A method for distinguishing the stages of shale structural fractures

Abstract

The present invention discloses a method for distinguishing the stages of shale structural fractures. First, different types of shale fracture images are obtained, and characteristic data of the fractures therein is extracted and classified into fracture categories, and the fracture characteristic data is output. The obtained fracture characteristic data is then normalized. A fracture discrimination model based on a BP neural network is then constructed, and the normalized fracture characteristic data is used as training samples for training. Finally, data acquisition equipment is deployed to randomly sample fracture data in the area to be tested and extract its characteristic quantities. The extracted characteristic quantities are input into the trained fracture discrimination model, and the discrimination result of the fracture stage in the area to be tested is output. Based on this solution, the BP neural network is used to train existing fracture characteristics, breaking through the problem that traditional imaging well methods require a large amount of core data as a basis to achieve rock fracture stage discrimination, and making up for the lack of coring data.

Description

Technical Field

[0001] The present invention relates to the field of geology, and in particular to a method for distinguishing the stages of shale structural fractures. Background Art

[0002] Downhole fractures are usually identified through coring and logging interpretation methods. Due to the limited number of cores, most fractures are identified through logging data. Imaging logging data has the advantage of two-dimensional imaging of the wellbore. Since resistivity imaging logging data reflects the information within the depth range of the wellbore, there are still many uncertainties and multiple solutions compared with core data, which requires a lot of core calibration and rich interpretation experience to complete.

[0003] Existing well logging interpretation methods primarily focus on identifying natural fractures, classifying them, calculating fracture parameters, evaluating fracture effectiveness, interpreting fracture parameters, and qualitatively evaluating their contribution to productivity. However, interpreting fracture stages using well logging data is limited to natural fractures and drilling-induced fractures, and there is no effective method for studying the stages of natural fractures. Geological studies of fracture stages primarily rely on core and field outcrop observations and analysis, requiring a significant amount of core data to obtain a clear picture. Imaged well data, on the other hand, provide insightful information, making fracture stage analysis using well logging data extremely challenging. Summary of the Invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for distinguishing the stages of shale structural fractures.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] A method for distinguishing the stages of shale structural fractures comprises the following steps:

[0007] S1. Acquire different types of shale fracture images, extract characteristic data of the fractures, classify the fracture types, and output the fracture characteristic data;

[0008] S2, normalizing the crack characteristic data obtained in step S1;

[0009] S3, constructing a crack discrimination model based on BP neural network, using the crack feature data normalized in step S2 as training samples for training;

[0010] S4. Deploy data acquisition equipment, randomly sample crack data in the area to be tested and extract its features, input the extracted features into the crack discrimination model trained in step S3, and output the discrimination result of the crack stage in the area to be tested.

[0011] Furthermore, the characteristic data of the cracks in S1 are extracted by using a threshold-based crack segmentation method, specifically in the following manner:

[0012] S11, a crack image Divided into different areas, among which is the pixel area in the image;

[0013] S12, selecting an image containing cracks in the divided pixel area using a threshold method;

[0014] S13. Extract crack features from the selected image containing cracks, and output crack dip, inclination, and strike data.

[0015] Furthermore, the specific method of S12 is:

[0016] S121. Calculate crack image Grayscale, and divide it into Levels, and the total number of pixels is obtained at the same time;

[0017] S122. Setting selection threshold , according to the selected threshold Divide the pixels in the image into and Two groups, and calculate the probability of each group;

[0018] S123, using the threshold selection function to calculate the square difference threshold If you choose the threshold Maximize the square difference threshold and output the selected threshold ;

[0019] S124: Select threshold value according to the output of S123 The image containing cracks is divided into pixels whose grayscale level is less than the selected threshold as the background image area, and pixels whose grayscale level is higher than the selected threshold as the crack image area;

[0020] S125. Extracting crack features from the crack area image, including crack inclination, dip, and strike data features.

[0021] Furthermore, the method for calculating the probability of each group being generated in S122 is:

[0022] ;

[0023] ;

[0024] in, for The classification probability of a group of pixels, Choose a function for the probability, is the probability of each point appearing, for Below the selection threshold The number of pixels; for The classification probability of a group of pixels, Number the pixels.

