis between each wave number point and property Y is performed to screen out the characteristic wave number bands related to the property [r 1 ,r2 ,…,r n ], the steps of correlation analysis and characteristic wave number band screening are as follows:

[0007] (2-1) Select X p The spectral data x corresponding to the first wave number point p1 , calculate the spectral data x p1 and the first property y 1 The correlation coefficient between 11 :

[0008]

[0009] in

[0010] (2-2) Calculate the spectrum X according to the method of step (2-1) p All wave number points and the first property y 1 The correlation coefficient between them is obtained by 11 ,ρ 12 ,…,ρ 1q ] T , where q is the total number of wave number points in the spectrum, and |ρ 1s |>0.5,s∈[1,2,…q] corresponds to the wave number point as the characteristic wave number segment r of the first property 1 ;

[0011] (2-3) Calculate the spectrum X according to the method of step (2-2) p The characteristic wave number band between all wave number points and n properties Y [r 1 ,r 2 ,…,r n ];

[0012] (3) Determine the target crude oil and k component crude oils for blending according to the production requirements of the refining enterprise, and collect the spectrum X of the target crude oil A and component crude oil spectrum X B ;

[0013] (4) To X A and X B Preprocessing is performed to obtain the spectrum and spectrum

[0014] (5) According to the selected relevant wave number band [r 1 ,r 2 ,…,r n ]Will and Divided into characteristic spectra corresponding to different properties and

[0015] (6) Use the least squares method to analyze the spectrum Blending is performed to calculate the blending ratio w of the component crude oils. The specific calculation steps are as follows:

[0016] (6-1) According to [r 1 ,r 2 ,…,r n ] to obtain the number of wave number points in the wave number bands with different characteristics [l 1 ,l 2 ,…,l n ];

[0017] (6-2) Use the following method to calculate the blending ratio w:

[0018]

[0019]

[0020] 0≤w i ≤1,i∈[1,2,…,k]

[0021] where α j Represents the weight of the jth property in the blending optimization algorithm; The characteristic spectrum representing the jth property of the component crude oil; represents the characteristic spectrum of the jth property of the target crude oil; w i Indicates the blending ratio of the i-th component crude oil. Beneficial effects:

[0022] The present invention discloses a crude oil blending method directly based on near infrared spectroscopy, which screens out spectral bands related to various properties through correlation analysis, directly blends the target crude oil using the relevant spectral bands of the component crude oils, and calculates the blending ratio. The method eliminates the cumbersome intermediate mathematical modeling process and greatly reduces the maintenance cost of the model, and has important application value for refining enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Implementation flow chart of the method of the present invention;

[0024] Figure 2 Crude oil is between 4000 and 4800 cm -1 Near infrared spectrum of

[0025] Figure 3 This is the near infrared spectrum of crude oil after pretreatment;

[0026] Figure 4(a) is a correlation coefficient diagram of crude oil spectral wave number points and API gravity properties;

[0027] FIG4( b ) is a correlation coefficient diagram of crude oil spectral wave number points and sulfur content properties;

[0028] Figure 4(c) is a correlation coefficient diagram of crude oil spectral wave number points and acid value properties;

[0029] Figure 4(d) is a correlation coefficient diagram of crude oil spectral wavenumber points and naphtha yield properties.

[0030] Specific implementation process

[0031] The following is a detailed calculation process and a specific operation flow to illustrate the implementation effect of the present invention using near infrared spectroscopy for crude oil blending. This implementation case is implemented based on the technical solution of the present invention, but the protection scope of the present invention is not limited to the following embodiments.

[0032] The present invention takes the crude oil blending scheme of a certain refining enterprise as an example. The enterprise has two sets of atmospheric and vacuum devices, and has different requirements for the properties of the processed crude oil (the properties mainly considered include API degree, sulfur content, acid value and naphtha yield). Currently, four kinds of crude oil components are mainly purchased from the international market for processing in atmospheric and vacuum devices.

[0033] In combination with this embodiment, the implementation steps of the present invention are as follows: Figure 1 As shown, the details are as follows:

[0034] (1) Extract 84 crude oils from the crude oil spectrum and property database at 4000-4800 cm -1 The near infrared spectrum of X=[x 1 ,x 2 ,…,x 84 ] and its related properties matrix Y = [y 1 ,y 2 ,y 3 ,y 4 ], where the spectrum is Figure 2 As shown, the property y 1 ,y 2 ,y 3 ,y 4 They correspond to API degree, sulfur content, acid value and naphtha yield respectively;

