Method and system for fracture segmentation and identification of imaging logging images based on machine learning

Abstract

The present invention relates to a method and system for segmenting and identifying fractures in imaging logging images based on machine learning. The method comprises the following steps: preprocessing imaging logging image data; training a constructed image segmentation and recognition model based on the preprocessed imaging logging image data to obtain a trained image segmentation and recognition model; using the trained image segmentation and recognition model to identify the imaging logging image to be identified to obtain the pore and fracture structure of the imaging logging image to be identified. The present invention uses a K-means clustering algorithm and a genetic algorithm to gradually optimize the threshold segmentation of the image, thereby reducing the noise points in the image after image segmentation as much as possible, so that the target area in the image is more prominent. Image segmentation makes the image quality higher, and in neural network image recognition, a higher recognition accuracy can be guaranteed when the number of samples is small. Therefore, the present invention can be widely used in the field of oil and gas well engineering.

Description

Technical Field

[0001] The invention relates to a method and system for segmenting and identifying fractures in imaging well logging images based on machine learning, and belongs to the field of oil and gas well engineering. Background Art

[0002] Fractures are important underground storage spaces and migration channels for oil and gas. Drilling accidents such as collapse and leakage are prone to occur in the area where the fractures are located. Therefore, determining whether there are cracks around the well wall is a very important research direction in drilling engineering, and it is also a major problem faced by this field.

[0003] Experts and scholars at home and abroad have proposed many methods for identifying cracks, which can be divided into two categories:

[0004] The first type is fracture identification based on traditional methods. This method mainly relies on the characteristic that natural fractures approximate a sine curve in the imaging logging image, and identifies the sine curve through methods such as Hough transform, and then identifies the natural fractures. However, in practical problems, the natural fractures in the measured imaging logging images often appear intermittently, and it is very difficult to identify natural fractures by identifying sine curves.

[0005] The second category is to identify fractures using image recognition algorithms based on machine learning. Among these methods, fracture identification is mainly based on neural networks. However, neural networks have high requirements for samples, requiring both a large number of samples and high sample quality. However, in actual problems, there are not many high-quality imaging logging images. When there are not many high-quality samples, the accuracy of fracture identification using neural networks is often low. Summary of the invention

[0006] In view of the above problems, the purpose of the present invention is to provide a method and system for segmenting and identifying fractures in imaging logging images based on machine learning, which can improve image quality through image segmentation when the number of samples is small, thereby applying neural networks to effectively identify fractures in small sample conditions.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for segmenting and identifying fractures in imaging logging images based on machine learning, comprising the following steps:

[0009] Preprocessing of imaging logging image data;

[0010] Based on the preprocessed imaging logging image data, the constructed image segmentation recognition model is trained to obtain a trained image segmentation recognition model;

[0011] The trained image segmentation recognition model is used to recognize the imaging logging image to be recognized, and the fracture recognition result of the imaging logging image to be recognized is obtained.

[0012] Preferably, a gamma transform method is used when preprocessing the imaging logging image.

[0013] Preferably, the gamma transform method includes normalization, pre-compensation and denormalization.

[0014] Preferably, the normalized calculation formula is as follows:

[0015]

[0016] Among them, R i is the normalized gray value of the i-th pixel, r i is the original grayscale value of the i-th pixel, r min and r max They are the minimum and maximum grayscale values ​​of all pixels respectively.

[0017] Preferably, the calculation formula of the pre-compensation is:

[0018]

[0019] Among them, G i is the value corresponding to the i-th pixel after gamma transformation, R i is the normalized grayscale value of the ith pixel, and γ is the gamma transform index.

[0020] Preferably, the method of training the constructed image segmentation and recognition model based on the preprocessed imaging logging image data to obtain the trained image segmentation and recognition model comprises:

[0021] Performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data;

[0022] The segmented imaging logging image data is randomly divided to obtain a training set and a test set;

[0023] The training set is used to train the image segmentation and recognition module, and the test set is used to test the effect of the image segmentation and recognition model to obtain a trained image segmentation and recognition model.

[0024] Preferably, the method of performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data comprises:

[0025] The K-means clustering algorithm is used to perform cluster segmentation on the preprocessed imaging logging image data;

[0026] The threshold segmentation method is used to perform threshold segmentation on the image after clustering segmentation.

