Reservoir opening fracture prediction method

Abstract

The invention provides a reservoir opening fracture prediction method, and belongs to the technical field of fracture prediction, and the method comprises the steps: S1, building a target stratum fracture information standardized data set through employing an original three-dimensional seismic data body, imaging logging information and geological data; s2, distinguishing the target layer crack information standardized data set, reserving values of data points consistent with open cracks, and establishing an open crack quantitative modeling data body; and S3, combining reservoir seismic inversion, combining the open fracture quantitative modeling data volume, and obtaining the density of the open fracture of the reservoir. By means of the method, the development degree of natural effective fractures containing glutenite can be effectively determined, information such as the fracture opening direction and the fracture density can be obtained, and then rich fracture information is provided for research of fractured reservoirs.

Description

Technical Field

[0001] The invention belongs to the technical field of fracture prediction, and in particular relates to a method for predicting reservoir opening fractures. Background Art

[0002] With the continuous advancement of sandy conglomerate gas reservoirs, the prediction of natural effective fractures has become an important part of the efficient development of sandy conglomerate. In particular, oil and gas in fractured reservoirs will account for an increasingly large proportion of oil and gas reserves. Therefore, fractures are a key factor affecting production capacity.

[0003] Due to tectonic movement, there are a large number of faults and natural cracks in the study area, which are channels for groundwater and acidic solutions. Through the summary and study of this actual geological law, a technical method for predicting open cracks based on geological modeling is proposed. The accurate prediction of open cracks can play a greater control role in avoiding drilling engineering risks, improving the drilling rate of reservoirs, and fracturing reservoir transformation.

[0004] Currently, the existing methods for predicting open fractures mainly include shear wave splitting, microseismic, core thin sections and FMI imaging logging technology.

[0005] The main reservoir spaces of the conglomerate-bearing rocks in the study area are open fractures and dissolution pores, and the corresponding reservoir types are open fractures and induced fractures.

[0006] The most commonly used natural effective crack prediction technology at present is the earthquake longitudinal wave crack opening prediction technology.

[0007] Seismic P-wave crack opening prediction technology can usually be divided into two categories: pre-stack and post-stack seismic data prediction according to the type of seismic data.

[0008] The pre-stack P-wave crack prediction technology uses the mathematical method of pre-stack 3D seismic data to analyze the correlation between seismic trace signals and quantitatively describe the heterogeneity characteristics of the formation to predict crack opening. Due to the relatively high cost of pre-stack P-wave crack prediction technology, it is not currently used.

[0009] The post-stack P-wave open fracture prediction technology mainly studies the lateral discontinuity of seismic traces through seismic attributes and rose diagrams. Based on the discontinuity study, the characteristics of the fracture-pore dual medium reservoir caused by faults, small fractures, etc. are studied.

[0010] In summary, the methods of using seismic data to predict underground cracks are all related to pre-stack and post-stack three-dimensional seismic data. However, it is difficult to use these methods to predict cracks, and the operation process is easily affected by various factors such as poor signal-to-noise ratio of seismic data. Specifically, when using conventional post-stack seismic data for crack prediction, it is often difficult to predict micro cracks due to the influence of stack averaging. Summary of the invention

[0011] In view of the problems existing in the above-mentioned prior art, the present invention provides a method for predicting reservoir opening fractures. The technical problem to be solved by the present invention is to provide a method for effectively predicting opening fractures conveniently, simply and accurately.

[0012] In order to solve the above technical problems, the present invention provides a reservoir open fracture prediction method, comprising the following steps:

[0013] Step S1: using the original 3D seismic data volume, imaging logging data and geological data, a standardized data set of fracture information of the target layer is established;

[0014] Step S2: discriminating the standardized data set of target layer fracture information, retaining the values ​​of data points consistent with the open fractures, and establishing a quantitative modeling data body for the open fractures;

[0015] Step S3: Combine reservoir seismic inversion and merge open fracture quantitative modeling data volumes to obtain reservoir open fracture density.

[0016] Furthermore, the original 3D seismic data volume in step S1 includes three data volumes of fracture similarity attributes, dip angle and strike.

