Method for predicting fluid probability through Epanechnikov kernel function well-seismic joint

Abstract

The invention discloses a method for predicting a fluid probability through well-to-seismic combination of an Epanechnikov kernel function. Comprising the following steps: S1, acquiring drilling coring data, geological stratification data, logging curve data, logging interpretation data and longitudinal wave impedance, transverse wave impedance and density data volume data obtained by pre-stack earthquake three-parameter inversion; s2, identifying fluid properties; s3, elastic parameters and fluid sensitive factor curve data are calculated; s4, oil and gas sensitive attributes in the target layer are selected; s5, making a label; s6, manufacturing a fluid probability measurement plate; s7, calculating a corresponding sensitive elastic parameter attribute data volume; s8, calculating a fluid probability body; s9, determining a probability threshold value for distinguishing fluid properties; and S10, predicting the fluid probability. The method has the advantage that the fluid property is predicted in a probabilistic mode.

Description

Technical Field

[0001] The invention relates to the technical field of petrochemical engineering, and in particular to a method for predicting fluid probability by combining well-seismic and Epanechnikov kernel functions. Background Art

[0002] Modern oil and gas exploration mainly uses geophysical technology for reservoir prediction and hydrocarbon detection, that is, based on seismic information, other data (geology, logging, reservoir, etc.) are used as constraints to predict the quality parameters of oil and gas reservoirs, such as geometric characteristics, geological characteristics, reservoir physical characteristics, etc. The main research content is roughly divided into four aspects: (1) lithofacies: predicting the phase belt that controls the development of reservoirs; (2) lithology: predicting the lithology, thickness and top structural morphology of reservoirs; (3) physical properties: predicting such as porosity and permeability; (4) hydrocarbon content: comprehensive analysis and research on the properties of fluids contained in the reservoir and their distribution, saturation, etc. Among them, hydrocarbon content prediction, that is, hydrocarbon detection, is extremely important but also the most difficult part. Reservoirs containing oil and gas are valuable reservoirs in industry.

[0003] The use of seismic data for hydrocarbon detection has evolved from the initial seismic profile bright spot technology to parameter inversion, low-frequency shadow, spectrum analysis and other technologies based on post-stack seismic data, and then to the current simultaneous inversion of elastic parameters based on pre-stack seismic data, various AVO analyses, and construction of various fluid sensitive factors. Hydrocarbon detection technology has received continuous attention and development, forming a systematic and comprehensive research method that mainly uses pre-stack seismic data detection technology and supplements it with post-stack seismic data detection technology.

[0004] The existing seismic data hydrocarbon detection technology has the following main problems: 1) Most of the existing technical solutions, whether based on post-stack seismic data methods or pre-stack seismic data methods, are predicted by a single seismic attribute that is sensitive to oil and gas. They are affected by the quality of seismic data and the properties of the sensitive attribute itself, and there are multiple solutions; 2) Most of the existing hydrocarbon detection technologies are based on qualitative interpretation, which is difficult to quantify. With the deepening of exploration and production needs, the quantification of oil and gas content has become one of the basic requirements for hydrocarbon detection. Summary of the invention

[0005] The object of the present invention is to provide a method for jointly predicting fluid probability using Epanechnikov kernel function and well seismic, which predicts fluid properties in a probabilistic manner.

[0006] To achieve the above object, the present invention adopts the following technical solution, including the following steps:

[0007] S1. Acquire drilling coring data, geological stratification data, logging curve data, logging interpretation data, and P-wave impedance, S-wave impedance, and density data obtained by pre-stack seismic three-parameter inversion, wherein the logging interpretation data includes logging oil and gas interpretation data and logging lithology interpretation data;

[0008] S2. Identify the fluid properties in the target layer on the well based on the drilling coring data, geological stratification data, logging curve data, and logging interpretation data;

[0009] S3, calculating elastic parameters or constructing a fluid indicator factor curve according to the P-wave velocity, S-wave velocity and density data in the well logging curve data;

[0010] S4. Select the oil and gas sensitive attributes in the target layer by performing logging rock physical analysis on the elastic parameter or fluid indicator factor curve;