[0025] Furthermore, the threshold selection function in S123 is expressed as:

[0026] ;

[0027] in, for The average grayscale value of the pixel points in the group, for The average grayscale value of the pixel points in the group, is the average grayscale value of all pixels, Select a function for the threshold.

[0028] Furthermore, the specific calculation method of S123 is:

[0029] ;

[0030] like Make The maximum value is output as the selection threshold.

[0031] Furthermore, the normalization method in S2 is:

[0032] ；

[0033] in, is the standardized value, The value range is [1,3], which respectively represents the crack inclination, dip and strike data. Represents the original attribute value, and Represents the minimum and maximum values of the original attribute respectively.

[0034] Furthermore, the number of neurons in the input layer of the crack discrimination model in S3 is 3, which input the data features of the crack inclination, dip angle and strike respectively. The number of neurons in the hidden layer is set to 3-7, and the number of neurons in the output layer is 2 or 4. Specifically:

[0035] The output of each neuron in the input layer is expressed as ,in, is the neuron number of the input layer, For the The standardized value input by the input layer neurons;

[0036] The input of the hidden layer is represented as ,in, is the hidden layer neuron number, is the transfer function between the input layer and the hidden layer, is the correction coefficient of the input layer neurons;

[0037] The hidden layer is represented as the output , For the training function trainlm;

[0038] The input of the output layer is represented as: ,in, is the sequence number of the output layer neuron, is the transfer function from the hidden layer to the output layer, is the correction coefficient of the hidden layer neurons;

[0039] The loss function is expressed as: .

[0040] Furthermore, the output result of the crack discrimination model is expressed as:

[0041] ;

[0042] in, is the output function of the discriminant model, , respectively represent the strike, dip and inclination data of the fracture, They are the proportions corresponding to the strike, dip and inclination data of the cracks respectively.

[0043] Furthermore, the crack discrimination model outputs different crack stage levels according to the number of neurons in the output layer. When the output layer is 2, the output crack stage levels are development and non-development levels, and the output results are:

[0044] ;

[0045] When the output layer is 4, the output is extremely underdeveloped, relatively developed, developed and very developed. The output result is:

[0046] .

[0047] The present invention has the following beneficial effects:

[0048] 1. The BP neural network is used to train the existing fracture characteristics, which breaks through the problem that the traditional imaging well method requires a large amount of core data as a basis to realize the identification of rock fracture stages, and makes up for the lack of coring data.

[0049] 2. The threshold method is used to extract and calculate the characteristics of rock fractures, which can make more accurate calculations on the obtained rock fracture image data, making the final prediction results more accurate.

[0050] 3. Reasonable setting of the BP neural network structure solves the problem that the discriminant model cannot obtain enough information from existing samples to reflect the basic rules. At the same time, it can fix the noise data in the samples to prevent oversaturation problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a schematic diagram of a method for distinguishing the stages of shale structural fractures according to the present invention.

[0052] Figure 2 This is a structural diagram of a BP neural network according to an embodiment of the present invention.

[0053] Figure 3 Another structural diagram of the BP neural network according to an embodiment of the present invention is shown in FIG. DETAILED DESCRIPTION

[0054] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0055] A method for distinguishing the stages of shale structural fractures, such as Figure 1 As shown, the following steps are included:

[0056] S1. Acquire different types of shale fracture images, extract characteristic data of the fractures, classify the fracture types, and output the fracture characteristic data;

[0057] Shale fracture images contain a variety of fracture feature data, including fractal data, fracture rate, fracture direction, fracture dip, etc. In this embodiment, the fracture direction, dip, and inclination that can be more intuitively displayed in the image are used as basic data to make the entire image recognition more accurate. The specific feature extraction method is as follows:

[0058] S11, a crack image Divided into different areas, among which is the pixel area in the image;

[0059] S12, selecting an image containing cracks in the divided pixel area using a threshold method;

[0060] In this embodiment, the Otsu threshold method is used for segmentation. The threshold of this method can be automatically selected. The essential idea is to maximize the inter-class variance between different sub-regions or minimize the intra-class variance within the same region based on the difference in grayscale values between the crack target and the background area. From a mathematical point of view, the variance of an image is equal to the sum of the inter-class variance and the intra-class variance. Therefore, when the inter-class variance value is the largest, the corresponding intra-class variance value must be the smallest.