[0017] Furthermore, the fracture similarity attribute data of the original 3D seismic data volume is preprocessed. When the calculated fracture similarity attribute is left-biased, the preprocessing calculation formula is as follows:

[0018] likelihood=(1-semblance 8 ) 2 (1)

[0019] In the formula, likelihood is the similarity attribute value, and semblance is the skewness;

[0020] When the calculated fracture similarity attribute is right-biased, the preprocessing calculation formula is as follows:

[0021]

[0022] Furthermore, in step S1, the value range of the inclination and strike of the fracture corresponding to the original 3D seismic data volume is searched, and the strike and inclination of the open fracture are determined using imaging logging data and geological data to establish a standardized data set of fracture information of the target layer.

[0023] Furthermore, the inclination angle ranges from 0 to 180 degrees, and the strike angle ranges from -90 to 90 degrees.

[0024] Furthermore, the dip angle of the fracture in the original 3D seismic data volume is standardized to 0 to 90 degrees. If the strike is greater than 90 degrees, the corrected dip angle is used as the standardized dip angle data; if the dip angle is less than 90 degrees, it remains unchanged. The calculation formula of the corrected dip angle is:

[0025] B=180°-A

[0026] Where, B is the corrected inclination, °; ​​A is the original inclination, °;

[0027] The strike of the fractures in the original 3D seismic data volume is standardized to 0 to 90 degrees. If the strike is less than 0 degrees, the first revised strike is used as the standardized strike data; if the strike is greater than 0 degrees, the second revised strike is used as the standardized strike data. The calculation formulas for the first revised strike and the second revised strike are as follows:

[0028] C=90°+E

[0029] Where, C is the first corrected direction, °; ​​E is the original direction, °;

[0030] D=90°-E

[0031] Where, D is the second corrected direction, °; ​​E is the original direction, °;

[0032] The inclination and strike of the fractures in the imaging logging data are standardized to 0-90 degrees. If the inclination and strike of the open fractures in the imaging logging data are within different setting ranges, the corresponding data in the fracture information standardization data set is assigned as an open fracture; otherwise, it is a closed fracture. The setting range is set according to the actual situation of the region. For example, the setting range of the inclination in a certain area is 60-90 degrees, and the setting range of the strike is 0-45 degrees. For other areas, the setting range should be selected according to the actual situation.

[0033] Furthermore, in step S2, a logistic logic binary machine learning discrimination program is used to discriminate open cracks on the standardized data set of target layer crack information.

[0034] Furthermore, in step S3, the invalid mudstone is filtered out in combination with the reservoir inversion result, the reservoir open fracture information is retained, and the open fracture quantitative modeling data volume is output.

[0035] Furthermore, when the impedance value is less than or equal to the set value, it is mudstone, and the open fracture density plane distribution map is output. The set value is equal to 9500g.m / (cm 3 .s).

[0036] The reservoir open fracture prediction method of the present invention can effectively determine the development degree of natural effective fractures in sandstone-containing conglomerate, obtain information such as the direction of open fractures and fracture density, and thus provide rich fracture information for the study of fracture-type reservoirs.

[0037] The reservoir open fracture prediction method of the present invention can effectively solve the problem of identifying natural effective fractures in sandy conglomerate and has the following beneficial effects:

[0038] 1. Solved the problem of multiple solutions in the prediction of natural effective fractures in sandy conglomerate;

[0039] 2. Improved the success rate of deployment demonstration and pre-exploration wells;

[0040] 3. Improve the accuracy of the main controlling factors and oil and gas content of Permian conglomerate reservoirs;

[0041] 4. It laid a foundation for the efficient development of the Middle and Lower Permian conglomerate oil and gas reservoirs;

[0042] 5. Provide theoretical basis for the selection of corresponding reserve parameters in reserve declaration;

[0043] 6. The fracture sweet spots predicted by this technology can greatly promote the construction of reserve production capacity. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a flow chart of a reservoir opening fracture prediction method and device.

[0045] Figure 2 This is an effect diagram of the fusion of M-well open fracture prediction and seismic profile in a reservoir open fracture prediction method of the present invention.

[0046] Figure 3 It is a planar distribution diagram of the density of open fractures in a reservoir open fracture prediction method of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0050] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be directly connected or indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] In order to better understand the purpose, structure and function of the present invention, a reservoir opening fracture prediction method of the present invention is further described in detail below with reference to the accompanying drawings.