[0027] Preferably, before clustering and segmenting the pre-processed imaging logging image data using the K-means clustering algorithm, the following steps are also included:

[0028] Genetic algorithm is used to determine the initial cluster centers.

[0029] Preferably, after performing threshold segmentation on the image after cluster segmentation by using the threshold segmentation method, the method further comprises the following steps: filtering the threshold segmentation result by using a median filtering method.

[0030] An imaging logging image fracture segmentation and identification system based on machine learning, comprising:

[0031] A preprocessing module, used for preprocessing imaging logging image data;

[0032] A model training module is used to train the constructed image segmentation recognition model based on the preprocessed imaging logging image data to obtain a trained image segmentation recognition model;

[0033] The fracture recognition module is used to recognize the imaging logging image to be recognized by using the trained image segmentation recognition model to obtain the pore and fracture structure of the imaging logging image to be recognized.

[0034] The present invention adopts the above technical solution, which has the following advantages:

[0035] 1. The present invention uses K-means clustering algorithm and genetic algorithm to gradually optimize the threshold segmentation of the image, thereby reducing the noise points in the image after image segmentation as much as possible, making the target area in the image more prominent.

[0036] 2. The present invention first performs image segmentation, and through image segmentation, the image quality is improved. When performing neural network image recognition, a higher recognition accuracy can be guaranteed when the number of samples is small compared with directly using a neural network for image recognition.

[0037] Therefore, the present invention can be widely applied to the field of oil and gas well engineering. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of a method for segmenting and identifying fractures in imaging logging images based on machine learning provided in one embodiment of the present invention;

[0039] Figure 2 is an image after image preprocessing in one embodiment of the present invention;

[0040] Figure 3 is an image after threshold segmentation filtering in one embodiment of the present invention;

[0041] Figure 4 is a clustering result diagram of K-means clustering optimized by genetic algorithm in one embodiment of the present invention;

[0042] Figure 5 is an image segmented by K-means clustering threshold optimized by genetic algorithm in one embodiment of the present invention;

[0043] Figure 6 It is a fracture identification result of an imaging logging long map of a production well in one embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0045] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0046] In some embodiments of the present invention, a method for segmenting and identifying fractures in imaging logging based on machine learning is provided, comprising the following steps: image preprocessing; image threshold segmentation; applying a K-means clustering algorithm to optimize the threshold segmentation, and applying a genetic algorithm to optimize the K-means clustering threshold segmentation again, and performing image segmentation using the optimized model; inputting the high-quality image after image segmentation into a deep neural network model for fracture identification. The present invention first performs image segmentation, and image segmentation makes the image quality higher. When performing neural network image recognition, a higher recognition accuracy can be guaranteed when the number of samples is small compared with directly using a neural network for image recognition.

[0047] Correspondingly, some other embodiments of the present invention provide an imaging logging image fracture segmentation and identification system based on machine learning.

[0048] Example 1

[0049] like Figure 1 As shown, this embodiment provides a method for segmenting and identifying fractures in imaging logging images based on machine learning, comprising the following steps:

[0050] Step S1, preprocessing imaging logging image data;

[0051] Step S2: training the constructed image segmentation recognition model based on the preprocessed imaging logging image data to obtain a trained image segmentation recognition model;

[0052] Step S3: using the trained image segmentation recognition model to recognize the imaging logging image to be recognized, and obtaining the fracture recognition result of the imaging logging image to be recognized.

[0053] Furthermore, in the above step S1, the process of preprocessing the imaging logging image is: applying gamma transformation to perform image enhancement processing, so that the fracture area in the imaging logging image is clearer than other areas.

[0054] Furthermore, the above gamma transformation can be divided into three steps, namely normalization, pre-compensation and denormalization.

[0055] Normalization is to transform the original grayscale value of the image into a value between 0 and 1. The formula is as follows:

[0056]

[0057] Among them, R i is the normalized gray value of the i-th pixel, r i is the original grayscale value of the i-th pixel, r min and r max They are the minimum and maximum grayscale values ​​of all pixels respectively.

[0058] The purpose of pre-compensation is to change the brightness of the image and enhance the image contrast. The formula is as follows:

[0059]

[0060] Among them, G i is the value corresponding to the i-th pixel after gamma transformation, R i is the normalized grayscale value of the ith pixel, γ is the gamma transform index, and γ can be specified manually according to actual needs.

