Machine learning-based method, system and storage medium for inverse analysis of in-situ stress field of rock mass

Abstract

The present invention relates to the technical field of machine learning, and more particularly to a method, system and storage medium for inverting the rock mass in-situ stress field based on machine learning. Step S1: Construct a three-dimensional virtual model of the target rock mass and obtain the geological data at each position in the inversion target area; Step S2: Obtain the geological types in the inversion area and obtain the rock mass parameters of the rock mass corresponding to the sampling points; Step S3: Obtain a number of groups of test data through the hydraulic fracturing method and the borehole method respectively, perform preprocessing, and obtain learning data and detection data; Step S4: Create a rock mass stress prediction model, the rock mass stress prediction model includes m sub-models, train the sub-models using the learning data, and verify the sub-models based on the detection data. The present invention solves the problem of inaccurate inversion of the rock mass stress field and improves the inversion accuracy of the rock mass in-situ stress field.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and more particularly to a method, system and storage medium for inverting the rock mass in-situ stress field based on machine learning. Background Art

[0002] Rock mass stress inversion is an important research content in the field of geotechnical engineering. It inversely calculates the stress state or mechanical parameters of the rock mass by analyzing the stress response of the rock mass under specific conditions. This technology plays a key role in the design, construction and stability evaluation of underground engineering. With the increasing demand for underground space development, the development of rock mass stress inversion technology becomes particularly important. Similar prior arts include a Chinese patent with the publication number CN116861704B, which proposes a method for dynamically and rapidly predicting the large deformation grade of a high in-situ stress soft rock tunnel, relating to the field of tunnel large deformation prediction, including: first, constructing a three-dimensional geological model based on the obtained geological data; then, performing in-situ stress inversion analysis based on the three-dimensional geological model and the measured results of in-situ stress to determine the magnitude and direction of the in-situ stress within the tunnel axis range; then, carrying out point load strength tests on test rock samples to calculate the uniaxial compressive strength of the rock for the section to be predicted; then, estimating the rock mass compressive strength according to the uniaxial compressive strength of the rock for the section to be predicted; finally, determining the large deformation grade of the tunnel based on the maximum in-situ stress in the in-situ stress inversion analysis results and the estimated rock mass compressive strength, and formulating corresponding construction methods and support parameters based on the determination results; it can consider the in-situ stress characteristics and surrounding rock properties of different prediction sections and can realize the dynamic and rapid prediction of large deformations of high in-situ stress soft rock tunnels. In addition, a similar prior art is a US patent with the publication number US11536864B1, which discloses a method for establishing the in-situ stress field distribution of a slope in a canyon area, including: obtaining the persistence rate of the fracture surface according to the structural plane trace length and rock bridge length of the fracture surface, and then obtaining the fracture stage of the crack according to the fracture stage of the fracture surface. According to the progressive failure characteristics of the rock mass and combining the characteristics of the fracture surface, the magnitude and direction of the maximum principal stress are obtained, and the in-situ stress field distribution is established. This method is simple to operate, does not require in-situ stress testing, does not require a large amount of manpower and material resources, does not require redundant capital investment, and can simply and effectively obtain in-situ stress field data. The above two documents have solved the problem of in-situ stress field inversion, but the accuracy is low and the rock mass stress distribution problem cannot be accurately inverted. Summary of the Invention

[0003] In order to better solve the above problems, the present invention provides a method for inverting the rock mass in-situ stress field based on machine learning, including the following steps:

[0004] Step S1: By scanning and imaging the target rock mass to obtain a first geological image, filtering and interpolating the first geological image to obtain a second geological image, and constructing a three-dimensional virtual model of the target rock mass based on the second geological image; obtaining geological data at each position in the inversion area based on the three-dimensional virtual model, where the geological data includes measurement data of faults, fractures, and topographic trends;

[0005] Step S2: Based on the three-dimensional virtual model, obtain the geological types in the inversion area, select several sampling points within the range corresponding to each geological type, and obtain the rock mass parameters corresponding to the sampling points by sampling and in-situ compression of the rock mass at each sampling point, and mark the rock mass parameters in the three-dimensional virtual model;

[0006] Step S3: Obtain several groups of test data by the hydraulic fracturing method and the borehole method respectively according to the sub-region where each sampling point is located, preprocess the test data, and generate learning data and detection data from the preprocessed test data, the rock mass parameters of the sampling points corresponding to the test data, and the geological data;

[0007] Step S4: Create a rock mass stress prediction model, where the rock mass stress prediction model includes m sub-models, the m sub-models are created based on the geological types, use the learning data to train the sub-models, and verify the sub-models based on the detection data, where each sub-model corresponds to a geological type.