[0052] Embodiment 1:

[0053] like Figure 1 As shown, a reservoir opening fracture prediction method of the present invention comprises the following steps:

[0054] Step S1: using the original 3D seismic data volume, imaging logging data and geological data, a standardized data set of fracture information of the target layer is established;

[0055] Step S2: discriminating the standardized data set of target layer fracture information, retaining the values ​​of data points consistent with the open fractures, and establishing a quantitative modeling data body for the open fractures;

[0056] Step S3: Combine reservoir seismic inversion and merge open fracture quantitative modeling data volumes to obtain reservoir open fracture density.

[0057] Embodiment 2:

[0058] like Figure 1 As shown, a reservoir opening fracture prediction method of the present invention comprises the following steps:

[0059] Step S1: using the original 3D seismic data volume, imaging logging data and geological data, a standardized data set of all fracture information of the target layer is established;

[0060] Step S2: using a logistic logic binary machine learning discriminant program to discriminate the target layer open fracture data body, retaining the values ​​of data points consistent with the open fractures, and establishing the open fracture quantitative modeling data body;

[0061] Step S3: Combine reservoir seismic inversion and merge open fracture quantitative modeling data volumes to obtain open fracture density.

[0062] The difference between this embodiment and the first embodiment is that:

[0063] In step S1, the value range of dip and strike corresponding to the original 3D seismic data volume is searched, the strike and dip of the open fractures are determined using imaging logging data and geological data, and a standardized data set of all fracture information of the target layer is established.

[0064] In step S1, the original three-dimensional seismic data volume includes likelihood attributes, dip angles and strikes.

[0065] In step S1, the original 3D seismic data volume is subjected to a fracture identification algorithm, i.e., an analysis window of a certain analysis point is defined. Theoretically, the window can be of any shape. For the convenience of calculation, a regular window (e.g., 3x3) centered on the analysis point is usually used. The skewness is estimated based on the calculated similarity. When the calculated fracture similarity attribute is left-skewed, the preprocessing calculation formula is as follows:

[0066] likelihood=(1-semblance 8 ) 2 (1)

[0067] In the formula, likelihood is the similarity attribute value, and semblance is the skewness.

[0068] When the calculated fracture similarity attribute is right-biased, the preprocessing calculation formula is as follows:

[0069]

[0070] Although the value range of the fracture similarity attribute calculated for different three-dimensional data bodies is 0-1, the skewness is different, so standardization is required to ensure the uniformity of the output drawings.

[0071] That is, the calculated similarity values ​​are classified as normal distribution, otherwise the final crack plane map will not be consistent.

[0072] In step S1, according to the value range of fracture dip and strike of each sampling point of the original 3D seismic data volume, a discriminant statement is used to classify the specific value of the dip and strike of the fracture response of the logging data. The specific method is as follows:

[0073] IFC(dip<90°,dip,180°-dip)

[0074] IFC(stike>=0°,90°-strike,strike+90°)

[0075] Finally, the data with a dip angle of -90°-90° in the original 3D seismic data volume are standardized to dip data of 0°-90°; the data with a strike angle of 0°-180° are standardized to strike data of 0°-90°.

[0076] In step S1, three standardized data (inclination, strike, and similarity attributes) of the sampling points are output, and the input data format is as follows:

[0077] The format is: inclination (x, y, data value); trend (x, y, data value); similarity attribute (x, y, data value);

[0078] In step S1, the three data (dip, strike and likelihood) of the original three-dimensional seismic data volume are combined into a comma-separated data file for storage and loading according to the three-dimensional seismic data volume with Gauss Krige geodetic coordinates (without the tape number). The specific file format is as follows:

[0079] The format is: (x, y, dip, strike, similarity), i.e. fracture_dip_strike.csv.

