Semantic segmentation-based roadway surrounding rock crack intelligent identification method, system and equipment

Abstract

The invention discloses a roadway surrounding rock crack intelligent identification method, system and device based on semantic segmentation, and the method comprises the steps: carrying out the semantic segmentation of a preprocessed roadway surrounding rock three-dimensional point cloud picture, and obtaining a plurality of roadway surrounding rock identification cracks; classifying the plurality of roadway surrounding rock identification fractures according to the number of the fractures identified by the same real fracture to obtain a first type of fractures and a second type of fractures; taking a plurality of roadway surrounding rock identification fractures identified belonging to the same real fracture in the second type of fractures as a group of fractures; fitting point cloud data in the preprocessed roadway surrounding rock three-dimensional point cloud picture to obtain a fracture reference surface; and converting the roadway surrounding rock identification fractures in the first type of fractures to the fracture reference surface to obtain a first fracture identification result, converting a group of roadway surrounding rock identification fractures in the second type of fractures to the fracture reference surface, and fitting the converted new fracture group to obtain a second fracture identification result. The accuracy of roadway surrounding rock crack recognition is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent identification technology for roadway surrounding rock fissures, and in particular to a method, system and device for intelligent identification of roadway surrounding rock fissures based on semantic segmentation. Background Technology

[0002] Stability assessment of surrounding rock in tunnels is a crucial step in ensuring safety, and accurate identification and analysis of rock fissures are key to evaluating rock stability. With the development of intelligent monitoring technology for underground engineering, rock fissure identification methods based on 3D point cloud data have become a research hotspot due to their ability to efficiently acquire spatial information about the rock surface.

[0003] However, existing technologies for identifying fractures in roadway surrounding rock face numerous challenges in practical applications. Due to the generally uneven surface of roadway surrounding rock after blasting, the same joint is often identified as two fractures. Furthermore, the unevenness of the roadway surrounding rock surface can cause identified fractures to not belong to the same plane, resulting in significant errors in the measurement of fracture spacing. These errors severely affect the accurate assessment of roadway surrounding rock stability and prevent the provision of reliable digital data for underground engineering safety monitoring.

[0004] Therefore, existing tunnel rock fracture identification technologies suffer from problems such as "multiple judgments for a single fracture" and large errors in fracture spacing measurement, resulting in relatively low identification accuracy. Summary of the Invention

[0005] This application aims to propose a method, system, and device for intelligent identification of rock fissures in roadways based on semantic segmentation, which can improve the accuracy of rock fissure identification and reduce the measurement error of fissure spacing.

[0006] In a first aspect, embodiments of this application provide a method for intelligent identification of rock fissures in roadways based on semantic segmentation, the method comprising:

[0007] Obtain the pre-processed 3D point cloud map of the surrounding rock of the tunnel;

[0008] Semantic segmentation is performed on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel to identify the fractures in the surrounding rock of the tunnel in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and multiple identified fractures in the surrounding rock of the tunnel are obtained.

[0009] The multiple roadway surrounding rock identified fractures are classified according to the number of fractures identified in the same real fracture, resulting in a first category of fractures and a second category of fractures. The first category of fractures is when only one roadway surrounding rock identification fracture is identified in the same real fracture, while the second category of fractures is when multiple roadway surrounding rock identification fractures are identified in the same real fracture.

[0010] Multiple roadway surrounding rock identified as fractures belonging to the same real fracture in the second category are grouped as a group of fractures.

[0011] By fitting the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway, the fracture reference surface is obtained.

[0012] The roadway surrounding rock identified fractures in the first category of fractures are converted to the fracture reference surface to obtain the first fracture identification result. In addition, a group of roadway surrounding rock identified fractures in the second category of fractures are converted to the fracture reference surface, and the new fracture group after conversion is fitted to obtain the second fracture identification result.

[0013] Compared with the prior art, the first aspect of this application has the following beneficial effects:

[0014] This method involves acquiring a pre-processed 3D point cloud map of the surrounding rock of a roadway; performing semantic segmentation on the pre-processed 3D point cloud map to identify fractures in the surrounding rock; obtaining multiple identified fractures; and classifying these fractures based on the number of fractures identified from the same real fracture, resulting in a first category and a second category. First category fractures are those where only one fracture from the same real fracture is identified, while second category fractures are those where only one fracture from the same real fracture is identified. Multiple roadway surrounding rock fractures were identified; multiple roadway surrounding rock fractures belonging to the same real fracture in the second category were grouped into a fracture group; the point cloud data in the preprocessed three-dimensional point cloud map of the roadway surrounding rock were fitted to obtain the fracture reference surface; the roadway surrounding rock fractures in the first category were converted to the fracture reference surface to obtain the first fracture identification result; and a group of roadway surrounding rock fractures in the second category were converted to the fracture reference surface, and the new fracture group after conversion was fitted to obtain the second fracture identification result. In this way, by classifying the fractures in the surrounding rock of the roadway identified by semantic segmentation, and taking multiple fractures identified in the surrounding rock of the roadway as a group of fractures, and then converting the multiple fractures to a fracture reference surface (that is, converting the three-dimensional point cloud data of the fractures to a two-dimensional fracture reference surface), and fitting them into a single fracture, the problems of "one fracture being identified as multiple fractures" and large measurement errors of fracture spacing are solved, thereby improving the accuracy of fracture identification in the surrounding rock of the roadway.

[0015] In some embodiments, obtaining the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway includes:

[0016] Obtain the maximum and minimum values ​​of the 3D point cloud data of the surrounding rock in the tunnel;

[0017] Scaling all the three-dimensional point cloud data of the surrounding rock of the tunnel to a value between the maximum and minimum values, where both are integers, yields a three-dimensional point cloud map of the surrounding rock of the tunnel.

[0018] The three-dimensional point cloud map of the surrounding rock of the tunnel is subjected to bilateral filtering for noise reduction to obtain a preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel.