[0008] As a preferred technical solution of the present invention, the filtering and interpolation of the first geological image to obtain the second geological image in step S1 includes:

[0009] Step S11: Detect abnormal pixels in the pixels of the first geological image, obtain the difference between the brightness value of each pixel and a first value, when the difference is greater than a set difference, the pixel is an abnormal pixel, and when the difference is less than or equal to the set difference, the pixel is a normal pixel, where the first value is the average of the maximum and minimum brightness values of the pixels in the first geological image;

[0010] Step S12: Taking the abnormal pixel as the center, within a set range, obtain the number of other abnormal pixels, and based on the number, obtain the normal pixels on the circle with the abnormal pixel as the center and a set radius around the abnormal pixel, and average and weight the brightness values of the normal pixels on the circle to obtain an interpolation value, where the more the number, the smaller the set radius, and vice versa, the larger, and the number of abnormal pixels within the set range is less than the total number of pixels within the set range;

[0011] Step S13: Replace the brightness value of each abnormal pixel with the interpolation value corresponding to the abnormal pixel.

[0012] As a preferred technical solution of the present invention, step S2 includes: obtaining the geological type of the rock mass in the three-dimensional virtual model, selecting a landmark sampling area within the range of each corresponding geological type area, selecting multiple sampling points at different positions in the sampling area, sampling the rock mass at each sampling point and performing in-situ compression to obtain the rock mass parameters corresponding to the sampling point, where the rock mass parameters include the Young's modulus E, Poisson's ratio v, strain hardening modulus H, peak strain μ, and post-peak drop modulus M of the rock mass, and the geological type at least includes folds, fractures, and faults.

[0013] As a preferred technical solution of the present invention, step S3 includes the following steps:

[0014] Step S31: Based on the three-dimensional virtual model, obtain the sub-region where the sampling point in step S2 is located, and uniformly select N positions in the sub-region as data collection positions, where N is a positive integer greater than or equal to 10;

[0015] Step S32: Obtain several groups of test data through the hydraulic fracturing method and the drilling method respectively, and mark the several groups of test data at the corresponding data collection positions in the three-dimensional virtual model;

[0016] Step S33: According to the geological data and rock mass parameters at the data collection position of the test data and the geological data and rock mass parameters at the data collection positions of other surrounding test data, obtain the magnitude relationship between the test data and other surrounding test data, and judge whether the test data needs to be corrected based on the magnitude relationship. When the test data satisfies the magnitude relationship, the test data does not need to be corrected. When the test data does not satisfy the magnitude relationship, the test data needs to be corrected, and weight the four test data closest to the data collection position of the test data in the front, back, left, and right directions, and use the weighted value as the correction data of the test data. Among them, the weights of the four test data are set according to the distance from the test data. The closer the distance, the higher the weight, and vice versa.

[0017] As a preferred technical solution of the present invention, after step S33, there is also step S34:

[0018] Based on the three-dimensional virtual model, combine the test data, the geological data at the data acquisition location of the test data, and the rock mass parameters of the sub-region where the test data is located to form training data, and select any one of any three consecutive pieces of the training data within the range corresponding to any geological type as the detection data, and the remaining two pieces of the training data as the learning data.

[0019] As a preferred technical solution of the present invention, the step S4 includes:

[0020] Create the rock mass stress prediction model, obtain m geological types of the inversion area through the geological data, and create m sub-models based on the geological types. Use the learning data within the sub-region corresponding to each geological type in the three-dimensional virtual model as the learning data of the corresponding sub-model, and train the sub-model through the learning data, where m is a positive integer greater than or equal to 3.

[0021] As a preferred technical solution of the present invention, the step S4 further includes: input the geological data and rock mass parameters in the detection data into the rock mass stress prediction model, obtain the geological type to which the detection data belongs according to the position of the detection data in the three-dimensional virtual model, input the corresponding sub-model according to the geological type, obtain the prediction result, compare the prediction result with the stress value in the detection data, and obtain the comparison difference. When the comparison difference is less than or equal to the preset difference, the prediction result is accurate. When the comparison difference is greater than the preset difference and the prediction result satisfies the magnitude relationship between the stress value in the detection data and the stress value in the nearest learning data in the sub-region where it is located, re-train the sub-model by increasing the number of learning data of the geological type. When the comparison difference is greater than the preset difference and the prediction result does not satisfy the magnitude relationship between the detection data and the nearest learning data in the sub-region where it is located, repeat the step S2 to re-obtain the rock mass parameters corresponding to the geological type and re-train the sub-model.

[0022] As a preferred technical solution of the present invention, the test data is the numerical value of the in-situ stress field at the data acquisition location.

[0023] The present invention also provides a machine learning-based rock mass in-situ stress field inversion system, which is used to implement the above method. The system includes:

[0024] A virtual model generation unit, configured to scan and image a target rock mass to obtain a first geological image, filter and interpolate the first geological image to obtain a second geological image, and construct a three-dimensional virtual model of the target rock mass based on the second geological image; obtain geological data at each position in the inversion area based on the three-dimensional virtual model, where the geological data includes measurement data of faults, fractures, and topographic trends;

[0025] A marking unit, configured to obtain geological types in the inversion area based on the three-dimensional virtual model, select a number of sampling points within the range of the area corresponding to each geological type, and obtain the rock mass parameters corresponding to the sampling points by sampling the rock mass at each sampling point and performing in-situ compression, and mark the rock mass parameters in the three-dimensional virtual model;

[0026] A data generation unit, configured to obtain several groups of test data respectively by the hydraulic fracturing method and the borehole method according to the sub-area where each sampling point is located, preprocess the test data, and generate learning data and detection data from the preprocessed test data, the rock mass parameters corresponding to the test data at the sampling points, and the geological data;

[0027] A model training unit, configured to create a rock mass stress prediction model, where the rock mass stress prediction model includes m sub-models, the m sub-models are created based on the geological types, use the learning data to train the sub-models, and verify the sub-models based on the detection data, where each sub-model corresponds to a geological type.