[0080] X,Y,dip,,strike,,likelihood

[0081] 334314,4980831.5,79.761,,28.97,,1

[0082] 334314,4980856.5,61.828,,21.77,,1

[0083] 334314,4980881.5,80.477,,57.543,,1

[0084] 334314,4980906.5,53.818,,56.328,,0.499

[0085] 334314,4980931.5,41.929,,43.257,,0.501

[0086] 334314,4980956.5,62.898,,21.77,,0.498

[0087] 334339，4980806.5，57.193，，36.057，，0.499

[0088] 334339，4980831.5，53.532，，14.287，，0.5

[0089] 334339，4980856.5，44.425，，48.569，，0.501

[0090] 334339，4980881.5，39.046，，64.63，，1

[0091] 334339，4980906.5，81.986，，57.543，，0.5

[0092] 334339，4980931.5，88.622，，57.827，，0.501

[0093] 334339，4980956.5，65.209，，21.77，，0.5

[0094] 334339，4980981.5，58.359，，21.77，，0.499

[0095] 334339，4981006.5，81.061，，7.483，，0.499

[0096] 334339，4981031.5，89.988，，0，，0.498

[0097] 334339，4981056.5，57.206，，36.057，，0.494

[0098] 334339，4981081.5，57.586，，36.057，，0.496

[0099] 334339，4981106.5，65.87，，75.595，，0.496

[0100] 334339，4981131.5，66.87，，21.77，，0.781

[0101] 334339，4981156.5，66.463，，7.483，，0.493

[0102] 334339，4981181.5，47.309，，0，，0.491

[0103] 334339,4981206.5,50.483,,0,,0.491

[0104] 334339,4981231.5,89.859,,36.057,,0.491

[0105] 334339,4981256.5,65.744,,50.343,,0.492

[0106] 334339,4981281.5,44.293,,50.343,,0.492

[0107] 334339,4981306.5,45.758,,50.343,,0.989

[0108] 334339,4981331.5,42.997,,56.828,,0.495

[0109] In step S1, the imaging logging data and geological data are used to determine the direction and inclination of the open fractures, and the judgment is made through the following python statement:

[0110] Import pandas as pd

[0111] data=pd.read_csv("logfracture_dip_strike.csv')

[0112] df = pd.DataFrame(data)

[0113] print(df.columns, df.shape)

[0114] / / Input three 3D seismic data volumes with Gauss Krige geodetic coordinates (without band numbers).

[0115] Import pandas as pd

[0116] Import matplotlib.pyplot as plt

[0117] data=pd.read_csv("fracture_dip_strike.csv')

[0118] df = pd.DataFrame(data)

[0119] color=[]

[0120] for i in df['class'][0:100]:

[0121] ifi == 'open':

[0122] color.append('red')

[0123] else:

[0124] color.append('blue')

[0125] plt.scatter(df['dip'][0:100]), df['strike'][0:100], color=color)

[0126] plt.xlabel('dip')

[0127] plt.ylabel('strike')

[0128] plt.show()

[0129] The above python statements are used to compare the dip and strike of the three data of the three-dimensional seismic data body with the imaging logging data and geological data using "shape" and "color". The dip and strike of the open fractures identified by the imaging logging data are used as the sampling point information of the open fractures in the seismic data body. The specific judgment method is to use the above python statements to determine whether the dip and strike are within a certain range as open fractures, and whether they are not within this range as closed fractures.

[0130] In step S2, the logistic binary machine learning discrimination program (python statement) is input as follows: the data of all fracture data volumes of the target layer are input into the program to perform seismic open fracture data discrimination.

[0131] The python statement is as follows:

[0132] import numpy as np

[0133] import pandas as pd

[0134] data=pd.read_csv('fracture_dip_strike.csv')

[0135] df = pd.DataFrame(data)

[0136] label=[]

[0137] for I in df['fracture'][0:400]: / / For example: input the first 400 sample points for training

[0138] if i == 'dip':

[0139] Label.append(θ)

[0140] Else:

[0141] Label.append(1)

[0142] X1=df['dip'][0:400]

[0143] X2=df['strike'][0:400]

[0144] Train_data=list(zip(x1,x2,label))

[0145] Class Logistic_Regression: / / python statement package, logistic binary discrimination Def_init_(self,traindata,alpha=0.01,circle=10,batchlength=20)

[0146] Self.traindata = traindata / / training data set

[0147] Self.alpha = alpha / / learning rate is 0.01

[0148] Self.circle = circle / / The number of learning times is 10 times

[0149] Self.batchlength = batchlength / / Divide 400 data into 20 parts, 20 data in each part

[0150] Self.w = np.random.normal (size = (3,1)) / / Randomly initialize parameter w