[0011] S5, calibrating the oil and gas sensitive attributes in S4 and the fluid properties in S2 in the well section and making labels;

[0012] S6, using the high-dimensional Epanechnikov kernel function to perform high-dimensional kernel density estimation on multiple oil and gas sensitive attributes at the well point, and using the label data prepared in S5 to prepare the fluid probability volume plate according to the Bayesian probability classification method;

[0013] S7, calculating the corresponding sensitive elastic parameter attribute data volume from the P-wave impedance, S-wave impedance and density data volume obtained by pre-stack seismic three-parameter inversion;

[0014] S8, projecting the sensitive elastic parameter attribute data volume calculated in S7 onto the fluid probability volume plate prepared in S6, and calculating the fluid probability volume;

[0015] S9, by analyzing the response characteristics of the fluid properties in the target layer section on the well in S2 to the fluid probability body, determining the probability threshold value for distinguishing the fluid properties;

[0016] S10. quantify the fluid probability data body obtained in S8 according to the probability threshold value for distinguishing fluid properties determined in S9, so as to predict the fluid properties of the reservoir in the target layer.

[0017] Preferably, in step S2, the top and bottom ranges of the target layer study segment on the well are determined by geological stratification data; the lithological characteristics of different depth segments in the target layer are determined by drilling coring data, logging curve data and logging lithological interpretation data; and the fluid properties of the favorable lithological segment are determined by logging oil and gas interpretation data.

[0018] Preferably, in step S3, the elastic parameters include longitudinal wave impedance, transverse wave impedance, longitudinal and transverse wave velocity ratio, Poisson's ratio, Poisson's damping factor, Lame constant, shear modulus, bulk modulus, and Young's modulus.

[0019] Preferably, in step S3, the fluid sensitivity factor curve is constructed by constructing the fluid indication factor Y with the help of the longitudinal wave impedance and the shear wave impedance in the well, and its expression is:

[0020] Y=AIcosθ+SIsinθ (1)

[0021] In formula (1), θ (0°≤θ≤360°) is the factor angle, AI is the longitudinal wave impedance, and SI is the transverse wave impedance. Under given longitudinal wave impedance AI and transverse wave impedance SI, as the angle θ changes, a series of factor curves can be obtained from the above formula. The logging curve related to the fluid is selected as the target curve, and a correlation scan is performed with the series of factor curves to obtain the factor curve with the highest correlation. This factor curve is the constructed fluid indication factor curve, and the corresponding factor angle is called the fluid indication factor angle.

[0022] Preferably, in step S4, the method of well logging petrophysical analysis is to select two well logging curves for intersection display, and select the oil and gas sensitive attributes in the target layer from a large number of well logging curve data through the clustering characteristics of the intersection diagram data.

[0023] Preferably, in step S6, the Bayesian probability classification method is:

[0024] Using probability to represent all types of uncertainty, for G categories, the Bayesian rule for category "i" is;

[0025]

[0026] In formula (2), X represents the sample point data, c i is a fluid property, p(c i |X) A sample point is classified as fluid property c i The posterior probability density, p(X|c i ) is the fluid property c i The conditional probability density of i ) is c i The prior probability density of is equal to the proportion of this fluid property in all fluid types; p(X) is the marginal probability density.

[0027] Preferably, in step S6, the fluid property c is calculated by a non-parametric kernel density estimation method. i The conditional probability density p(X|c i );

[0028] Let x 1 ,x 2 ,…,x nare n independent and identically distributed sample points, whose probability density is f, then the kernel density estimation formula of the one-dimensional probability distribution function is;

[0029]

[0030] Where n is the number of sample points, x i is the observed value of a sample, K is the selected kernel function (non-negative, integral is 1, conforms to the probability density property, and has a mean of 0), and h is the smoothing bandwidth (h>0) that controls the smoothing amount; the kernel with the subscript h is called the scaling kernel, which is defined as

[0031] Select multiple fluid sensitive attribute parameters and use high-dimensional kernel density estimation. The formula is:

[0032]

[0033] Where D is the number of selected parameters, d(xi,X i ) is x and sample X in D-dimensional space i distance, h is the smoothing bandwidth parameter, and K is the kernel function;

[0034] The Epanechnikov kernel function is selected, and its formula is as follows;

[0035]

[0036] Substituting the Epanechnikov kernel function into equation (3), we obtain the high-dimensional kernel density estimation formula:

[0037]

[0038] Among them, C D is the volume of the D-dimensional hypersphere, which is expressed as;

[0039]

[0040] here,

[0041] Preferably, in step S6, the observation value X is determined to belong to fluid c according to the following principle: i or fluid c j ;

[0042] (1) If p(X|c i )p(c i )>p(X|c j )p(c j ), then X belongs to fluid c i ;

[0043] (2) If p(X|cj )p(c j )>p(X|c i )p(c i ), then X belongs to fluid c j .

[0044] Preferably, in step S7, the corresponding elastic parameter attribute data body is calculated according to the elastic parameter definition in step S3, and the longitudinal wave impedance data body and the shear wave impedance data body can be converted into a fluid indicator factor data body according to the fluid indicator factor angle determined by the logging curve in step S3.

[0045] The beneficial effects of the present invention are:

[0046] 1. Based on the high-dimensional Epanechnikov kernel function, multiple fluid sensitive attributes are estimated by high-dimensional kernel density, and the fluid probability density estimation is obtained through the Bayesian probability classification method to predict the fluid properties in a probabilistic way;

[0047] 2. The fluid property data predicted by combining well-seismic data with multiple fluid sensitive attributes is highly reliable. This method is universal and can effectively solve the problem of difficult fluid detection in carbonate rocks, igneous rocks and unconventional reservoirs in the industry;

[0048] 3. By calibrating with actual drilling and logging data, the probability threshold value for distinguishing oil and gas in the research target reservoir is determined, and the distribution characteristics of fluids in the reservoir space are quantitatively studied, which will take hydrocarbon detection results one step further and improve the reliability of the basis for deploying exploration wells in oil and gas reservoirs in the target area, thereby increasing the success rate of exploration wells in oil and gas reservoirs. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Schematic diagram of the process of the present invention;

[0050] Figure 2 Schematic diagram of the construction principle of fluid sensitive indicator factors;

[0051] Figure 3 Obtain the fluid indicator factor angle correlation analysis diagram based on the target curve;

[0052] Figure 4 Fluid indicator factor curve analysis chart;

[0053] Figure 5 Optimizing fluid selection and sensitive elastic parameter attribute crosstalk using well logging rock physics analysis Figure 1 ;

[0054] Figure 6 Optimizing fluid selection and sensitive elastic parameter attribute crosstalk using well logging rock physics analysis Figure 2 ;

[0055] Figure 7Optimizing fluid selection and sensitive elastic parameter attribute crosstalk using well logging rock physics analysis Figure 3 ;

[0056] Figure 8 Calibrate the fluid properties of the selected sensitive elastic parameter attributes and produce a label data feature map;

[0057] Fig. 9 Fluid probability scale based on Epanechnikov kernel function probability prediction technology Figure 1 ;

[0058] Fig.10 Fluid probability scale based on Epanechnikov kernel function probability prediction technology Figure 2 ;

[0059] Fig.11 This is a fluid probability density data volume result diagram obtained based on the technical solution of the present invention. DETAILED DESCRIPTION

[0060] The invention is further described in detail below in conjunction with the accompanying drawings so that those skilled in the art can implement it according to the description.

[0061] It should be understood that terms such as “having”, “including” and “comprising” used herein do not exclude the existence or addition of one or other elements or combinations thereof.

[0062] Example

[0063] like Figure 1 As shown, the present invention provides a method for predicting fluid probability based on Epanechnikov kernel function combined with well seismic multi-attribute, including the following steps:

[0064] S1. Acquire drilling coring data, geological stratification data, logging curve data, logging interpretation data, and P-wave impedance, S-wave impedance, and density data obtained by pre-stack seismic three-parameter inversion, wherein the logging interpretation data includes logging oil and gas interpretation data and logging lithology interpretation data;.