[0019] In some embodiments, the semantic segmentation of the preprocessed 3D point cloud map of the surrounding rock of the tunnel is performed to identify fractures in the surrounding rock of the tunnel in the preprocessed 3D point cloud map, resulting in multiple identified fractures in the surrounding rock of the tunnel, including:

[0020] Obtain the training dataset and validation dataset;

[0021] A rock fissure identification model is constructed, and the training dataset is used to train the rock fissure identification model to obtain the trained rock fissure identification model.

[0022] The trained surrounding rock fracture identification model was validated using a validation dataset to obtain the target surrounding rock fracture identification model.

[0023] The target surrounding rock fracture identification model is used to perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and the tunnel surrounding rock fractures in the preprocessed three-dimensional point cloud map of the surrounding rock are identified to obtain multiple tunnel surrounding rock fractures.

[0024] In some embodiments, training the surrounding rock fracture identification model using the training dataset to obtain the trained surrounding rock fracture identification model includes:

[0025] Obtain the number of pixels belonging to the crack and the number of pixels not belonging to the crack in all labeled images in the training dataset;

[0026] Calculate the ratio between the number of pixels with cracks and the number of pixels without cracks;

[0027] Based on the ratio, construct the training loss function;

[0028] Based on the training loss function, the surrounding rock fracture identification model is trained using the training dataset to obtain the trained surrounding rock fracture identification model.

[0029] In some implementations, constructing the training loss function based on the ratio includes:

[0030] Multiply the reciprocal of the ratio by the number of misclassified crack pixels to obtain the first result;

[0031] The first result is added to the number of non-cracked pixels that were misclassified to obtain the second result;

[0032] The second result is compared with the total number of pixels in the three-dimensional point cloud map of the surrounding rock of the tunnel to construct the training loss function.

[0033] In some embodiments, the step of converting the identified roadway surrounding rock fractures in the first category of fractures to the fracture reference surface to obtain a first fracture identification result, and converting a group of identified roadway surrounding rock fractures in the second category of fractures to the fracture reference surface, and fitting the converted new fracture group to obtain a second fracture identification result, includes:

[0034] For the first type of fracture and the second type of fracture, calculate the angle between the joint surface corresponding to each identified fracture in the surrounding rock of the roadway and the fracture reference surface, and calculate the height difference between each identified fracture in the surrounding rock of the roadway and the fracture reference surface;

[0035] Based on the included angle and height difference corresponding to the first type of fracture, the roadway surrounding rock identified fractures in the first type of fracture are converted to the fracture reference plane to obtain the first fracture identification result;

[0036] Based on the included angle and height difference corresponding to the second type of fractures, a group of roadway surrounding rock identified fractures in the second type of fractures are converted to the fracture reference plane to obtain a new fracture group after conversion;

[0037] The transformed new fracture group was fitted using the least squares method to obtain the second fracture identification result.

[0038] In some embodiments, the step of converting the identified roadway surrounding rock fractures in the first category of fractures to the fracture reference plane based on the included angle and height difference corresponding to the first category of fractures to obtain the first fracture identification result includes:

[0039] Calculate the first conversion distance based on the included angle and height difference corresponding to the first type of crack;

[0040] Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and the first plane equation perpendicular to the joint surface and the fracture reference plane;

[0041] Calculate the equation of the first phase intersection line between the first plane equation and the fracture reference plane;

[0042] Based on the first intersection line equation and the first transformation distance, the identified fractures in the roadway surrounding rock of the first type of fracture are transformed to the fracture reference surface to obtain the first fracture identification result.

[0043] In some embodiments, the step of converting a group of roadway surrounding rock identification fractures in the second category of fractures to the fracture reference plane based on the included angle and height difference corresponding to the second category of fractures, to obtain a new fracture group after conversion, includes:

[0044] Calculate the second conversion distance based on the included angle and height difference corresponding to the second type of crack;

[0045] Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and the equation of the second plane perpendicular to the joint surface and the fracture reference plane;

[0046] Calculate the second phase intersection line equation between the second plane equation and the fracture reference plane;

[0047] Based on the second intersection line equation and the second transformation distance, a group of roadway surrounding rock identified fractures in the second category of fractures are transformed to the fracture reference surface to obtain a new fracture group after transformation.

[0048] Secondly, embodiments of this application also provide a roadway surrounding rock fracture intelligent identification system based on semantic segmentation, the system comprising:

[0049] The data acquisition unit is used to acquire the pre-processed three-dimensional point cloud map of the surrounding rock of the tunnel.

[0050] The semantic segmentation unit is used to perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, identify the fractures in the surrounding rock of the tunnel in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and obtain multiple identified fractures in the surrounding rock of the tunnel.

[0051] The fracture classification unit is used to classify the fractures identified in the surrounding rock of the multiple roadways according to the number of fractures identified in the same real fracture, to obtain a first category of fractures and a second category of fractures. The first category of fractures is when only one fracture in the surrounding rock of the same real fracture is identified, and the second category of fractures is when multiple fractures in the surrounding rock of the same real fracture are identified.

[0052] The fracture grouping unit is used to group multiple roadway surrounding rock identified fractures belonging to the same real fracture in the second category of fractures into a group of fractures.

[0053] The data fitting unit is used to fit the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway to obtain the fracture reference surface.

[0054] The fracture identification unit is used to convert the roadway surrounding rock identified fractures in the first category of fractures to the fracture reference surface to obtain a first fracture identification result, and to convert a group of roadway surrounding rock identified fractures in the second category of fractures to the fracture reference surface, and to fit the new fracture group after conversion to obtain a second fracture identification result.

[0055] Thirdly, embodiments of this application also provide an electronic device, including at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor, which are executed by the at least one control processor to enable the at least one control processor to perform a semantic segmentation-based intelligent identification method for roadway surrounding rock fissures as described above.

[0056] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the above-described semantic segmentation-based intelligent identification method for roadway surrounding rock fissures.