[0028] The present invention also provides a computer storage medium, where the storage medium stores program instructions, and when the program instructions run, the device where the storage medium is located is controlled to execute the above method.

[0029] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0030] The present invention obtains a first geological image of a target rock mass through rock mass scanning technology, filters and denoises the above first geological image, obtains abnormal pixels in the above first geological image, obtains the number of other above abnormal pixels within a set range, and filters and interpolates the above abnormal pixels according to the number of abnormal pixels within the above set range to obtain a second geological image, and obtains accurate geological data based on the second geological image. According to the geological type of the rock mass in the above three-dimensional virtual model, landmark sampling areas are selected within the corresponding area range of each geological type, and multiple sampling points are selected at different positions in the above sampling areas, and the rock mass parameters of each above sampling point are obtained, so as to obtain relatively rich and comprehensive sampling rock mass parameters for each geological type. Based on the sub-areas where each sampling point is located in step S2 in the above three-dimensional virtual model, N data acquisition positions are evenly selected in the above sub-areas, and several groups of test data are obtained through the hydraulic fracturing method and drilling, and the above test data are marked in the above three-dimensional virtual model. Since the sampling points and the above data acquisition positions are in the same sub-area, the accuracy of the rock mass stress prediction model is improved by combining the rock mass parameters of the above sampling points and the above test data into learning data and detection data. The stress magnitude relationship between two relatively positions in the above three-dimensional virtual model can be judged through the stress distribution law, and whether the above test data needs to be corrected is judged through the above magnitude relationship. When correction is needed, the four closest test data before, after, left, and right of the above test data are weighted to obtain relatively accurate stress field data, laying a foundation for further improving the accuracy of the rock mass stress prediction model. Create multiple above sub-models, and fully train the learning data corresponding to each above geological type through the above sub-models, and the stress prediction values corresponding to each geological type can be obtained. Through the mutual cooperation of the above technical solutions, the accuracy of the in-situ stress field inversion can be improved. Description of the Drawings

[0031] Figure 1 is a flowchart of the method for in-situ stress field inversion of rock mass based on machine learning according to the present invention;

[0032] Figure 2 is a structural diagram of the system for in-situ stress field inversion of rock mass based on machine learning according to the present invention. Detailed Embodiments

[0033] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0034] The present invention provides a method for in-situ stress field inversion of rock mass based on machine learning, as Figure 1 shown, including the following steps:

[0035] Step S1: By scanning and imaging the target rock mass to obtain a first geological image, filtering and interpolating the first geological image to obtain a second geological image, and constructing a three-dimensional virtual model of the target rock mass based on the second geological image; obtaining geological data at each position in the inversion target area based on the three-dimensional virtual model, where the geological data includes measurement data of faults, fractures, and terrain trends.

[0036] Specifically, a first geological image of the target rock mass is obtained through rock mass scanning technology. Since noise interference is inevitably introduced during the scanning process, there will be noise points, i.e., abnormal pixels. In order to obtain more accurate geological data, it is necessary to filter and denoise the above-mentioned first geological image. Obtain the difference between the brightness value, i.e., the gray value, of each pixel in the above-mentioned first geological image and a first value. The above-mentioned first value is the average of the maximum and minimum pixel brightness values. Based on the above difference, accurately identify the abnormal pixels and normal pixels in the above-mentioned first geological image. Also, since the structure and geological features of the rock formation are expressed by the brightness values of the pixels in the above-mentioned first geological image, and due to the continuity between pixels, therefore, within a set range, obtain the number of other above-mentioned abnormal pixels, and filter and interpolate the above-mentioned abnormal pixels according to the number of abnormal pixels within the above-mentioned set range. Taking the above-mentioned abnormal pixel as the center, the average value of the normal pixels on the circle with a set radius is used as the brightness value of the above-mentioned abnormal pixel. The more the number of the above-mentioned abnormal pixels, the more accurate the interpolation of the brightness value of the above-mentioned abnormal pixel needs to be obtained. On the contrary, when the number of the above-mentioned abnormal pixels is small, it needs to be consistent with the brightness value in the current area. Therefore, the more the above-mentioned abnormal quantity, the larger the corresponding above-mentioned set radius. Through the above technical solution, more accurate interpolation of the above-mentioned abnormal pixel points can be obtained, thereby obtaining a more accurate three-dimensional virtual model, laying a foundation for further obtaining accurate geological data.

[0037] Step S2: Based on the three-dimensional virtual model, obtain the geological types in the inversion area, select several sampling points within the range corresponding to each geological type, and obtain the rock mass parameters corresponding to the sampling points by sampling and in-situ compression of the rock mass at each sampling point, and mark the rock mass parameters in the three-dimensional virtual model.