[0061] When using the Otsu threshold method for crack image segmentation, if a crack target is mistakenly regarded as the background or some background pixels are mistakenly segmented as cracks, the result will be inaccurate. Therefore, when the inter-class variance between the target area and the background area of the image is the largest, the probability of misclassification during segmentation is minimized. The specific method is:

[0062] S121. Calculate crack image Grayscale, and divide it into Levels, and the total number of pixels is obtained at the same time;

[0063] S122. Setting selection threshold , according to the selected threshold Divide the pixels in the image into and There are two groups, and the probability of each group is calculated separately. The probability of each group is:

[0064] ;

[0065] ;

[0066] in, for The classification probability of a group of pixels, Choose a function for the probability, is the probability of each point appearing, for Below the selection threshold The number of pixels; for The classification probability of a group of pixels, Number the pixels.

[0067] S123, using the threshold selection function to calculate the square difference threshold If you choose the threshold Maximize the square difference threshold and output the selected threshold ;

[0068] The threshold selection function is expressed as:

[0069] ;

[0070] in, for The average grayscale value of the pixel points in the group, for The average grayscale value of the pixel points in the group, The average grayscale value of all pixels. Select a function for the threshold.

[0071] like Make The maximum value is output As the selection threshold, it is expressed as:

[0072] .

[0073] S124: Select threshold value according to the output of S123 The image containing cracks is divided into pixels whose grayscale level is less than the selected threshold as the background image area, and pixels whose grayscale level is higher than the selected threshold as the crack image area;

[0074] S125. Extracting crack features from the crack area image, including crack inclination, dip, and strike data features.

[0075] S13. Extract crack features from the selected image containing cracks, and output crack dip, inclination, and strike data.

[0076] S2, normalizing the crack characteristic data obtained in step S1;

[0077] The normalization method is:

[0078] ；

[0079] in, is the standardized value, The value range is [1,3], which respectively represents the crack inclination, dip and strike data. Represents the original attribute value, and Represents the minimum and maximum values of the original attribute respectively.

[0080] S3, constructing a crack discrimination model based on BP neural network, using the crack feature data normalized in step S2 as training samples for training;

[0081] The number of neurons in the input layer of the crack discrimination model in S3 is 3, which inputs the data features of the crack inclination, dip angle and strike respectively. The number of neurons in the hidden layer is set to 3-7, and the number of neurons in the output layer is 2. Figure 2 As shown, or 4, such as Figure 3 As shown, specifically:

[0082] The output of each neuron in the input layer is expressed as ,in, is the neuron number of the input layer, For the The standardized value input by the input layer neurons;

[0083] The input of the hidden layer is represented as ,in, is the hidden layer neuron number, is the transfer function between the input layer and the hidden layer, is the correction coefficient of the input layer neurons;

[0084] The hidden layer is the output represented as , For the training function trainlm;

[0085] The input of the output layer is represented as: ,in, is the sequence number of the output layer neuron, is the transfer function from the hidden layer to the output layer, is the correction coefficient of the hidden layer neurons;

[0086] The loss function is expressed as: .

[0087] The neural network design can be completed in Matlab. The transfer function is set to tansig and logsig respectively. The training function can use Traingd, Traingdm and trainlm. The training error is 0.001. The number of hidden layer neurons is 3, 5, 7 and 9 respectively. When the output result is 4 levels, the number of hidden layer neurons is 7. When the training function uses traingdx or trainm, it can converge to the required error accuracy, while trainlm has the fastest convergence speed. When the output result is 2 levels, such as Figure 2 As shown, when the training function is trainlm and the number of neurons is 7 or 9, the network can achieve a thin face and a given accuracy. Therefore, this embodiment can consider using a network model with 4 output categories, 7 hidden layer neurons, trainlm as the training function, and tandig and logsig as the transfer functions to construct a BP network.

[0088] The output of the crack discrimination model is expressed as:

[0089] ;

[0090] in, is the output function of the discriminant model, , respectively represent the strike, dip and inclination data of the fracture, They are the proportions corresponding to the strike, dip and inclination data of the cracks respectively.

[0091] Different crack stage levels are output according to the number of neurons in the output layer. When the output layer is 2, the output crack stage levels are development and non-development levels. The output results are:

[0092] ;

[0093] When the output layer is 4, the output is extremely underdeveloped, relatively developed, developed and very developed. The output result is:

[0094] .