[0057] It is understood that the beneficial effects of the second to fourth aspects compared with the related technologies are the same as the beneficial effects of the first aspect compared with the related technologies. Please refer to the relevant description in the first aspect above, which will not be repeated here. Attached Figure Description

[0058] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0059] Figure 1 This is a flowchart illustrating an embodiment of the intelligent identification method for roadway surrounding rock fissures based on semantic segmentation provided in this application;

[0060] Figure 2 This is a schematic diagram of the same crack being repeatedly identified in the best embodiment of the intelligent identification method for roadway surrounding rock fissures based on semantic segmentation provided in this application;

[0061] Figure 3 This is a schematic diagram of the transformation of identified fractures to a fracture reference surface in the best embodiment of the intelligent identification method for roadway surrounding rock based on semantic segmentation provided in this application;

[0062] Figure 4 This is a schematic diagram of the best embodiment of the intelligent identification method for roadway surrounding rock fissures based on semantic segmentation provided in this application;

[0063] Figure 5 This is a schematic diagram of the structure of an embodiment of the electronic device provided in this application. Detailed Implementation

[0064] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0065] In the description of this application, the use of terms such as "first," "second," etc., is for the purpose of distinguishing technical features only and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated.

[0066] In the description of this application, it should be understood that the orientation descriptions, such as up, down, etc., are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0067] In the description of this application, it should be noted that, unless otherwise explicitly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0068] Because the surrounding rock surface of tunnels is generally uneven after blasting, the same joint is often identified as two fissures. Furthermore, the unevenness of the surrounding rock surface can cause the identified fissures to not belong to the same plane in space, resulting in significant errors in the measurement of fissure spacing. This error severely affects the accurate assessment of the stability of the surrounding rock in tunnels, and fails to provide reliable digital data for underground engineering safety monitoring.

[0069] To address the problems of "multiple judgments for a single fracture" and large measurement errors in fracture spacing in existing technologies, this application proposes a method, system, and device for intelligent identification of fractures in roadway surrounding rock based on semantic segmentation.

[0070] Reference Figure 1 This application provides a schematic flowchart of a semantic segmentation-based intelligent identification method for roadway surrounding rock fissures. This semantic segmentation-based method is applied to an electronic device, such as a server or a mobile terminal. Figure 1 As shown, the intelligent identification method for roadway surrounding rock fissures based on semantic segmentation may include the following steps:

[0071] Step S100: Obtain the pre-processed 3D point cloud map of the surrounding rock of the tunnel;

[0072] Step S200: Perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, identify the fractures in the surrounding rock of the tunnel in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and obtain multiple identified fractures in the surrounding rock of the tunnel.

[0073] Step S300: Classify the multiple roadway surrounding rock identified fractures according to the number of fractures identified by the same real fracture to obtain a first category fracture and a second category fracture. The first category fracture is when only one roadway surrounding rock identified fracture is the same real fracture, and the second category fracture is when multiple roadway surrounding rock identified fractures are the same real fracture.

[0074] Step S400: Collect multiple roadway surrounding rock identified as true fractures in the second category of fractures as a group of fractures;

[0075] Step S500: Fit the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway to obtain the fracture reference surface.

[0076] Step S600: Convert the identified fractures in the surrounding rock of the roadway in the first category of fractures to the fracture reference plane to obtain the first fracture identification result; and convert a group of identified fractures in the surrounding rock of the roadway in the second category of fractures to the fracture reference plane, and fit the new fracture group after conversion to obtain the second fracture identification result.

[0077] In this embodiment, a pre-processed 3D point cloud map of the surrounding rock of the roadway is acquired; semantic segmentation is performed on the pre-processed 3D point cloud map of the surrounding rock of the roadway to identify roadway surrounding rock fractures, resulting in multiple identified roadway surrounding rock fractures; these multiple identified roadway surrounding rock fractures are classified according to the number of fractures identified for the same real fracture, resulting in a first category of fractures and a second category of fractures. The first category of fractures consists of only one identified roadway surrounding rock fracture for the same real fracture, while the second category consists of fractures for the same real fracture. Multiple roadway surrounding rock fractures were identified; multiple roadway surrounding rock fractures belonging to the same real fracture in the second category were grouped into a fracture group; the point cloud data in the preprocessed three-dimensional point cloud map of the roadway surrounding rock was fitted to obtain the fracture reference surface; the roadway surrounding rock fractures in the first category were converted to the fracture reference surface to obtain the first fracture identification result; and a group of roadway surrounding rock fractures in the second category were converted to the fracture reference surface, and the new fracture group after conversion was fitted to obtain the second fracture identification result. In this way, by classifying the fractures in the surrounding rock of the roadway identified by semantic segmentation, and taking multiple fractures identified in the surrounding rock of the roadway as a group of fractures, and then converting the multiple fractures to a fracture reference surface (that is, converting the three-dimensional point cloud data of the fractures to a two-dimensional fracture reference surface), and fitting them into a single fracture, the problems of "one fracture being identified as multiple fractures" and large measurement errors of fracture spacing are solved, thereby improving the accuracy of fracture identification in the surrounding rock of the roadway.

[0078] The semantic segmentation of the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel can be performed by using the EAFNet surrounding rock fracture identification model. The EAFNet surrounding rock fracture identification model is the EAFNet network model in the existing technology for semantic segmentation. Its specific network model structure is not described in detail in this embodiment.

[0079] The point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel can be obtained by fitting the preprocessed point cloud data of the surrounding rock of the tunnel using the least squares method.

[0080] In some implementations, obtaining a pre-processed three-dimensional point cloud map of the surrounding rock of the roadway includes:

[0081] Obtain the maximum and minimum values ​​of the 3D point cloud data of the surrounding rock in the tunnel;

[0082] Scaling all the 3D point cloud data of the surrounding rock of the tunnel to between the maximum and minimum values, and both being integers, yields a 3D point cloud map of the surrounding rock of the tunnel.

[0083] The three-dimensional point cloud map of the surrounding rock of the tunnel is subjected to bilateral filtering for noise reduction to obtain the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel.

[0084] In this embodiment, the maximum and minimum values ​​of the 3D point cloud data of the surrounding rock of the tunnel are obtained; all the 3D point cloud data of the surrounding rock of the tunnel are scaled to between the maximum and minimum values ​​and are integers, to obtain a 3D point cloud map of the surrounding rock of the tunnel; the 3D point cloud map of the surrounding rock of the tunnel is then subjected to bilateral filtering for noise reduction to obtain a preprocessed 3D point cloud map of the surrounding rock of the tunnel. Thus, by performing infinitesimal hardening processing on the 3D point cloud data of the surrounding rock of the tunnel to obtain a 3D point cloud map of the surrounding rock of the tunnel, and then performing bilateral filtering for noise reduction on the 3D point cloud map of the surrounding rock of the tunnel, a good data foundation can be laid for correct semantic segmentation in the later stage.