[0061] Denormalization is to restore the grayscale value corresponding to the pre-compensated pixel to the true grayscale value. The formula is as follows:

[0062] R' i =G i ·(r max -r min )+r min (3)

[0063] Among them, R'i is the true gray value corresponding to the i-th pixel after denormalization, G i is the value corresponding to the i-th pixel after gamma transformation, r min and r max They are the minimum and maximum grayscale values ​​of all pixels respectively.

[0064] Furthermore, the above step S2 specifically includes the following steps:

[0065] Step S2.1, performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data;

[0066] Step S2.2, randomly dividing the segmented imaging logging image data to obtain a training set and a test set;

[0067] Step S2.3: Use the training set to train the constructed image segmentation and recognition module, and use the test set to test the effect of the image segmentation and recognition model to obtain a trained image segmentation and recognition model.

[0068] Furthermore, in the above step S2.1, when threshold segmentation is performed on the preprocessed imaging logging image data, the following steps are included: clustering segmentation of the image is performed using a genetic algorithm combined with a K-means clustering algorithm; and threshold segmentation is performed on the image data after cluster segmentation using a threshold segmentation method.

[0069] Furthermore, when the genetic algorithm is combined with the K-means clustering algorithm to perform clustering segmentation of the image, it includes: applying the genetic algorithm based on the preprocessed image to determine the initial cluster center, and performing K-means clustering image segmentation according to the determined initial cluster center.

[0070] This is because the selection of the initial cluster center has a great influence on the clustering results during K-means clustering, so the genetic algorithm is first applied to calculate the initial cluster center before K-means clustering. Among them, the image grayscale values ​​after image clustering segmentation of the preprocessed imaging logging image data using the K-means clustering algorithm are limited to a few categories. When solving the segmentation threshold in threshold segmentation, it is easier to find a suitable threshold, thereby achieving better segmentation results.

[0071] Furthermore, when the threshold segmentation method is used to perform threshold segmentation on the image data after K-means clustering segmentation, the following steps are included: applying an iterative algorithm to calculate the segmentation threshold T; classifying all pixels with grayscale values ​​greater than T into one category, and classifying all pixels with grayscale values ​​less than T into another category, to obtain the target area and the non-target area. The essence of threshold segmentation is to classify the pixels that make up the image into two categories according to the size of the grayscale value. The final image has two colors, black and white, with the target area being white and the non-target area being black.

[0072] Furthermore, after performing threshold segmentation on the image obtained by cluster segmentation using the threshold segmentation method, the following steps are also included: filtering the threshold segmentation result using the median filtering method to remove noise points.

[0073] This is because, in the image obtained by using the threshold segmentation method for image threshold segmentation, there are many pixels in non-target areas that will be mistakenly judged as pixels in the target area. In order to reduce such noise points, this embodiment uses median filtering to perform pixel filtering to reduce noise points in the segmented image.

[0074] Specifically, the median filtering method includes the following steps:

[0075] A filtering window consisting of an odd number of pixels is selected, and the grayscale value of the pixel to be filtered (i.e., the central pixel determined in the filtering window) is replaced by the median of the grayscale values ​​of the pixels in the filtering window area.

[0076] Example 2

[0077] Based on the machine learning imaging logging image fracture segmentation and identification method proposed in Example 1, this example studies the Mishrif reservoir of the H oil field in the Middle East. The H oil field has developed multiple sets of different strata, and dissolved pores and fractures are relatively developed in the Mishrif reservoir.

[0078] In this embodiment, 108 imaging logging images containing pores, induced fractures, high conductivity fractures and non-pore fracture structures are segmented, and the 108 segmented images are randomly divided into training sets and test sets, and finally 78 training set images and 30 test set images are obtained. Finally, the trained image segmentation recognition model is applied to recognize a 10-meter-long imaging logging image, and the areas where the pores, induced fractures and high conductivity fractures are located are marked.

[0079] like Figure 2 As shown, the specific process includes the following steps:

[0080] During image preprocessing, gamma transform is applied, which is divided into three steps. First, grayscale normalization is performed. According to the above formula (1), the original grayscale value of the image is transformed from 0 to 255 to a value between 0 and 1. Then, the normalized grayscale value is pre-compensated according to formula (2). The purpose of pre-compensation is to change the brightness of the image and enhance the image contrast effect. Finally, de-normalization is performed according to formula (3). De-normalization is to restore the value corresponding to the pixel point after pre-compensation to the true grayscale value.