[0065] S2. Identify the fluid properties in the target layer section on the well according to the drilling coring data, geological stratification data, logging curve data, and logging interpretation data; wherein, determine the top and bottom range of the target layer study section on the well according to the geological stratification data, such as Figure 4 As shown in the fourth column, the geological layer E3d2U at a depth of 2900 meters is taken as the top boundary of the target layer, and the geological layer E3d3 at a depth of 3900 meters is taken as the bottom boundary, which determines the range of the target layer Dongying Formation Dongsan Section (E3d2U to E3d3). The logging curves and logging data in the subsequent analysis are all taken from this target layer. According to the drilling coring data, logging curve data ( Figure 4 Column 1) and well logging lithology interpretation data ( Figure 4 Column 4) determines the lithological characteristics of different depth sections in the target layer, based on the logging oil and gas interpretation data ( Figure 4 The filling curve in the fifth column is the oil saturation, indicating that the formation contains oil) to determine the favorable lithology section ( Figure 4 The fluid nature of the sandstone in column 4 is oil, gas or water.

[0066] S3, according to the longitudinal wave velocity (V p ), shear wave velocity (V s ) and density (ρ) data to calculate elastic parameter curves or construct fluid indicator factor curve data. Commonly used elastic parameters include longitudinal wave impedance, transverse wave impedance, longitudinal and transverse wave velocity ratio, Poisson's ratio, Poisson's damping factor, Lame constant, shear modulus, bulk modulus, and Young's modulus. The definitions and geological meanings of the above elastic parameters are as follows:

[0067] Longitudinal wave impedance AI: AI = ρ-V p ;

[0068] Shear wave impedance SI: SI = ρ * V s ;

[0069] P-wave velocity ratio VP / VS: VP / VS=V p / V s

[0070] Poisson's ratio σ:

[0071] Poisson damping factor PDF:

[0072] Lamé constant λ:

[0073] Shear modulus μ:

[0074] Bulk modulus BULK:

[0075] Young's modulus E:

[0076] The construction of the fluid sensitivity factor curve is based on Connolly's theory on the relationship between elastic parameters and longitudinal wave impedance and shear wave impedance. The target correlation algorithm is used to construct the fluid indicator factor Y with the help of the longitudinal wave impedance and shear wave impedance in the well. Its expression is:

[0077] Y=AI cosθ+SI sinθ (1)

[0078] In formula (1), θ (0°≤θ≤360°) is the rotation angle, also known as the factor angle.

[0079] The construction principle of fluid indicator factors is as follows Figure 2 As shown in the figure, under given longitudinal and transverse wave impedance curves, as the angle θ changes, a series of factor curves can be obtained from the above formula. Studies have shown that the cross-correlation calculations using the elastic parameters in the table as the target curve and a series of factor curves that change with angles can all be highly correlated with the factor curve of a certain factor angle, and the correlation is close to 1 (e.g. Figure 2 As shown in the table on the right). Generally, the water saturation curve, resistivity curve or natural potential curve is selected as the target curve, and a correlation scan is performed with the series of factor curves calculated by formula (1) to obtain the factor curve with the highest correlation. This curve is called the fluid indication factor curve, and the corresponding factor angle is called the fluid indication factor angle. Figure 3 This is the fluid indicator factor angle obtained by using multiple wells with the water saturation curve as the target curve in the example. The different types of dotted lines in the figure represent the correlation curves (the horizontal axis is the angle, and the vertical axis is the correlation coefficient) of different wells (BZ2-1-2 well, BZ1-1-2 well, BZ1-1-1 well) as the angle θ (0°≤θ≤360°) changes. The solid line is the average correlation curve of multiple wells. The fluid indicator factor angle determined in this example is 300 degrees. Figure 4 is the fluid factor curve Figure 4 ②Compared with the measured curve Figure 4 ① and water saturation curve ( Figure 4 ⑤) The comparative analysis shows that the fluid factor curve is in good agreement with the fluid.