[0085] The maximum and minimum values ​​of the three-dimensional point cloud data of the surrounding rock of the tunnel can be obtained by using a three-dimensional point cloud scanning device to obtain the three-dimensional point cloud data of the surrounding rock of the tunnel, or by taking pictures from multiple angles and using algorithms to generate the three-dimensional point cloud data of the surrounding rock of the tunnel, and then obtaining the maximum and minimum values ​​of the three-dimensional point cloud data of the surrounding rock of the tunnel. The minimum value of the three-dimensional point cloud data of the surrounding rock of the tunnel can be 0, and the maximum value can be 255.

[0086] In some implementations, semantic segmentation is performed on the preprocessed 3D point cloud map of the surrounding rock of the roadway to identify roadway surrounding rock fractures in the preprocessed 3D point cloud map, resulting in multiple identified roadway surrounding rock fractures, including:

[0087] Obtain the training dataset and validation dataset;

[0088] A rock fracture identification model was constructed, and the model was trained using a training dataset to obtain the trained rock fracture identification model.

[0089] The trained surrounding rock fracture identification model was validated using a validation dataset to obtain the target surrounding rock fracture identification model.

[0090] The target surrounding rock fracture identification model is used to perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway, and to identify the roadway surrounding rock fractures in the preprocessed three-dimensional point cloud map of the surrounding rock, resulting in multiple roadway surrounding rock fracture identifications.

[0091] In this embodiment, a rock fracture identification model is constructed by acquiring a training dataset and a validation dataset. The training dataset is used to train the model, resulting in a trained model. The validation dataset is then used to validate the trained model, yielding a target rock fracture identification model. This target model is then used to perform semantic segmentation on a preprocessed 3D point cloud image of the surrounding rock in the tunnel, identifying rock fractures within the preprocessed image and obtaining multiple identified rock fractures. By evaluating the results using the validation dataset, the optimal rock fracture identification model is obtained as the target model, thus improving the accuracy of the target rock fracture identification model.

[0092] In some implementations, a training dataset is used to train the surrounding rock fracture identification model to obtain a trained surrounding rock fracture identification model, including:

[0093] Obtain the number of pixels belonging to cracks and the number of pixels not belonging to cracks in all labeled images in the training dataset;

[0094] Calculate the ratio between the number of pixels with cracks and the number of pixels without cracks;

[0095] Construct a training loss function based on the ratio;

[0096] Based on the training loss function, the surrounding rock fracture identification model is trained using the training dataset to obtain the trained surrounding rock fracture identification model.

[0097] In this embodiment, the number of pixels belonging to fractures and the number of pixels not belonging to fractures in all labeled images in the training dataset are obtained; the ratio between the number of pixels belonging to fractures and the number of pixels not belonging to fractures is calculated; a training loss function is constructed based on the ratio; and the rock fracture identification model is trained using the training dataset based on the training loss function to obtain the trained rock fracture identification model. Thus, by constructing the training loss function using the ratio between the number of pixels belonging to fractures and the number of pixels not belonging to fractures, the identification accuracy of the trained rock fracture identification model is improved.

[0098] In some implementations, a training loss function is constructed based on the ratio, including:

[0099] Multiply the reciprocal of the ratio by the number of misclassified crack pixels to obtain the first result;

[0100] The first result is added to the number of non-slit pixels that were misclassified to obtain the second result;

[0101] The second result is compared with the total number of pixels in the 3D point cloud map of the surrounding rock of the tunnel to construct the training loss function.

[0102] In this embodiment, a first result is obtained by multiplying the reciprocal of the ratio by the number of misclassified fracture pixels; a second result is obtained by adding the first result to the number of misclassified non-fracture pixels; and a training loss function is constructed by comparing the second result with the total number of pixels in the 3D point cloud map of the surrounding rock. Thus, by combining the number of misclassified fracture pixels and the number of misclassified non-fracture pixels, the constructed training loss function can better assist in training the surrounding rock fracture identification model, further improving the identification accuracy of the trained model.

[0103] In some implementations, the identified fractures in the surrounding rock of the roadway in the first category of fractures are converted to a fracture reference plane to obtain a first fracture identification result; and a group of identified fractures in the surrounding rock of the roadway in the second category of fractures are converted to a fracture reference plane, and the converted new fracture group is fitted to obtain a second fracture identification result, including:

[0104] For the first category of fractures and the second category of fractures, calculate the angle between the joint surface and the fracture reference surface corresponding to each fracture in the surrounding rock of the roadway, and calculate the height difference between each fracture in the surrounding rock of the roadway and the fracture reference surface;

[0105] Based on the included angle and height difference corresponding to the first category of fractures, the fractures in the surrounding rock of the roadway in the first category of fractures are converted to the fracture reference plane to obtain the first fracture identification result;

[0106] Based on the included angle and height difference corresponding to the second category of fractures, a group of roadway surrounding rock identified fractures in the second category of fractures are converted to the fracture reference plane to obtain a new fracture group after conversion;

[0107] The least squares method was used to fit the transformed new fracture group to obtain the second fracture identification result.

[0108] In this embodiment, for both first-category and second-category fractures, the angle between the joint surface and the fracture reference surface corresponding to each fracture in the surrounding rock of the roadway is calculated, as well as the height difference between each fracture and the fracture reference surface. Based on the angle and height difference corresponding to the first-category fractures, the fractures in the surrounding rock of the roadway in the first-category fractures are converted to the fracture reference surface to obtain the first fracture identification result. Based on the angle and height difference corresponding to the second-category fractures, a group of fractures in the surrounding rock of the roadway in the second-category fractures are converted to the fracture reference surface to obtain a new fracture group after conversion. The new fracture group after conversion is fitted using the least squares method to obtain the second fracture identification result. By fitting the new fracture group after conversion using the least squares method, a group of fractures in the surrounding rock of the roadway belonging to the same real fracture will have only one fracture identification result, solving the problem of "multiple judgments for one fracture". Furthermore, converting the three-dimensional point cloud data of the fractures to the two-dimensional fracture reference surface can solve the problem of large measurement errors in fracture spacing, thereby improving the accuracy of fracture identification in the surrounding rock of the roadway.