[0151] Def data_process(self): / / random gradient descent, disrupt the data order, and divide all data into several batches

[0152] Np.random.shuffle(self.traindata)

[0153] Data=[self.traindata[i:i+self.batchlength]]

[0154] For I in range(0,len(self.traindata),self.batchlength)]

[0155] Return data

[0156] Def train1(self):

[0157] Def train1(self):

[0158] / / Perform stochastic gradient descent based on loss function 1

[0159] For I in range(self.circle):

[0160] Batches = self.data_process()

[0161] Print('the{}epoch'.format(i))

[0162] For batch in batches:

[0163] D_w = np.zeros(shape+(3,1))

[0164] For j in batch:

[0165] X0=np.mat(x0).T

[0166] Y=j[2]

[0167] Dw=(self.sigmoid(self.wT*x)-y)[0,0]*x

[0168] D_w+=dw

[0169] Self.w-=self.alpha*d_w / self.batchlength

[0170] Gradient descent solution for the root mean square loss function

[0171] Def train2(self):

[0172] For I in range(self.circle):

[0173] Batchs=self.data_process()

[0174] Print(‘the{}epoch’.format(i))

[0175] For batch in batches:

[0176] D_w=np.zeros(shape=(3,1)

[0177] For j in batch:

[0178] X0=np.r_[j[0:2],1]

[0179] X=np.mat(x0).T

[0180] Y=j[2]

[0181] Dw=((self.sigmoid(self.w.T*x)-y)*self.sigmoid(self.w.T*x)*(1-self.sigmoid(self.w.T*x)))[0.0]*x

[0182] D_w+=dw

[0183] Self.w-=self.alpha*d_w / self.batchlength

[0184] Def sigmoid(self,x):

[0185] Return 1 / (1+np.exp(-x))

[0186] Def predict(self,x):

[0187] S=self.sigmoid(self.w.T*x)

[0188] If s>=0.5:

[0189] Return 1

[0190] Elif s<0.5:

[0191] Return o

[0192] If__name__==’__main__’:

[0193] Regr=Logistic_Regression(traindata=train_data)

[0194] Regr.train1()

[0195] If it is an open crack, the similarity attribute value is retained. If not, the similarity attribute value is reset to zero. The judgment of the open crack is performed using the Python statement as follows:

[0196] If fracture=1

[0197] Return fracture

[0198] Elif fracture=0:

[0199] Return

[0200] In step S3, invalid mudstone is filtered out according to the reservoir inversion result, and the reservoir open fracture information is retained.

[0201] Mudstone can be filtered in the following ways:

[0202] if(sand_IMP>11500)

[0203] {fracture = fracture;}

[0204] else

[0205] {fracture=0;}

[0206] Impedance value greater than 11500g.m / (cm 3 .s) is sandy conglomerate, otherwise, it is mudstone.

[0207] Embodiment 3:

[0208] like Figure 2 and Figure 3 As shown, the present invention has been well applied in a certain well area M, and the specific contents include:

[0209] First, from the 3D seismic data volume SEGY file, the output format is a 32-bit floating point format with 12 headers and amplitude-preserving and stacked 3D seismic data volume.

[0210] The seismic data were processed using dip imaging enhancement, and high-resolution data volumes were generated according to large- and medium-scale faults of earthquake levels;

[0211] By extracting the dip, strike and similarity attributes of the high-resolution data volume, three seismic structural fracture attribute data volumes are obtained.

[0212] Second, under the control of the interpreted horizon, the target layer segment time window (dip, strike and similarity attributes) is extracted and output as a standardized data set of target layer fracture information.

[0213] Third, according to the strike and dip of the open fractures determined by the imaging logging data and geological data, the similarity attributes of the open fractures in the standardized data set of all fracture information are retained under the python statement logistic binary discrimination program, and the rest are filtered, so as to establish the quantitative modeling data body of the open fractures in the target layer.

[0214] According to the impedance measurement of sandstone and mudstone of Fengcheng Formation in the Permian System in the study area, it is greater than 10500g.m / (cm 3 .s) can be judged as sandy conglomerate, and less than 9500g.m / (cm3.s) can be basically judged as mudstone. This study simulates the opening of cracks in sandy conglomerate with impedance greater than 10500g.m / (cm3.s).