[0038] Specifically, according to the geological types of the rock masses in the above three-dimensional virtual model, and within the corresponding regions of each geological type, iconic sampling regions are selected, that is, different regions with iconic characteristics are selected as sampling regions. For example, at different equidistant positions and different distance positions on both sides of the fold region are used as sampling regions, and multiple sampling points are selected at different positions in the above sampling regions, and the rock mass parameters of each sampling point are obtained. Through the above technical solution, relatively rich and comprehensive sampled rock mass parameters of each geological type can be obtained, laying a foundation for further obtaining accurate inversion results of the in-situ stress field of the target rock mass.

[0039] Step S3: According to the sub-region where each sampling point is located, several groups of test data are obtained respectively by the hydraulic fracturing method and the borehole method, the test data is preprocessed, and the preprocessed test data, the rock mass parameters of the sampling points corresponding to the test data, and the geological data are used to generate learning data and detection data;

[0040] Specifically, based on the above three-dimensional virtual model, in the sub-region where each sampling point is located in step S2, N data acquisition positions are evenly selected in the above sub-region, and several groups of test data are obtained by the hydraulic fracturing method and the borehole method. The above test data is the stress at the above data acquisition positions, and the above test data is marked on the above three-dimensional virtual model. Since the sampling points and the above data acquisition positions are in the same sub-region, the accuracy of the rock mass stress prediction model is improved by combining the rock mass parameters of the above sampling points and the above test data into learning data and detection data, and then a relatively accurate inversion of the in-situ stress field is obtained. The magnitude relationship of the stresses at two relatively positions in the above three-dimensional virtual model can be judged through the stress distribution law, and whether the above test data needs to be corrected is judged through the above magnitude relationship. When the above test data satisfies the above magnitude relationship, the above test data does not need to be corrected. When the above test data does not satisfy the above magnitude relationship, the test data needs to be corrected, and the four nearest test data before, after, left, and right of the above test data are weighted, where the sum of the weights of the left and right test data is 0.5, and the sum of the weights of the front and rear test data is also 0.5, and the closer the distance to the above test data, the higher the weight, and vice versa. Through the above technical solution, relatively accurate stress field data can be obtained, laying a foundation for further improving the accuracy of the rock mass stress prediction model.

[0041] Step S4: Create a rock mass stress prediction model. The rock mass stress prediction model includes m sub-models. The m sub-models are created based on the geological types, and the sub-models are trained using the learning data and the sub-models are verified based on the detection data, where each sub-model corresponds to a geological type.

[0042] Specifically, by creating multiple above-mentioned sub-models and fully training the learning data corresponding to each above-mentioned address type through the above-mentioned sub-models, the stress prediction values corresponding to each geological type can be obtained, thereby improving the accuracy of stress inversion. By inputting the input data in the detection data of the corresponding sub-model into the above-mentioned rock mass stress prediction model, the above-mentioned input data being the rock mass parameters and geological data in the detection data, the prediction result, i.e., the predicted stress, is obtained, and the above-mentioned predicted stress is compared with the corresponding stress value in the detection data, and the comparison difference is obtained. Then, the above-mentioned rock mass stress prediction model is corrected according to the above-mentioned comparison difference. When the comparison difference is less than or equal to the preset difference, it is considered that the prediction result is relatively accurate and does not need to be corrected. When the comparison difference is greater than the preset difference and the prediction result satisfies the size relationship between the detection data and the nearest learning data in the sub-region where it is located, it is considered that although the prediction result is not accurate enough, the error is relatively small, which may be caused by the insufficient number of learning data. The accuracy of the corresponding sub-model is improved by increasing the number of learning data. When the comparison difference is greater than the preset difference and the prediction result does not satisfy the size relationship between the detection data and the nearest learning data in the sub-region where it is located, it indicates that the error is relatively large and the accuracy of the sub-model is poor. It may be that the rock mass parameters in the learning data are not accurate enough. By repeating the above step S2 to re-obtain the above-mentioned rock mass parameters, and re-obtaining the above-mentioned learning data and detection data, and re-training the above-mentioned sub-model, the accuracy of the above-mentioned sub-model is improved. Through the above technical solution, a rock mass stress model with higher accuracy can be obtained, thereby obtaining an accurate inverted rock mass geostress field.

[0043] Further, the step of obtaining the second geological image by filtering and interpolating the first geological image in step S1 includes:

[0044] Step S11: Detect abnormal pixels for the pixels in the set area in the middle of the first geological image, obtain the difference between the brightness value of each pixel and the first value. When the difference is greater than the set difference, the pixel is an abnormal pixel; when the difference is less than or equal to the set difference, the pixel is a normal pixel. Among them, the first value is the average of the maximum and minimum values of the pixel brightness in the first geological image;

[0045] Specifically, when scanning and imaging the above-mentioned target rock mass, noise will inevitably be introduced, so there will be noise points, that is, abnormal pixels. In order to obtain more accurate geological data, it is necessary to filter and denoise the above-mentioned first geological image. First, abnormal pixel detection is carried out. Since when the brightness of the above-mentioned pixel is the largest, it is all white, that is, the brightest, and when the brightness of the above-mentioned pixel is the smallest, that is, all black, neither can represent the specific image content. The pixel brightness that can represent the image content should be in the middle of the maximum and minimum brightness values. Therefore, the difference between the brightness value, that is, the gray value, of each pixel in the above-mentioned first geological image and the above-mentioned first value is obtained. The above-mentioned first value is the average value of the maximum and minimum pixel brightness values. When the above-mentioned difference is less than or equal to the above-mentioned set difference, that is, when it is relatively close to the above-mentioned first value, the above-mentioned pixel is a normal pixel. When the above-mentioned difference is greater than the above-mentioned set difference, the above-mentioned pixel is an abnormal pixel. Through the above technical solution, the abnormal pixels and normal pixels in the above-mentioned first geological image can be accurately identified.