[0080] S4. Select the oil and gas sensitive attributes in the target layer by performing logging rock physics analysis on the elastic parameter or fluid indicator factor curve. Select two logging curves for intersection display, and the grayscale of the sample point is represented by the fluid property determined in S2 (such as water saturation, 1 means 100% water, 0 means 100% oil or gas without water). The oil and gas sensitive attributes in the target layer are selected from a large number of logging curves (including measured curves, interpretation curves, elastic parameters, fluid indicator factor curves, etc.) through the clustering characteristics of the intersection diagram data.

[0081] Figure 5 This is a cross-plot of longitudinal wave impedance (Zp) and longitudinal and transverse wave velocity ratio (Vp / Vs ratio) under water saturation. It can be seen from the figure that in this example, the samples with Vp / Vs values ​​less than 1.9 have low water content (<70%), and oil and water can be distinguished. The analysis shows that the longitudinal and transverse wave velocity ratio is a fluid-sensitive property.

[0082] same Figure 6 This is a cross-plot of longitudinal wave impedance (Zp) and fluid indicator factor (FI) at water saturation. In this example, when the FI value is below 390, the sample contains oil and gas. Analysis shows that the fluid indicator factor is also a fluid-sensitive property.

[0083] Figure 7 is a cross-plot of the P-wave velocity ratio and the fluid indicator factor at water saturation. It is further determined from the cross-plot that the P-wave velocity ratio and the fluid indicator factor are both fluid sensitive parameter attributes in this example;

[0084] S5, calibrate the oil and gas sensitive attributes in S4 and the fluid properties in S2 in the well section and make labels; Figure 8 As shown;

[0085] S6. In the well section, the P-wave velocity ratio and fluid indicator factor fluid sensitivity curve selected in S4 are discretized into sample points according to depth, X j (x 1 , x 2 , …, x n ), j = 1, 2 correspond to the ratio of longitudinal and transverse wave velocities and fluid indicator factors, n is the number of sample points, x i is the observed value of a sample. First, multiple oil and gas sensitive attributes are estimated at the sample point using the high-dimensional Epanechnikov kernel function. The calculation formula of the Epanechnikov kernel function is as follows:

[0086]

[0087] Substituting formula (2) into the high-dimensional kernel density estimation formula (3) we get the high-dimensional kernel density estimation calculation formula (4):

[0088]

[0089] Where D is the number of selected parameters (in the embodiment, D=2), d(x, X i ) is x and sample X in D-dimensional space i distance, h is the smoothing bandwidth parameter, and K is the kernel function.

[0090]

[0091] Among them, C D is the volume of the D-dimensional hypersphere, which is expressed as

[0092]

[0093] here,

[0094] At the sample point, the ratio of longitudinal and transverse wave velocities and the oil and gas sensitive attributes of fluid indicator factors are used to perform high-dimensional kernel density estimation using formula (4) to calculate the conditional probability p(X|c i ), where c i is a fluid property. Then, using the label data created in S5 and ci The prior probability density p(c i ) and marginal probability density p(X) are used to make the fluid probability volume version according to the Bayesian probability classification formula (6):

[0095]

[0096] Finally, the observation X is determined to belong to fluid c according to the following principle: i or fluid c j ;

[0097] (1) If p(X|c i )p(c i )>p(X|c j )p(c j ), then X belongs to fluid c i ;

[0098] (2) If p(X|c j )p(c j )>p(X|c i )p(c i ), then X belongs to fluid c j .