[0035] (2) Preprocess X to obtain spectrum X p The preprocessing methods include vector normalization and Savitzky-Golay first-order convolution derivative, spectral X p like Figure 3 As shown; respectively find X p The correlation coefficients ρ between each wave number point and the four properties are selected, and the wave number points corresponding to |ρ|>0.5 are selected to form the characteristic wave number segment [r 1 ,r 2 ,r3 ,r 4 ], where [r 1 ,r 2 ,r 3 ,r 4 ] are the characteristic wavenumber bands corresponding to API degree, sulfur content, acid value and naphtha yield, respectively. Figure 4(a)-Figure 4(d) X p The correlation coefficients of each wave number point and the four properties, where the horizontal line is the threshold line where the absolute value of the correlation coefficient is equal to 0.5;

[0036] (3) According to the plan of the refining enterprise, four kinds of crude oil were used to blend the target crude oil, among which the near infrared spectrum of the target crude oil was X A Near infrared spectra of crude oil and its components B ;

[0037] (4) To X A and X B Perform the preprocessing in step (2) respectively to obtain the spectrum and spectrum

[0038] (5) According to the characteristic wave number band [r 1 ,r 2 ,…,r n ]Will and Divided into characteristic spectra corresponding to different properties and And record the number of wave number points of each characteristic spectrum [l 1 ,l 2 ,l 3 ,l 4 ], where l 1 =167,l 2 =37,l 3 =97,l 4 =44;

[0039] (6) Calculate the blending ratio w using the following method:

[0040]

[0041]

[0042] 0≤w i ≤1

[0043] The blending optimization weight setting is: 1 =1,α 2 =5,α 3 =5,α 4 =10.

[0044] The results of calculating the crude oil blending ratio are shown in Table 1:

[0045] Table 1 Crude oil blending ratio

[0046]

[0047] The properties of the blended crude oil were tested and compared with the target crude oil. The results are shown in Table 2:

[0048] Table 2 Comparison of properties of target crude oil and blended crude oil

[0049]

[0050] As can be seen from Table 2, the properties of the blended crude oil and the target crude oil are similar, which fully meets the processing requirements of the company's atmospheric and vacuum distillation unit. It can be seen that for a specific refining enterprise, the crude oil blending method directly based on near-infrared spectroscopy proposed in the present invention can bypass the spectral property modeling process of crude oil and can be applied to the crude oil blending production process of the refining enterprise.

Claims

1. A crude oil blending method directly based on near infrared spectroscopy, characterized in that The following steps are involved: (1) Collect m types of crude oil 4000~4800cm -1 The near infrared spectrum X=[x1,x2,…,x m ], collect n properties of crude oil and construct the property matrix Y = [y1, y2, ..., y n ]; where n=4, y1, y2, y3, y4 correspond to API gravity, sulfur content, acid value and naphtha yield respectively; (2) Preprocess X to obtain spectrum X p , for X p The correlation analysis between each wave number point and property Y is performed to screen out the characteristic wave number bands [r1, r2, ..., r n ]; the preprocessing methods include vector normalization and Savitzky-Golay first-order convolution derivative; correlation analysis and wave number point screening are performed in the following manner: (2-1) Select X p The spectral data x corresponding to the first wave number point p1 , calculate the spectral data x p1 The correlation coefficient ρ between the first property y1 11 : in x pi1 represents the spectral data corresponding to the first wave number point of the i-th component crude oil; y i1 represents the first property of the i-th component crude oil; (2-2) Calculate the spectrum X according to the method of step (2-1) p The correlation coefficients between all wave number points and the first property y1 are obtained as the correlation coefficient vector [ρ 11 ,ρ 12 ,...,ρ 1q ] T , where q is the total number of wave number points in the spectrum, and ρ is selected 1s |>0.5, the wave number point corresponding to s∈[1,2,...q] is taken as the characteristic wave number segment r1 of the first property; (2-3) Calculate the spectrum X according to the method of step (2-2) p The characteristic wave number range [r1, r2, ..., r n ]; (3) Determine the target crude oil and k component crude oils for blending according to the production requirements of the refining enterprise, and collect the spectrum X of the target crude oil A and component crude oil spectrum X B ; (4) To X A and X B After preprocessing, we get and (5) According to the selected wave number band [r1, r2, ..., r n ]Will and Divided into characteristic spectra corresponding to different properties and (6) Use the least squares method to directly calculate the spectrum Blending is performed to calculate the blending ratio w of the component crude oils; the blending ratio w is calculated in the following manner: (6-1) According to [r1,r2,...,r n ] to obtain the number of wave number points of different characteristic wave number bands [l1,l2,...,l n ]; (6-2) Use the following method to calculate the blending ratio w: 0≤w i ≤1,i∈[1,2,…,k] where α j Represents the weight of the jth property in the blending optimization algorithm; The characteristic spectrum representing the jth property of the component crude oil; represents the characteristic spectrum of the jth property of the target crude oil; w i Represents the blending ratio of the i-th component crude oil.

Images