[0109] In some implementations, based on the included angle and height difference corresponding to the first type of fracture, the identified fractures in the surrounding rock of the roadway within the first type of fracture are converted to a fracture reference plane to obtain the first fracture identification result, including:

[0110] Calculate the first conversion distance based on the included angle and height difference corresponding to the first type of crack;

[0111] Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and calculate the first plane equation perpendicular to the joint surface and the fracture reference plane;

[0112] Calculate the equation of the first phase intersection line between the first plane equation and the fracture reference plane;

[0113] Based on the first phase intersection line equation and the first transformation distance, the identified fractures in the roadway surrounding rock of the first category of fractures are transformed to the fracture reference plane to obtain the first fracture identification result.

[0114] In this embodiment, a first conversion distance is calculated based on the included angle and height difference corresponding to the first type of fracture; the fracture points on the identified fractures in the surrounding rock of the roadway and their projection points on the fracture reference plane are calculated, and a first plane equation perpendicular to the joint surface and the fracture reference plane is calculated; a first intersection line equation between the first plane equation and the fracture reference plane is calculated; based on the first intersection line equation and the first conversion distance, the identified fractures in the surrounding rock of the roadway in the first type of fracture are converted to the fracture reference plane to obtain the first fracture identification result. Thus, by converting the identified fractures in the surrounding rock of the roadway in the first type of fracture to the fracture reference plane, and converting the three-dimensional point cloud data of the fractures to the two-dimensional fracture reference plane, the measurement error of the fracture spacing can be reduced.

[0115] In some implementations, based on the included angle and height difference corresponding to the second category of fractures, a group of roadway surrounding rock identified fractures in the second category are converted to a fracture reference plane to obtain a new fracture group after conversion, including:

[0116] Calculate the second conversion distance based on the included angle and height difference corresponding to the second type of fracture;

[0117] Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and calculate the second plane equation perpendicular to the joint surface and the fracture reference plane;

[0118] Calculate the equation of the second phase intersection line between the second plane equation and the fracture reference plane;

[0119] Based on the second phase intersection line equation and the second transformation distance, a group of roadway surrounding rock identified fractures in the second category of fractures are transformed to the fracture reference plane to obtain a new fracture group after transformation.

[0120] In this embodiment, a second conversion distance is calculated based on the included angle and height difference corresponding to the second type of fractures; the fracture points on the identified fractures in the surrounding rock of the roadway and their projection points on the fracture reference plane are calculated, and a second plane equation perpendicular to the joint surface and the fracture reference plane is calculated; a second intersection line equation between the second plane equation and the fracture reference plane is calculated; based on the second intersection line equation and the second conversion distance, a group of identified fractures in the surrounding rock of the roadway in the second type of fractures is converted to the fracture reference plane, resulting in a new fracture group after conversion. Thus, by converting a group of identified fractures in the surrounding rock of the roadway in the second type of fractures to the fracture reference plane, a good data foundation is laid for subsequent fracture fitting, which can improve the accuracy of fracture fitting results. Furthermore, converting the three-dimensional point cloud data of the fractures to the two-dimensional fracture reference plane can reduce the measurement error of fracture spacing.

[0121] To facilitate understanding by those skilled in the art, a set of preferred embodiments is provided below:

[0122] Stability assessment of surrounding rock in tunnels is a crucial step in ensuring safety, and accurate identification and analysis of rock fissures are key to evaluating rock stability. With the development of intelligent monitoring technology for underground engineering, rock fissure identification methods based on 3D point cloud data have become a research hotspot due to their ability to efficiently acquire spatial information about the rock surface.

[0123] However, existing technologies for identifying fractures in roadway surrounding rock face numerous challenges in practical applications. Due to the generally uneven surface of the roadway surrounding rock after blasting, the same joint is often identified as two fractures, such as... Figure 2 As shown, Figure 2 Image (a) is a schematic diagram of a real fracture. Figure 2 (b) shows a schematic diagram of the two identified cracks. Furthermore, the unevenness of the surrounding rock surface can cause the identified cracks to not belong to the same plane in space, resulting in significant errors in the measurement of crack spacing. This error severely affects the accurate assessment of the stability of the surrounding rock in the tunnel, making it impossible to provide reliable digital data for underground engineering safety monitoring.

[0124] There is an urgent need for a method that can accurately identify fissures in the surrounding rock of tunnels and effectively optimize the identification results, in order to solve the problems of repeated identification of fissures and measurement errors in spacing caused by uneven surrounding rock surfaces, and promote the intelligent upgrading of underground engineering safety monitoring.

[0125] Therefore, this embodiment proposes an intelligent identification method for roadway surrounding rock fissures based on semantic segmentation of 3D point cloud data. This method includes dimensionless point cloud data processing and bilateral filtering denoising, used to process the collected 3D point cloud data of the roadway surrounding rock into a 3D point cloud map of the roadway surrounding rock, followed by denoising. A trained EAFNet surrounding rock fissure identification model is used to classify the preprocessed 3D point cloud map of the roadway surrounding rock into two categories: fissures and background. Then, surrounding rock fissure optimization technology is used to optimize the fissures identified by the EAFNet surrounding rock fissure identification model. This embodiment uses the EAFNet semantic segmentation network, which has a high-precision fissure identification effect, to extract roadway surrounding rock fissures. The surrounding rock fissure optimization technology further optimizes repeatedly extracted fissures, solving the problem of repeated fissure identification caused by uneven roadway surrounding rock surfaces and reducing the measurement error of fissure spacing caused by fissures not being on the same plane. This achieves intelligent and high-precision identification of roadway surrounding rock fissures, providing a more reliable digital basis for roadway surrounding rock stability assessment and contributing to the intelligent upgrade of underground engineering safety monitoring.