[0081] like Figure 2 As shown, this is the preprocessed image obtained after preprocessing. The dissolution pores, induced fractures and high-conductivity fracture areas in the figure are more clearly visible than the original image.

[0082] After image preprocessing, threshold segmentation is used to divide all pixels in the image into two categories, where the target area is white and the non-target area is black. After threshold segmentation, the image obtained by threshold segmentation is subjected to median filtering to reduce the noise points in the image. The median filter first selects a filter window consisting of an odd number of pixels, and then replaces the grayscale value of the target pixel with the median of the pixel grayscale value in the window. The image after threshold segmentation and filtering is shown in the figure below. Figure 3 As shown, through Figure 3 Comparing before and after filtering, we can see that filtering can reduce the noise points in the image and make the target area clearer.

[0083] The grayscale value of the pixels in the original image is a series of continuously changing values. It is not easy to get a good critical threshold when determining the threshold for threshold segmentation. Therefore, the K-means clustering algorithm is applied to classify the originally continuously changing grayscale values ​​into a limited number of categories through clustering, and then determine the critical threshold for threshold segmentation. The selection of the initial cluster center in the K-means clustering algorithm has an important influence on the clustering results. In order to reasonably determine the initial cluster center, a genetic algorithm is introduced to obtain the initial cluster center, and then K-means clustering threshold segmentation is performed. The clustering results of K-means clustering after genetic algorithm optimization are shown in Figure 2. Figure 4 As shown in the figure, the optimized threshold segmentation result is as follows Figure 5 As shown in Figure 2, the dissolved pores, induced fractures, and high-conductivity fracture areas are more prominent in the optimized segmented image. Using the optimized model, 108 sample images were segmented one by one to obtain 108 high-quality sample images after segmentation.

[0084] The 108 sample images obtained by image segmentation are divided into 78 training images and 30 test images. The 78 training images are input into the deep neural network model for model training, and then the 30 test images are input into the trained deep neural network model to verify the results. Finally, 27 of the 30 test images are correctly identified and 3 are incorrectly identified. The accuracy of model recognition is 90.0%. The recognition results are shown in Table 1.

[0085] Table 1 Analysis of hole and crack recognition results of deep neural network

[0086] Imaging logging image categories Sample size The number of correct prediction samples Number of prediction errors Accuracy High seam guide 7 7 0 100.0％ Dissolution pores 11 9 2 81.8％ Induction seam 6 6 0 100.0％ Non-porous seam 6 5 1 83.3％

[0087] Finally, the trained model was applied to the actual case analysis, and a 10-meter-long imaging logging image of a well was identified, which contained multiple pore structures. After processing the image, the image was automatically divided into 15 sub-areas according to the gray value change trend. Finally, 9 sub-areas in the image were correctly identified, and the area with high conductivity fractures in the image was accurately identified. The recognition results are as follows: Figure 6 shown.

[0088] Example 3

[0089] The above-mentioned embodiment 1 provides a method for imaging logging image fracture segmentation and recognition based on machine learning. Correspondingly, this embodiment provides a system for imaging logging image fracture segmentation and recognition based on machine learning. The recognition system provided in this embodiment can implement a method for imaging logging image fracture segmentation and recognition based on machine learning in embodiment 1. The recognition system can be implemented by software, hardware, or a combination of software and hardware. For example, the system may include integrated or separate functional modules or functional units to execute the corresponding steps in each method of embodiment 1. Since the recognition system of this embodiment is basically similar to that of embodiment 1, the description process of this embodiment is relatively simple. For relevant parts, please refer to the partial description of embodiment 1. The embodiment of the system of this embodiment is only illustrative.

[0090] This embodiment provides a machine learning-based imaging logging image fracture segmentation and identification system, including:

[0091] A preprocessing module, used for preprocessing imaging logging image data;

[0092] A model training module is used to train the constructed image segmentation recognition model based on the preprocessed imaging logging image data to obtain a trained image segmentation recognition model;

[0093] The fracture recognition module is used to recognize the imaging logging image to be recognized by using the trained image segmentation recognition model to obtain the fracture recognition result of the imaging logging image to be recognized.

[0094] Example 4

[0095] This embodiment provides a processing device corresponding to the imaging logging image fracture segmentation and identification method based on machine learning provided in this embodiment 1. The processing device can be a processing device for a client, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the method of embodiment 1.