[0095] S4. Deploy data acquisition equipment, randomly sample crack data in the area to be tested and extract its features, input the extracted features into the crack discrimination model trained in step S3, and output the discrimination result of the crack stage in the area to be tested.

[0096] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0097] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0099] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

[0100] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A method for distinguishing the stages of shale structural fractures, characterized in that: The steps include: S1. Acquire different types of shale fracture images, extract characteristic data of the fractures, classify the fracture types, and output the fracture characteristic data. The fracture characteristic data is extracted using a threshold-based fracture segmentation method, specifically as follows: S11, a crack image Divided into different areas, among which is the pixel area in the image; S12, using a threshold method to select an image containing cracks in the divided pixel area, specifically in the following manner: S121. Calculate crack image Grayscale, and divide it into levels, and the total number of pixels is obtained at the same time; S122. Setting selection threshold , according to the selected threshold Divide the pixels in the image into and Two groups, and calculate the probability of each group separately, the specific method is: ; ; in, for The classification probability of a group of pixels, Choose a function for the probability, is the probability of each point appearing, for Below the selection threshold The number of pixels; for The classification probability of a group of pixels, Number the pixels; S123, using the threshold selection function to calculate the square difference threshold If you choose the threshold Maximize the square difference threshold and output the selected threshold , the threshold selection function is expressed as: ； in, for The average grayscale value of the pixel points in the group, for The average grayscale value of the pixel points in the group, is the average grayscale value of all pixels, Select a function for the threshold; S124: Select threshold value according to the output of S123 The image containing cracks is divided into pixels whose grayscale level is less than the selected threshold as the background image area, and pixels whose grayscale level is higher than the selected threshold as the crack image area; S125, extracting crack features from the crack area image, including crack inclination, dip angle, and strike data features; S13, extracting crack features from the selected image containing cracks, and outputting crack dip, inclination, and strike data; S2, normalizing the crack characteristic data obtained in step S1; S3, constructing a crack discrimination model based on BP neural network, using the crack feature data normalized in step S2 as training samples for training; S4. Deploy data acquisition equipment, randomly sample crack data in the area to be tested and extract its features, input the extracted features into the crack discrimination model trained in step S3, and output the discrimination result of the crack stage in the area to be tested.

2. A method for distinguishing the stages of shale structural fractures according to claim 1, characterized in that: The S123 square difference threshold The specific calculation method is: ； like Make The maximum value is output as the selection threshold.

3. A method for distinguishing the stages of shale structural fractures according to claim 2, characterized in that: The normalization method in S2 is: ； in, is the standardized value, The value range is [1,3], which respectively represents the crack inclination, dip and strike data. Represents the original attribute value, and Represents the minimum and maximum values of the original attribute respectively.

4. A method for distinguishing the stages of shale structural fractures according to claim 3, characterized in that: The number of neurons in the input layer of the crack discrimination model in S3 is 3, which input the data features of the crack's inclination, dip angle and strike respectively. The number of neurons in the hidden layer is set to 3-7, and the number of neurons in the output layer is 2 or 4. Specifically: The output of each neuron in the input layer is expressed as ,in, is the neuron number of the input layer, For the The normalized value input by the input layer neurons; The input of the hidden layer is represented as ,in, is the hidden layer neuron number, is the transfer function between the input layer and the hidden layer, is the correction coefficient of the input layer neurons; The hidden layer is represented as the output , For the training function trainlm; The input of the output layer is represented as: ,in, is the sequence number of the output layer neuron, is the transfer function from the hidden layer to the output layer, is the correction coefficient of the hidden layer neurons; The loss function is expressed as: .

5. A method for distinguishing the stages of shale structural fractures according to claim 4, characterized in that: The output result of the crack discrimination model is expressed as: ; in, is the output function of the discriminant model, , respectively represent the strike, dip and inclination data of the fracture, They are the proportions corresponding to the strike, dip and inclination data of the cracks respectively.

6. A method for distinguishing the stages of shale structural fractures according to claim 5, characterized in that: The crack discrimination model outputs different crack stage levels according to the number of neurons in the output layer. When the output layer is 2, the output crack stage levels are development and non-development levels. The output results are: ； When the output layer is 4, the output is extremely underdeveloped, relatively developed, developed and very developed. The output result is: 。

Images