[0126] The technical solution of this embodiment specifically includes the following:

[0127] Step 1: Preprocess the three-dimensional point cloud data of the surrounding rock of the tunnel to obtain a three-dimensional point cloud image of the surrounding rock of the tunnel.

[0128] Specifically, the maximum and minimum values ​​of the 3D point cloud data of the surrounding rock are obtained. Based on the maximum and minimum values, the 3D point cloud data is scaled to 0 to 255, meaning the minimum value of the 3D point cloud data is 0 and the maximum value is 255. All data are then rounded to integers to form a 3D point cloud map of the roadway surrounding rock. Bilateral filtering is then applied to the 3D point cloud image of the roadway surrounding rock for noise reduction.

[0129] Because the data range of 3D point cloud data is not fixed, and the greater the elevation difference of the collected object, the larger the data range of the 3D point cloud data, the data range of the 3D point cloud data is scaled so that the pixel values ​​of the obtained 3D point cloud image of the roadway surrounding rock are within the range of [0, 255], and all pixel values ​​are integers. Preprocessing the 3D point cloud data into a 3D point cloud image in this way facilitates subsequent deep learning processing, thus laying a good data foundation for later image semantic segmentation.

[0130] Step 2: Perform semantic segmentation on the three-dimensional point cloud image of the surrounding rock of the tunnel, thereby dividing the point cloud data in the three-dimensional point cloud image of the surrounding rock into two categories: cracks and background.

[0131] Specifically, the training of the EAFNet surrounding rock fracture identification model includes the following steps:

[0132] At least 1000 preprocessed 3D point cloud images of the surrounding rock of the tunnel are obtained for model training.

[0133] Use annotation software or image processing technology to mark the pre-processed three-dimensional point cloud image of the surrounding rock of the tunnel.

[0134] The labeled 3D point cloud images of the surrounding rock of the tunnel were augmented by rotating them 90°, 180°, 270° and flipping them horizontally to form a 3D point cloud image dataset of the surrounding rock of the tunnel.

[0135] The 3D point cloud image dataset of the surrounding rock of the tunnel was divided into a training dataset and a validation dataset in an 8:2 ratio.

[0136] Calculate the ratio of the number of pixels belonging to the crack to the number of pixels belonging to other classes (i.e., the number of non-crack pixels) in all labeled images of the training dataset, and denote it as 1:A. Then, construct the model training loss (i.e., the training loss function) as: Model training loss = (A × number of pixels misclassified as crack class + number of pixels misclassified as other classes) / total number of pixels in the image.

[0137] The training parameters for the EAFNet rock fissure identification model are mainly the learning rate, training period, and batch size.

[0138] Based on the model training loss, the EAFNet surrounding rock fracture identification model is trained cyclically using the training dataset. After each cycle, the model performance is evaluated using the validation dataset. The calculation method is: model performance = 1 - model training loss. The larger the model performance value, the better the performance of the EAFNet surrounding rock fracture identification model.

[0139] After training, the optimal EAFNet surrounding rock fracture identification model (i.e., the target surrounding rock fracture identification model) is obtained based on the evaluation results of the validation dataset.

[0140] The optimal EAFNet surrounding rock fracture identification model is used to perform semantic segmentation on the three-dimensional point cloud image of the surrounding rock in the roadway, thereby dividing the point cloud data in the three-dimensional point cloud image of the surrounding rock into two categories: fracture and background.

[0141] Step 3: Optimize the identification results of the surrounding rock fissures in the tunnel.

[0142] Specifically, the least squares method is used to fit the point cloud data in the three-dimensional point cloud map of the surrounding rock of the tunnel to form a plane, which serves as the reference surface for the fracture.

[0143] Based on the three-dimensional point cloud map of the surrounding rock of the tunnel, determine whether each identified crack (i.e., the crack identified after semantic segmentation by the EAFNet surrounding rock crack identification model in step 2) is generated by the same real crack as its similar identified crack; this process can be judged manually.

[0144] Based on the above judgment results, the identified cracks are divided into two categories: the first category is where only one crack is identified for a real crack, and the second category is where multiple cracks are identified for a real crack, and the multiple cracks identified by the real crack are grouped together.

[0145] Calculate the angle between the joint surface corresponding to each fracture and the fracture reference surface in the three-dimensional point cloud map of the surrounding rock of the tunnel, and calculate the height difference between the fracture and the fracture reference surface.

[0146] Based on the included angle and height difference, the identified cracks are transferred to the crack reference plane, such as... Figure 3 As shown. The angle between the joint surface corresponding to each fracture and the fracture reference surface can be obtained by calculating using the three-dimensional point cloud data of the joint surface, and the height difference between the fracture and the fracture reference surface can be obtained by calculating using the three-dimensional point cloud data of the fracture.

[0147] Specifically, the location and elevation of the fracture point are extracted from the three-dimensional point cloud map of the surrounding rock of the tunnel, and the height difference H between the fracture point and the fracture reference surface is calculated based on the location and elevation. The location and elevation of the point on the joint surface corresponding to the identified fracture where the fracture point is located are extracted from the three-dimensional point cloud map of the surrounding rock of the tunnel, and then the equation of the joint surface is fitted. The number of fracture points should be greater than or equal to 3. More fracture points result in more accurate fitting results, but increase computational complexity. Therefore, the number of fracture points can be selected according to the actual situation. This embodiment does not impose a specific limitation. Calculate the angle θ between the two surfaces based on the equations of the joint surface and the reference surface (the normal vectors of the two surfaces can be calculated first, and then the angle can be calculated based on the dot product formula of the normal vectors). Find the equation of the plane passing through the fracture point and its projection point on the fracture reference surface, perpendicular to the joint surface and the fracture reference surface, and find the equation of the intersection line between this plane equation and the fracture reference surface. Calculate the transformation distance L based on the height difference H between the fracture point and the fracture reference surface and the angle θ between the joint surface and the reference surface. The transformation point of the fracture point on the fracture reference surface can be obtained based on the intersection line equation and the transformation distance L, thus realizing the transformation of the identified fracture onto the fracture reference surface. The height difference H, the equation of the joint surface, the included angle θ, the plane equation of the fracture reference surface, the equation of the intersection line, and the transformation distance L can all be calculated using calculation methods known to those skilled in the art. This embodiment does not describe the specific calculation process in detail. For example, the least squares method can be used to fit the equation of the joint surface based on the position and elevation of the points on the joint surface.