[0046] Step S12: Taking the abnormal pixel as the center, within a set range, obtain the number of other abnormal pixels, and based on the number, obtain the normal pixels on the circle with the abnormal pixel as the center and a set radius around the abnormal pixel. Average and weight the gray values of the normal pixels on the circle to obtain an interpolation value. Among them, the more the number, the smaller the set radius, and vice versa, the larger. The number of abnormal pixels within the set range is less than the total number of pixels within the set range;

[0047] Step S13: Replace the brightness value of each abnormal pixel with the corresponding interpolation value.

[0048] Specifically, since the structure and geological features of the rock formation are expressed by the brightness values of the pixels in the above-mentioned first geological image, and due to the continuity between the pixels, therefore, within a set range, obtain the number of other above-mentioned abnormal pixels. Among them, the more the number, the greater the density of abnormal pixels within the set range, and vice versa, the smaller the density. And filter and interpolate the above-mentioned abnormal pixels according to the number of abnormal pixels within the set range. Taking the above-mentioned abnormal pixel as the center of the circle, use the average value of the normal pixels on the circle with a set radius as the brightness value of the above-mentioned abnormal pixel. The more the number of the above-mentioned abnormal pixels, the more accurate the interpolation of the brightness value of the above-mentioned abnormal pixel needs to be obtained. On the contrary, when the number of the above-mentioned abnormal pixels is small, it needs to be consistent with the brightness value in the current area. Therefore, the more the above-mentioned abnormal number, the larger the corresponding set radius. Through the above technical solution, a more accurate interpolation of the above-mentioned abnormal pixel points can be obtained, so as to obtain a more accurate three-dimensional virtual model, laying a foundation for further obtaining accurate geological data.

[0049] Further, the step S2 includes: obtaining the geological type of the rock mass in the three-dimensional virtual model, selecting a landmark sampling area within the corresponding area of each geological type, selecting a plurality of sampling points at different positions in the sampling area, sampling the rock mass at each sampling point and performing in-situ compression to obtain the rock mass parameters corresponding to the sampling point, where the rock mass parameters include the Young's modulus E, Poisson's ratio v, strain hardening modulus H, peak strain μ, and post-peak drop modulus M of the rock mass, and the geological types at least include folds, fractures, and faults.

[0050] Specifically, according to the geological type of the rock mass in the above three-dimensional virtual model, and selecting a landmark sampling area within the corresponding area of each geological type, that is, selecting different areas with landmark characteristics as the sampling area. For example, at different equidistant positions and different distance positions on both sides of the fold area as the sampling area, and selecting a plurality of sampling points at different positions in the above sampling area, and obtaining the rock mass parameters of each above sampling point. Through the above technical solution, it is possible to obtain relatively rich and comprehensive sampling rock mass parameters for each geological type, laying a foundation for further obtaining an accurate inversion result of the in-situ stress field of the above target rock mass.

[0051] Further, the step S3 includes the following steps:

[0052] Step S31: Based on the three-dimensional virtual model, obtain the sub-region where the sampling points in step S2 are located, and uniformly select N positions in the sub-region as data collection positions, where N is a positive integer greater than or equal to 10;

[0053] Step S32: Obtain several groups of test data respectively by the hydraulic fracturing method and the borehole method, and mark the several groups of test data at the corresponding data collection positions in the three-dimensional virtual model;

[0054] Specifically, based on the above three-dimensional virtual model, for each sub-region where the sampling points in step S2 are located, and uniformly select N data collection positions in the above sub-region, and obtain several groups of test data by the hydraulic fracturing method and the borehole method. The above test data is the stress at the above data collection positions, and mark the above test data in the above three-dimensional virtual model. Since the sampling points and the above data collection positions are in the same sub-region, by combining the rock mass parameters of the above sampling points and the above test data into learning data and detection data, the accuracy of the rock mass stress prediction model can be improved, and thus a more accurate inversion of the in-situ stress field can be obtained.

[0055] Step S33: Based on the geological data and rock mass parameters at the data acquisition location of the test data and the geological data and rock mass parameters at the data acquisition locations of other surrounding test data, obtain the magnitude relationship between the test data and other surrounding test data, and determine whether the test data needs to be corrected based on the magnitude relationship. When the test data satisfies the magnitude relationship, the test data does not need to be corrected. When the test data does not satisfy the magnitude relationship, the test data needs to be corrected, and weight the four test data closest to the data acquisition location of the test data in the front, back, left, and right directions, and use the weighted value as the corrected data of the test data. Among them, the weights of the four test data are set according to the distance from the test data. The closer the distance, the higher the weight, and vice versa.