[0215] When conducting well location demonstration, there is only one exploration well, CP24, in the Fengcheng Formation of the Permian System in this area. Fractures are well developed. Well logging interpretation shows that the inclination of the opening fractures is 65 to 90 degrees and the strike is 0 to 40 degrees. Seismic opening fractures are well developed. After the above steps, the density of seismic opening fractures in the sandy conglomerate in this area is predicted. The results are as follows: Figure 3 As shown in the figure, three favorable targets with large open fracture density, SP3, SP4 and SP5, are selected for deployment and drilling. The drilling results show that these three wells are all located in the location where open fractures are developed, have good oil and gas shows, and obtain industrial oil flow. This method of predicting open fractures in deep reservoirs provides important technical support for providing drilling targets and optimizing well locations.

[0216] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A method for predicting reservoir opening fractures, characterized in that The following steps are involved: Step S1: using the original 3D seismic data volume, imaging logging data and geological data, a standardized data set of fracture information of the target layer is established; Step S2: discriminating the standardized data set of target layer fracture information, retaining the values ​​of data points consistent with the open fractures, and establishing a quantitative modeling data body for the open fractures; Step S3: Combine reservoir seismic inversion and merge open fracture quantitative modeling data volumes to obtain reservoir open fracture density.

2. The reservoir open fracture prediction method according to claim 1, characterized in that: The original 3D seismic data volume in step S1 includes three data volumes of fracture similarity attributes, dip angle and strike.

3. The reservoir open fracture prediction method according to claim 2, characterized in that: The fracture similarity attribute data of the original 3D seismic data volume is preprocessed. When the calculated fracture similarity attribute is left-biased, the preprocessing calculation formula is as follows: likelihood=(1-semblance 8 ) 2 (1) In the formula, likelihood is the similarity attribute value, and semblance is the skewness; When the calculated fracture similarity attribute is right-biased, the preprocessing calculation formula is as follows:

4. The reservoir opening fracture prediction method according to claim 1, characterized in that: In step S1, the value range of the inclination and strike of the fracture corresponding to the original 3D seismic data volume is searched, the strike and inclination of the open fracture are determined using imaging logging data and geological data, and a standardized data set of fracture information of the target layer is established.

5. The reservoir opening fracture prediction method according to claim 4, characterized in that: The dip angle has a value range of 0 to 180 degrees, and the strike angle has a value range of -90 to 90 degrees.

6. The reservoir open fracture prediction method according to claim 5, characterized in that: The dip angle of the fracture in the original 3D seismic data volume is standardized to 0-90 degrees. If the strike is greater than 90 degrees, the corrected dip angle is used as the standardized dip angle data; if the dip angle is less than 90 degrees, it remains unchanged. The calculation formula for the corrected dip angle is: B=180°-A Where, B is the corrected inclination, °; ​​A is the original inclination, °; The strike of the fractures in the original 3D seismic data volume is standardized to 0 to 90 degrees. If the strike is less than 0 degrees, the first revised strike is used as the standardized strike data; if the strike is greater than 0 degrees, the second revised strike is used as the standardized strike data. The calculation formulas for the first revised strike and the second revised strike are as follows: C=90°+E Where, C is the first corrected direction, °; ​​E is the original direction, °; D=90°-E Where, D is the second corrected direction, °; ​​E is the original direction, °; The inclination and strike of the fractures in the imaging logging data are standardized to 0 to 90 degrees. If the inclination and strike of the open fractures in the imaging logging data are within different setting ranges, the corresponding data in the fracture information standardized data set is assigned as an open fracture; otherwise, it is a closed fracture. The setting range is set according to the actual situation in the region.

7. The reservoir open fracture prediction method according to claim 1, characterized in that: In step S2, a logistic binary machine learning discriminant program is used to discriminate open cracks on the target layer crack information standardized data set.

8. The reservoir open fracture prediction method according to claim 1, characterized in that: In step S3, the invalid mudstone is filtered out in combination with the reservoir inversion result, the reservoir open fracture information is retained, and the open fracture quantitative modeling data volume is output.

9. The reservoir opening fracture prediction method according to claim 8, characterized in that: If the impedance value is less than or equal to the set value, it is mudstone, and the plane distribution map of open fracture density is output.

Images