[0148] Joint surfaces, also known as structural surfaces, are natural fissures or fracture surfaces in rock masses. Their geometric characteristics (such as occurrence, roughness, and spacing) are crucial for engineering geological stability analysis.

[0149] For the first type of fracture, the new fracture formed on the fracture reference surface is taken as the final fracture identification result (i.e., the first fracture identification result); for the second type of fracture group, the new fracture group formed on the fracture reference surface is fitted to form a new fracture, which is taken as the final fracture identification result (i.e., the second fracture identification result).

[0150] Compared with the prior art, the method of this embodiment has the following advantages:

[0151] This embodiment can solve the problems of repeated crack identification and large spacing measurement errors caused by uneven surrounding rock surfaces, thereby improving the accuracy of crack identification in roadway surrounding rock.

[0152] Reference Figure 4 This application also provides a semantic segmentation-based intelligent identification system for roadway surrounding rock fractures. The system includes a data acquisition unit 100, a semantic segmentation unit 200, a fracture classification unit 300, a fracture grouping unit 400, a data fitting unit 500, and a fracture identification unit 600, wherein:

[0153] Data acquisition unit 100 is used to acquire the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway;

[0154] Semantic segmentation unit 200 is used to perform semantic segmentation on the pre-processed three-dimensional point cloud map of the surrounding rock of the tunnel, identify the fractures in the surrounding rock of the tunnel in the pre-processed three-dimensional point cloud map of the surrounding rock of the tunnel, and obtain multiple identified fractures in the surrounding rock of the tunnel.

[0155] The fracture classification unit 300 is used to classify multiple roadway surrounding rock fractures according to the number of fractures identified by the same real fracture, resulting in a first category fracture and a second category fracture. The first category fracture is when only one roadway surrounding rock fracture is identified by the same real fracture, while the second category fracture is when multiple roadway surrounding rock fractures are identified by the same real fracture.

[0156] The fracture grouping unit 400 is used to group multiple roadway surrounding rock identified fractures belonging to the same real fracture in the second category of fractures into a group of fractures.

[0157] Data fitting unit 500 is used to fit the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway to obtain the fracture reference surface.

[0158] The fracture identification unit 600 is used to convert the fractures identified in the surrounding rock of the roadway in the first category of fractures to the fracture reference plane to obtain the first fracture identification result, and to convert a group of fractures identified in the surrounding rock of the roadway in the second category of fractures to the fracture reference plane, and to fit the new fracture group after conversion to obtain the second fracture identification result.

[0159] It should be noted that since the intelligent identification system for roadway surrounding rock fissures based on semantic segmentation in this embodiment is based on the same inventive concept as the intelligent identification method for roadway surrounding rock fissures based on semantic segmentation described above, the corresponding content in the method embodiment is also applicable to this system embodiment, and will not be described in detail here.

[0160] Reference Figure 5 This application also provides an electronic device, which includes:

[0161] At least one memory;

[0162] At least one processor;

[0163] At least one program;

[0164] The program is stored in memory, and the processor executes at least one program to implement the above-described method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation.

[0165] This electronic device can be any smart terminal, including mobile phones, tablets, personal digital assistants (PDAs), and in-vehicle computers.

[0166] The electronic devices according to embodiments of this application will now be described in detail.

[0167] The processor 1600 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this disclosure.

[0168] The memory 1700 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1700 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1700 and is called and executed by the processor 1600 to execute the semantic segmentation-based intelligent identification method for roadway surrounding rock fractures according to the embodiments of this disclosure.

[0169] The input / output interface 1800 is used to implement information input and output.

[0170] The communication interface 1900 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0171] Bus 2000 transmits information between various components of the device (e.g., processor 1600, memory 1700, input / output interface 1800, and communication interface 1900);

[0172] The processor 1600, memory 1700, input / output interface 1800 and communication interface 1900 are connected to each other within the device via bus 2000.

[0173] This disclosure also provides a storage medium, which is a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the above-described method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation.

[0174] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0175] The embodiments described in this disclosure are for the purpose of more clearly illustrating the technical solutions of this disclosure and do not constitute a limitation on the technical solutions provided by this disclosure. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by this disclosure are also applicable to similar technical problems.

[0176] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this disclosure, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0177] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0178] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0179] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0180] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0181] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0182] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0183] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0184] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. The embodiments of this application have been described in detail above with reference to the accompanying drawings, but this application is not limited to the above embodiments. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of this application.

[0185] The embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of this application.

Claims

1. A method for intelligent identification of fractures in roadway surrounding rock based on semantic segmentation, characterized in that, The method includes: Obtain the pre-processed 3D point cloud map of the surrounding rock of the tunnel; Semantic segmentation is performed on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel to identify the fractures in the surrounding rock of the tunnel in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and multiple identified fractures in the surrounding rock of the tunnel are obtained. The multiple roadway surrounding rock identified fractures are classified according to the number of fractures identified in the same real fracture, resulting in a first category of fractures and a second category of fractures. The first category of fractures is when only one roadway surrounding rock identification fracture is identified in the same real fracture, while the second category of fractures is when multiple roadway surrounding rock identification fractures are identified in the same real fracture. Multiple roadway surrounding rock identified as fractures belonging to the same real fracture in the second category are grouped as a group of fractures. By fitting the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway, the fracture reference surface is obtained. The roadway surrounding rock identified fractures in the first category of fractures are converted to the fracture reference surface to obtain the first fracture identification result. In addition, a group of roadway surrounding rock identified fractures in the second category of fractures are converted to the fracture reference surface, and the new fracture group after conversion is fitted to obtain the second fracture identification result.

2. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 1, characterized in that, The acquisition of the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway includes: Obtain the maximum and minimum values ​​of the 3D point cloud data of the surrounding rock in the tunnel; Scaling all the three-dimensional point cloud data of the surrounding rock of the tunnel to a value between the maximum and minimum values, where both are integers, yields a three-dimensional point cloud map of the surrounding rock of the tunnel. The three-dimensional point cloud map of the surrounding rock of the tunnel is subjected to bilateral filtering for noise reduction to obtain a preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel.

3. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 1, characterized in that, The preprocessed 3D point cloud map of the surrounding rock of the tunnel is semantically segmented to identify fractures in the surrounding rock, resulting in multiple identified fractures, including: Obtain the training dataset and validation dataset; A rock fissure identification model is constructed, and the training dataset is used to train the rock fissure identification model to obtain the trained rock fissure identification model. The trained surrounding rock fracture identification model was validated using a validation dataset to obtain the target surrounding rock fracture identification model. The target surrounding rock fracture identification model is used to perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and the tunnel surrounding rock fractures in the preprocessed three-dimensional point cloud map of the surrounding rock are identified to obtain multiple tunnel surrounding rock fractures.

4. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 3, characterized in that, The step of training the surrounding rock fracture identification model using the training dataset to obtain the trained surrounding rock fracture identification model includes: Obtain the number of pixels belonging to the crack and the number of pixels not belonging to the crack in all labeled images in the training dataset; Calculate the ratio between the number of pixels with cracks and the number of pixels without cracks; Based on the ratio, construct the training loss function; Based on the training loss function, the surrounding rock fracture identification model is trained using the training dataset to obtain the trained surrounding rock fracture identification model.

5. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 4, characterized in that, The step of constructing the training loss function based on the ratio includes: Multiply the reciprocal of the ratio by the number of misclassified crack pixels to obtain the first result; The first result is added to the number of non-cracked pixels that were misclassified to obtain the second result; The second result is compared with the total number of pixels in the three-dimensional point cloud map of the surrounding rock of the tunnel to construct the training loss function.

6. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 1, characterized in that, The process of converting the identified roadway surrounding rock fractures in the first category of fractures to the fracture reference surface to obtain a first fracture identification result, and converting a group of identified roadway surrounding rock fractures in the second category of fractures to the fracture reference surface, and fitting the converted new fracture group to obtain a second fracture identification result, includes: For the first type of fracture and the second type of fracture, calculate the angle between the joint surface corresponding to each identified fracture in the surrounding rock of the roadway and the fracture reference surface, and calculate the height difference between each identified fracture in the surrounding rock of the roadway and the fracture reference surface; Based on the included angle and height difference corresponding to the first type of fracture, the roadway surrounding rock identified fractures in the first type of fracture are converted to the fracture reference plane to obtain the first fracture identification result; Based on the included angle and height difference corresponding to the second type of fractures, a group of roadway surrounding rock identified fractures in the second type of fractures are converted to the fracture reference plane to obtain a new fracture group after conversion; The transformed new fracture group was fitted using the least squares method to obtain the second fracture identification result.

7. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 6, characterized in that, The step of converting the identified roadway surrounding rock fractures in the first category of fractures to the fracture reference plane based on the included angle and height difference corresponding to the first category of fractures to obtain the first fracture identification result includes: Calculate the first conversion distance based on the included angle and height difference corresponding to the first type of crack; Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and the first plane equation perpendicular to the joint surface and the fracture reference plane; Calculate the equation of the first phase intersection line between the first plane equation and the fracture reference plane; Based on the first intersection line equation and the first transformation distance, the identified fractures in the roadway surrounding rock of the first type of fracture are transformed to the fracture reference surface to obtain the first fracture identification result.

8. The method for intelligent identification of roadway surrounding rock fissures based on semantic segmentation according to claim 6, characterized in that, The step of converting a group of roadway surrounding rock identification fractures in the second category of fractures to the fracture reference plane based on the included angle and height difference corresponding to the second category of fractures, to obtain a new fracture group after conversion, includes: Calculate the second conversion distance based on the included angle and height difference corresponding to the second type of crack; Calculate the fracture points on the identified fractures in the surrounding rock of the tunnel and the projection points of the fracture points on the fracture reference plane, and calculate the second plane equation perpendicular to the joint surface and the fracture reference plane; Calculate the second phase intersection line equation between the second plane equation and the fracture reference plane; Based on the second intersection line equation and the second transformation distance, a group of roadway surrounding rock identified fractures in the second category of fractures are transformed to the fracture reference surface to obtain a new fracture group after transformation.

9. A semantic segmentation-based intelligent identification system for roadway surrounding rock fissures, characterized in that, The system includes: The data acquisition unit is used to acquire the pre-processed three-dimensional point cloud map of the surrounding rock of the tunnel. The semantic segmentation unit is used to perform semantic segmentation on the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, identify the fractures in the surrounding rock of the tunnel in the preprocessed three-dimensional point cloud map of the surrounding rock of the tunnel, and obtain multiple identified fractures in the surrounding rock of the tunnel. The fracture classification unit is used to classify the fractures identified in the surrounding rock of the multiple roadways according to the number of fractures identified in the same real fracture, to obtain a first category of fractures and a second category of fractures. The first category of fractures is when only one fracture in the surrounding rock of the same real fracture is identified, and the second category of fractures is when multiple fractures in the surrounding rock of the same real fracture are identified. The fracture grouping unit is used to group multiple roadway surrounding rock identified fractures belonging to the same real fracture in the second category of fractures into a group of fractures. The data fitting unit is used to fit the point cloud data in the preprocessed three-dimensional point cloud map of the surrounding rock of the roadway to obtain the fracture reference surface. The fracture identification unit is used to convert the roadway surrounding rock identified fractures in the first category of fractures to the fracture reference surface to obtain a first fracture identification result, and to convert a group of roadway surrounding rock identified fractures in the second category of fractures to the fracture reference surface, and to fit the new fracture group after conversion to obtain a second fracture identification result.

10. An electronic device, characterized in that, It includes at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor, which, when executed by the at least one control processor, enable the at least one control processor to perform the semantic segmentation-based intelligent identification method for roadway surrounding rock fractures as described in any one of claims 1 to 8.

Images