[0099] Fig. 9 and Fig.10 The fluid probability scale plate made according to the present invention, the left figure is an example of the fluid (oil and water) probability scale plate related to the longitudinal and transverse wave velocity ratio, and the right figure is an example of the fluid (oil and water) probability scale plate related to the fluid indicator factor. The horizontal axis and the vertical axis in the probability scale plate are the attributes of the mark, the solid points are determined to be water-containing samples, the hollow points are determined to be oil-containing samples, the curve around the solid points of the water-containing samples is the water-containing probability curve, and the curve around the hollow points of the oil-containing samples is the oil-containing probability curve. In this way, the measured samples are used to make a fluid probability scale plate by using the high-dimensional kernel density estimation of the high-dimensional Epanechnikov kernel function and according to the Bayesian probability classification method. A group of attribute values ​​are put into the scale plate to read the oil or water probability value of the formation where the group of attributes is located;

[0100] S7, calculate the corresponding sensitive elastic parameter attribute data volume from the P-wave impedance, S-wave impedance and density data volume obtained by pre-stack seismic three-parameter inversion; the pre-stack three-parameter synchronous inversion can invert the P-wave impedance, S-wave impedance and density data volume, so that the corresponding elastic parameter attribute data volume can be calculated according to the elastic parameter attribute definition mentioned in S3. The P-wave impedance data volume and S-wave impedance data volume can be converted into the fluid indicator factor data volume according to the fluid indicator factor angle determined by the logging curve in S3;

[0101] S8, projecting the sensitive elastic parameter attribute data volume calculated in S7 onto the fluid probability volume plate prepared in S6, and calculating the fluid probability volume;

[0102] S9, by analyzing the response characteristics of the fluid properties in the target layer section on the well in S2 to the fluid probability body, determining the probability threshold value for distinguishing the fluid properties;

[0103] S10, quantify the fluid probability data body obtained in S8 according to the probability threshold value for distinguishing fluid properties determined in S9, so as to predict the fluid probability and the distribution characteristics of the fluid in the reservoir space, such as Fig.11 As shown, the high-probability (dark) strata on the seismic section are oil-bearing strata.

[0104] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for predicting fluid probability by combining wellbore and seismic analysis using Epanechnikov kernel function, characterized in that: The following steps are involved: S1. Acquire drilling coring data, geological stratification data, logging curve data, logging interpretation data, and P-wave impedance, S-wave impedance, and density data obtained by pre-stack seismic three-parameter inversion, wherein the logging interpretation data includes logging oil and gas interpretation data and logging lithology interpretation data; S2. Identify the fluid properties in the target layer on the well based on the drilling coring data, geological stratification data, logging curve data, and logging interpretation data; S3, calculating elastic parameters or constructing a fluid indicator factor curve according to the P-wave velocity, S-wave velocity and density data in the well logging curve data; S4. Select the oil and gas sensitive attributes in the target layer by performing logging rock physical analysis on the elastic parameter or fluid indicator factor curve; S5, calibrating the oil and gas sensitive attributes in S4 and the fluid properties in S2 in the well section and making labels; S6, using the high-dimensional Epanechnikov kernel function to perform high-dimensional kernel density estimation on multiple oil and gas sensitive attributes at the well point, and using the label data prepared in S5 to prepare the fluid probability volume plate according to the Bayesian probability classification method; S7, calculating the corresponding sensitive elastic parameter attribute data volume from the P-wave impedance, S-wave impedance and density data volume obtained by pre-stack seismic three-parameter inversion; S8, projecting the sensitive elastic parameter attribute data volume calculated in S7 onto the fluid probability volume plate prepared in S6, and calculating the fluid probability volume; S9, by analyzing the response characteristics of the fluid properties in the target layer section on the well in S2 to the fluid probability body, determining the probability threshold value for distinguishing the fluid properties; S10. quantify the fluid probability data body obtained in S8 according to the probability threshold value for distinguishing fluid properties determined in S9, so as to predict the fluid properties of the reservoir in the target layer.

2. The method for joint prediction of fluid probability by Epanechnikov kernel function and well seismic according to claim 1 is characterized by: In step S2, the top and bottom ranges of the target layer study section on the well are determined by geological stratification data; the lithological characteristics of different depth sections in the target layer are determined by drilling coring data, well logging curve data and well logging lithological interpretation data; The fluid properties of favorable lithology sections are determined through logging oil and gas interpretation data.