[0056] Specifically, through the above three-dimensional virtual model, the above test data, geological data, and rock mass parameters at each position can be intuitively seen. Although it is impossible to individually determine whether the test data is accurate, the magnitude relationship of the stresses at two relative positions can be judged through the stress distribution law, and whether the above test data needs to be corrected can be judged through the above magnitude relationship. When the above test data satisfies the above magnitude relationship, the above test data does not need to be corrected. When the above test data does not satisfy the above magnitude relationship, the test data needs to be corrected, and weight the four test data closest to the above test data in the front, back, left, and right directions. Among them, the sum of the weights of the two test data on the left and right and the sum of the weights of the two test data on the front and back are both 0.5, and the closer the distance to the above test data, the higher the weight, and vice versa. Through the above technical solution, more accurate stress field data can be obtained, laying a foundation for further improving the accuracy of the rock mass stress prediction model.

[0057] Further, after step S33, there is also step S34:

[0058] Based on the three-dimensional virtual model, combine the test data, the geological data at the data acquisition location of the test data, and the rock mass parameters of the sub-region where the test data is located into training data, and select any one of any three consecutive training data within the range corresponding to any geological type as the test data, and the remaining two training data as the learning data.

[0059] Specifically, through the distribution position of the test data in the above three-dimensional virtual model, combine the geological data, rock mass parameters, and the above test data at the position of the above test data into training data, and select any one of any three consecutive training data within the range corresponding to any geological type as the test data, and the remaining other two training data as the learning data. Through the above technical solution, obtaining more comprehensive learning data and test data for each geological type improves the accuracy of the rock mass stress prediction model.

[0060] Further, step S4 includes:

[0061] Create the rock mass stress prediction model, obtain m geological types of the inversion area through the geological data, create m sub-models based on the geological types, use the learning data in each sub-region corresponding to each geological type in the three-dimensional virtual model as the learning data of the corresponding sub-model, and train the sub-model through the learning data, where m is a positive integer greater than or equal to 3.

[0062] Specifically, create multiple above-mentioned sub-models, and fully train the learning data corresponding to each above-mentioned geological type through the above-mentioned sub-models, so as to obtain the stress prediction value corresponding to each geological type, thereby improving the accuracy of stress inversion.

[0063] Further, step S4 also includes: input the geological data and rock mass parameters in the detection data into the rock mass stress prediction model, obtain the geological type to which the detection data belongs according to the position of the detection data in the three-dimensional virtual model, input the corresponding sub-model according to the geological type, obtain the prediction result, compare the prediction result with the stress value in the detection data, and obtain the comparison difference. When the comparison difference is less than or equal to the preset difference, the prediction result is accurate. When the comparison difference is greater than the preset difference and the prediction result satisfies the magnitude relationship between the stress value in the detection data and the stress value in the nearest learning data in the sub-region where it is located, re-train the sub-model by increasing the number of learning data of the geological type. When the comparison difference is greater than the preset difference and the prediction result does not satisfy the magnitude relationship between the detection data and the nearest learning data in the sub-region where it is located, repeat step S2 to re-obtain the geological type corresponding to the rock mass parameters and re-train the sub-model.

[0064] Specifically, by inputting the input data in the detection data of the corresponding sub-model into the above-mentioned rock mass stress prediction model, where the input data is the rock mass parameters and geological data in the detection data, the prediction result, i.e., the predicted stress, is obtained. Then, the predicted stress is compared with the corresponding stress value in the detection data, and the comparison difference is obtained. According to the comparison difference, the above-mentioned rock mass stress prediction model is corrected. When the comparison difference is less than or equal to the preset difference, it is considered that the prediction result is relatively accurate and does not need to be corrected. When the comparison difference is greater than the preset difference and the prediction result satisfies the size relationship between the detection data and the nearest learning data in the sub-region where it is located, it is considered that although the prediction result is not accurate enough, the error is relatively small, which may be caused by the insufficient number of learning data. The accuracy of the corresponding sub-model is improved by increasing the number of learning data. When the comparison difference is greater than the preset difference and the prediction result does not satisfy the size relationship between the detection data and the nearest learning data in the sub-region where it is located, it indicates that the error is relatively large and the accuracy of the sub-model is poor. It is possible that the rock mass parameters in the learning data are not accurate enough. By repeating the above step S2 to re-obtain the above-mentioned rock mass parameters, re-obtain the above-mentioned learning data and detection data, and re-train the above-mentioned sub-model, the accuracy of the above-mentioned sub-model is improved. Through the above technical solution, a rock mass stress model with higher accuracy can be obtained, so as to obtain an accurate inverse rock mass geostress field.

[0065] Furthermore, the test data is the numerical value of the geostress field at the data acquisition location.