[0096] The processing device includes a processor, a memory, a communication interface and a bus, and the processor, the memory and the communication interface are connected through the bus to complete mutual communication. The memory stores a computer program that can be run on the processor, and the processor executes a method for segmenting and identifying fractures in imaging logging images based on machine learning provided in Example 1 when running the computer program.

[0097] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.

[0098] In some other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors of various types, which are not limited herein.

[0099] Example 5

[0100] An imaging logging fracture segmentation and identification method of this embodiment 1 can be specifically implemented as a computer program product. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the imaging logging fracture segmentation and identification method described in this embodiment 1 are loaded.

[0101] Computer readable storage media can be tangible devices that hold and store instructions used by instruction execution devices. Computer readable storage media can be, for example, but not limited to, electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any combination thereof.

[0102] It should be noted that the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present application. Each box in the flowchart or block diagram may represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

[0104] The above embodiments are only used to illustrate the present invention, wherein the structure, connection mode and manufacturing process of each component may be changed. Any equivalent transformations and improvements based on the technical solution of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A method for fracture segmentation and identification of imaging logging images based on machine learning, characterized in that The following steps are involved: Preprocessing of imaging logging image data; Based on the preprocessed imaging logging image data, the constructed image segmentation recognition model is trained to obtain a trained image segmentation recognition model, including: Performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data; The segmented imaging logging image data is randomly divided to obtain a training set and a test set; The training set is used to train the image segmentation and recognition module, and the test set is used to test the effect of the image segmentation and recognition model to obtain a trained image segmentation and recognition model; The method of performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data comprises: The K-means clustering algorithm is used to perform cluster segmentation on the preprocessed imaging logging image data; The threshold segmentation method is used to perform threshold segmentation on the image after cluster segmentation; Before clustering and segmenting the pre-processed imaging logging image data using the K-means clustering algorithm, the following steps are also included: Apply genetic algorithm to determine the initial cluster centers; The trained image segmentation recognition model is used to recognize the imaging logging image to be recognized, and the fracture recognition result of the imaging logging image to be recognized is obtained.

2. The imaging logging image fracture segmentation and identification method based on machine learning according to claim 1, characterized in that: When the imaging logging image is preprocessed, a gamma transform method is used.

3. The method for fracture segmentation and identification of imaging logging images based on machine learning according to claim 2, characterized in that: The gamma transform method includes normalization, pre-compensation and denormalization.

4. The method for fracture segmentation and identification of imaging logging images based on machine learning as claimed in claim 3, characterized in that: The normalized calculation formula is as follows: in, It is The normalized gray value of each pixel, It is The original gray value of the pixel, and They are the minimum and maximum grayscale values ​​of all pixels respectively.

5. The method for fracture segmentation and identification of imaging logging images based on machine learning according to claim 3, characterized in that: The calculation formula of the pre-compensation is: in, It is The corresponding value of the pixel after gamma transformation, It is The normalized gray value of each pixel, is the gamma transform exponent.

6. The imaging logging image fracture segmentation and identification method based on machine learning according to claim 1, characterized in that: After the threshold segmentation method is used to perform threshold segmentation on the cluster segmented image, the method further includes the following steps: using a median filtering method to filter the threshold segmentation result.

7. A machine learning-based imaging logging image fracture segmentation and identification system, characterized in that: include: A preprocessing module, used for preprocessing imaging logging image data; The model training module is used to train the constructed image segmentation and recognition model based on the preprocessed imaging logging image data to obtain a trained image segmentation and recognition model, including: Performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data; The segmented imaging logging image data is randomly divided to obtain a training set and a test set; The training set is used to train the image segmentation and recognition module, and the test set is used to test the effect of the image segmentation and recognition model to obtain a trained image segmentation and recognition model; The step of performing threshold segmentation on the preprocessed imaging logging image data to obtain segmented imaging logging image data includes: The K-means clustering algorithm is used to perform cluster segmentation on the preprocessed imaging logging image data; The threshold segmentation method is used to perform threshold segmentation on the image after cluster segmentation; Before clustering and segmenting the pre-processed imaging logging image data using the K-means clustering algorithm, the method further includes: Apply genetic algorithm to determine the initial cluster centers; The fracture recognition module is used to recognize the imaging logging image to be recognized by using the trained image segmentation recognition model to obtain the pore and fracture structure of the imaging logging image to be recognized.

Images