3. The method for joint prediction of fluid probability by Epanechnikov kernel function and well seismic according to claim 1 is characterized by: In step S3, the elastic parameters include longitudinal wave impedance, transverse wave impedance, longitudinal and transverse wave velocity ratio, Poisson's ratio, Poisson's damping factor, Lame constant, shear modulus, bulk modulus, and Young's modulus.

4. The method for joint prediction of fluid probability by Epanechnikov kernel function and seismic analysis according to claim 1 is characterized by: In step S3, the fluid sensitivity factor curve is constructed by using the longitudinal wave impedance and the shear wave impedance in the well to construct the fluid indication factor Y, and its expression is: Y=AI cosθ+SI sinθ (1) In formula (1), θ (0°≤θ≤360°) is the factor angle, AI is the longitudinal wave impedance, and SI is the transverse wave impedance. Under given longitudinal wave impedance AI and transverse wave impedance SI, as the angle θ changes, a series of factor curves can be obtained from the above formula. The logging curve related to the fluid is selected as the target curve, and a correlation scan is performed with the series of factor curves to obtain the factor curve with the highest correlation. This factor curve is the constructed fluid indication factor curve, and the corresponding factor angle is called the fluid indication factor angle.

5. The method for joint prediction of fluid probability by Epanechnikov kernel function and well seismic according to claim 1 is characterized by: In step S4, the method of well logging petrophysical analysis is to select two well logging curves for cross-plot display, and select the oil and gas sensitive attributes in the target layer from a large number of well logging curve data through the clustering characteristics of the cross-plot data.

6. The method for joint prediction of fluid probability by Epanechnikov kernel function and seismic analysis according to claim 1, characterized in that: In step S6, the Bayesian probability classification method is: Using probability to represent all types of uncertainty, for G categories, the Bayesian rule for category "i" is; In formula (2), X represents the sample point data, c i is a fluid property, p(c i |X) A sample point is classified as fluid property c i The posterior probability density, p(X|c i ) is the fluid property c i The conditional probability density of i ) is c i The prior probability density of is equal to the proportion of this fluid property in all fluid types; p(X) is the marginal probability density.

7. The method for predicting fluid probability by combining well and seismic data with the Epanechnikov kernel function according to claim 6, characterized in that: In step S6, the fluid property c is calculated by a nonparametric kernel density estimation method. i The conditional probability density p(X|c i ); Let x 1, x 2,…， x n are n independent and identically distributed sample points, whose probability density is f, then the kernel density estimation formula of the one-dimensional probability distribution function is; Where n is the number of sample points, x i is the observed value of a sample, K is the selected kernel function (non-negative, integral is 1, conforms to the probability density property, and has a mean of 0), and h is the smoothing bandwidth (h>0) that controls the smoothing amount; the kernel with the subscript h is called the scaling kernel, which is defined as Select multiple fluid sensitive attribute parameters and use high-dimensional kernel density estimation. The formula is: Where D is the number of selected parameters, d(xi,X i ) is x and sample X in D-dimensional space i distance, h is the smoothing bandwidth parameter, and K is the kernel function; The Epanechnikov kernel function is selected, and its formula is as follows; Substituting the Epanechnikov kernel function into equation (3), we obtain the high-dimensional kernel density estimation formula: when If true, otherwise 0 (6) Among them, C D is the volume of the D-dimensional hypersphere, which is expressed as; here, 8. The method for joint prediction of fluid probability by Epanechnikov kernel function and seismic analysis according to claim 7 is characterized by: In step S6, the observation value X is determined to belong to fluid c according to the following principle: i or fluid c j ; (1) If p(X|c i )p(c)>p(X|c j )p(c j ), then X belongs to fluid c i ; (2) If p(X|c j )p(c j )>p(X|c i )p(c), then X belongs to fluid c j .

9. The method for joint prediction of fluid probability by Epanechnikov kernel function and well seismic according to claim 8 is characterized by: In step S7, the corresponding elastic parameter attribute data volume is calculated according to the elastic parameter definition in step S3, and the longitudinal wave impedance data volume and the shear wave impedance data volume can be converted into a fluid indicator factor data volume according to the fluid indicator factor angle determined by the logging curve in step S3.

Images