[0066] The present invention also provides a rock mass geostress field inversion system based on machine learning. The system is used to implement the above method, as Figure 2 shown. The system includes:

[0067] A virtual model generation unit, which is used to obtain a first geological image by scanning and imaging the target rock mass, filter and interpolate the first geological image to obtain a second geological image, and construct a three-dimensional virtual model of the target rock mass based on the second geological image; based on the three-dimensional virtual model, obtain the geological data at each position in the inversion area, and the geological data includes measurement data of faults, fractures, and terrain trends;

[0068] A marking unit, which is used to obtain the geological types in the inversion area based on the three-dimensional virtual model, select several sampling points within the range corresponding to each geological type area, and obtain the rock mass parameters corresponding to the sampling points by sampling and in-situ compression of the rock mass at each sampling point, and mark the rock mass parameters on the three-dimensional virtual model;

[0069] A data generation unit, configured to respectively obtain several groups of test data through the hydraulic fracturing method and the drilling method according to the sub-region where each sampling point is located, preprocess the test data, and generate learning data and detection data from the preprocessed test data, the rock mass parameters of the sampling points corresponding to the test data, and geological data;

[0070] A model training unit, configured to create a rock mass stress prediction model, where the rock mass stress prediction model includes m sub-models, the m sub-models are created based on the geological types, and the learning data is used to train the sub-models, and the detection data is used to verify the sub-models, where each sub-model corresponds to a geological type.

[0071] The present invention also provides a computer storage medium, where the storage medium stores program instructions, and when the program instructions run, the device where the storage medium is located is controlled to execute the above method.

[0072] In summary, the present invention obtains a first geological image of a target rock mass through a rock mass scanning technology, filters and denoises the above first geological image, obtains abnormal pixels in the above first geological image, within a set range, obtains the number of other above abnormal pixels, and filters and interpolates the above abnormal pixels according to the number of abnormal pixels within the above set range to obtain a second geological image, and obtains accurate geological data based on the second geological image, according to the geological types of the rock masses in the above three-dimensional virtual model, and selects landmark sampling areas within the corresponding regions of each above geological type, and selects multiple sampling points at different positions in the above sampling areas, and obtains the rock mass parameters of each above sampling point, so as to obtain relatively rich and comprehensive sampling rock mass parameters for each geological type. Based on the above three-dimensional virtual model, the sub-regions where each sampling point is located in step S2, and N data acquisition positions are evenly selected in the above sub-regions, and several groups of test data are obtained through the hydraulic fracturing method and the drilling method, and the above test data is marked in the above three-dimensional virtual model. Since the sampling points and the above data acquisition positions are in the same sub-region, the rock mass parameters of the above sampling points and the above test data are combined into learning data and detection data, thereby improving the accuracy of the rock mass stress prediction model. The stress magnitude relationship between two relatively positioned points in the above three-dimensional virtual model can be judged through the stress distribution law, and whether the above test data needs to be corrected is judged through the above magnitude relationship. When correction is required, the four nearest test data before, after, left, and right of the above test data are weighted to obtain relatively accurate stress field data, laying a foundation for further improving the accuracy of the rock mass stress prediction model. Create multiple above sub-models, and fully train the learning data corresponding to each above address type through the above sub-models, and the stress prediction values corresponding to each geological type can be obtained. Through the mutual cooperation of the above technical solutions, the accuracy of the in-situ stress field inversion can be improved.

[0073] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as falling within the scope described in this specification.

[0074] The above-described embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent for the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent for the present invention should be subject to the appended claims.

[0075] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A rock mass in-situ stress field inversion method based on machine learning, characterized in that: The steps include: Step S1: scanning and imaging the target rock mass and acquiring a first geological image, filtering and interpolating the first geological image to acquire a second geological image, and constructing a three-dimensional virtual model of the target rock mass based on the second geological image; acquiring geological data at each position in the inversion area based on the three-dimensional virtual model, wherein the geological data includes measurement data of faults, fractures and terrain trends; Step S2: Based on the three-dimensional virtual model, the geological type in the inversion area is obtained, and a number of sampling points are selected within the area corresponding to each geological type, and the rock mass parameters corresponding to the sampling point are obtained by sampling the rock mass at each sampling point and performing in-situ compression, and the rock mass parameters are marked in the three-dimensional virtual model; Step S3: obtaining a plurality of test data sets respectively by hydraulic fracturing and drilling according to the sub-region where each sampling point is located, preprocessing the test data, and generating learning data and detection data from the preprocessed test data, the rock mass parameters of the sampling points corresponding to the test data, and the geological data; Step S4: creating a rock mass stress prediction model, wherein the rock mass stress prediction model includes m sub-models, wherein the m sub-models are created based on the geological type, and the sub-models are trained using the learning data, and the sub-models are verified based on the detection data, wherein each of the sub-models corresponds to a geological type; Wherein, step S3 comprises: Step S31: based on the three-dimensional virtual model, the sub-region where the sampling point in step S2 is located is obtained, and N positions are uniformly selected in the sub-region as data collection positions, where the value of N is a positive integer greater than or equal to 10; Step S32: obtaining a plurality of groups of test data by the hydraulic fracturing method and the drilling method respectively, and marking the plurality of groups of test data at positions corresponding to the data acquisition positions in the three-dimensional virtual model; Step S33: according to the geological data and rock mass parameters at the data acquisition position of the test data and the geological data and rock mass parameters at the data acquisition positions of other surrounding test data, obtain the size relationship between the test data and other surrounding test data, and judge whether the test data needs to be corrected based on the size relationship. When the test data meets the size relationship, the test data does not need to be corrected. When the test data does not meet the size relationship, the test data needs to be corrected, and based on the four test data closest to the data acquisition position of the test data, the weighted values ​​are used as the correction data of the test data, wherein the weights of the four test data are set by the distance from the test data, the closer the distance, the higher the weight, and vice versa; Step S34: Based on the three-dimensional virtual model, the test data, the geological data at the data collection location of the test data and the rock parameters of the sub-area where the test data is located are combined into training data, and any three consecutive training data belonging to the corresponding area of ​​any geological type are selected as detection data, and the remaining two training data are used as learning data.

2. The method according to claim 1, characterized in that In step S1, filtering and interpolating the first geological image to obtain the second geological image includes: Step S11: abnormal pixel detection is performed on pixels in the first geological image, and a difference between the brightness value of each pixel and a first value is obtained. When the difference is greater than a set difference, the pixel is an abnormal pixel, and when the difference is less than or equal to the set difference, the pixel is a normal pixel, wherein the first value is the average of the maximum and minimum brightness values ​​of the pixels in the first geological image; Step S12: With the abnormal pixel as the center, within a set range, obtain the number of other abnormal pixels, and based on the number, obtain the normal pixels on a circle with a set radius around the abnormal pixel with the abnormal pixel as the center, and average the brightness values ​​of the normal pixels on the circle to obtain an interpolation value, wherein the larger the number, the smaller the set radius, and vice versa, the larger the radius, the number of abnormal pixels in the set range is less than the total number of pixels in the set range; Step S13: replacing the brightness value of each abnormal pixel with the interpolation value corresponding to the abnormal pixel.

3. The method according to claim 1, characterized in that The step S2 comprises: obtaining the geological type of the rock mass in the three-dimensional virtual model, selecting a landmark sampling area within the area corresponding to each geological type, selecting a plurality of sampling points at different positions in the sampling area, sampling the rock mass at each sampling point and performing in-situ compression to obtain the rock mass parameters corresponding to the sampling point, wherein the rock mass parameters include Young's modulus E, Poisson's ratio v, strain hardening modulus H, peak strain μ, and post-peak drop modulus M of the rock mass, and the geological types include at least folds, cracks, and faults.

4. The method according to claim 1, characterized in that: The step S4 comprises: Create the rock stress prediction model, obtain m geological types in the inversion area through the geological data, and create m sub-models based on the geological types, use the learning data in the sub-area corresponding to each geological type in the three-dimensional virtual model as the learning data of the corresponding sub-model, and train the sub-model through the learning data, wherein the value of m is a positive integer greater than or equal to 3.

5. The method according to claim 1, characterized in that The step S4 also includes: inputting the geological data and rock mass parameters in the detection data into the rock mass stress prediction model, and obtaining the geological type to which the detection data belongs according to the position of the detection data in the three-dimensional virtual model, inputting the corresponding sub-model according to the geological type, obtaining a prediction result, comparing the prediction result with the stress value in the detection data, and obtaining a comparison difference. When the comparison difference is less than or equal to a preset difference, the prediction result is accurate. When the comparison difference is greater than the preset difference and the prediction result satisfies the size relationship between the stress value in the detection data and the stress value in the closest learning data in the sub-region, the sub-model is retrained by increasing the number of learning data of the geological type. When the comparison difference is greater than the preset difference and the prediction result does not satisfy the size relationship between the detection data and the closest learning data in the sub-region, the step S2 is repeated to re-obtain the rock mass parameters corresponding to the geological type and retrain the sub-model.

6. The method according to claim 1, characterized in that The test data is the value of the geostress field at the data collection location.

7. A rock mass in-situ stress field inversion system based on machine learning, the system being used to implement the method according to any one of claims 1 to 6, characterized in that: The system comprises: A virtual model generation unit is used to obtain a first geological image by scanning and imaging the target rock mass, filter and interpolate the first geological image to obtain a second geological image, and construct a three-dimensional virtual model of the target rock mass based on the second geological image; and obtain geological data at each position in the inversion area based on the three-dimensional virtual model, wherein the geological data includes measurement data of faults, fractures and terrain trends; A marking unit, used to obtain the geological type in the inversion area based on the three-dimensional virtual model, select a number of sampling points within the area corresponding to each geological type, and obtain rock mass parameters corresponding to the sampling point by sampling and in-situ compressing the rock mass at each sampling point, and mark the rock mass parameters in the three-dimensional virtual model; A data generating unit, configured to obtain a plurality of test data respectively through hydraulic fracturing and drilling according to the sub-region where each sampling point is located, pre-process the test data, and generate learning data and detection data from the pre-processed test data, the rock mass parameters of the sampling points corresponding to the test data, and the geological data; A model training unit is used to create a rock stress prediction model, wherein the rock stress prediction model includes m sub-models, wherein the m sub-models are created based on the geological type, the sub-models are trained using the learning data, and the sub-models are verified based on the detection data, wherein each of the sub-models corresponds to a geological type.

8. A computer storage medium, characterized in that The storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

Images