Method for detecting internal defects of lining based on three-dimensional point cloud

Abstract

The application provides a lining internal defect detection method based on a three-dimensional point cloud, comprising: scanning a tunnel lining by using a three-dimensional laser scanning device, constructing initial three-dimensional point cloud data of the tunnel lining, constructing a three-dimensional lining structure of the tunnel lining by using the initial three-dimensional point cloud data, identifying a normal vector and a curvature value of each initial three-dimensional point cloud data, determining an abnormal area in the tunnel lining, performing fine scanning on the abnormal area by using the three-dimensional laser scanning device respectively, obtaining corresponding fine three-dimensional point cloud data, constructing a regional structure corresponding to the abnormal area, determining a defect appearance of the abnormal area according to a presentation feature of the regional structure and displaying the defect appearance, and solving problems such as low detection precision, great interference, and inability to accurately quantify defect parameters existing in the existing lining internal defect detection method, so that high-precision, comprehensive and quantitative detection of lining internal defects is realized.

Description

Technical Field

[0001] This invention relates to the field of defect identification technology, and in particular to a method for detecting internal defects in linings based on three-dimensional point clouds. Background Technology

[0002] In various engineering structures, the lining serves as a crucial support and protective structure, and its quality directly impacts the safety and service durability of the project. During the construction phase and throughout its entire lifecycle, the lining structure is prone to internal defects due to the combined effects of multiple factors, including typical abnormal defects such as voids, cracks, and spalling. If these defects are not identified and addressed in a timely manner, they will deteriorate and expand over time, significantly weakening the load-bearing capacity and mechanical stability of the lining structure, and in severe cases, potentially triggering catastrophic engineering accidents.

[0003] Currently, commonly used methods for detecting internal defects in linings include ultrasonic testing, rebound hammer testing, and electromagnetic induction testing. Ultrasonic testing determines defects by emitting ultrasonic waves and receiving the reflected waves, but its accuracy is limited by the inhomogeneity of the lining material. Rebound hammer testing relies on the relationship between the surface hardness and strength of concrete, only reflecting surface conditions and unable to detect deep internal defects. Electromagnetic induction testing is sensitive to metal components; in cases with abundant steel reinforcement and other metal structures within the lining, interference can easily occur, affecting the accuracy of defect identification. Therefore, improving the accuracy of internal defect detection in linings has become an urgent problem to be solved.

[0004] Therefore, the present invention provides a method for detecting internal defects in linings based on three-dimensional point clouds. Summary of the Invention

[0005] This invention provides a method for detecting internal defects in linings based on three-dimensional point clouds. This method addresses the problems of low detection accuracy, high susceptibility to interference, and inability to accurately quantify defect parameters in existing methods for detecting internal defects in linings. It achieves high-precision, comprehensive, and quantitative detection of internal defects in linings.

[0006] This invention provides a method for detecting internal defects in linings based on three-dimensional point clouds, including:

[0007] Step 1: Use a 3D laser scanning device to scan the tunnel lining, and use the collected geometric coordinate information and reflection intensity information to construct the initial 3D point cloud data of the tunnel lining;

[0008] Step 2: Construct the three-dimensional lining structure of the tunnel lining based on the initial three-dimensional point cloud data, and identify the normal vector and curvature value corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure;

[0009] Step 3: Identify the abnormal areas of the tunnel lining based on the normal vector and the curvature value, and use the three-dimensional laser scanning device to perform a fine scan on each abnormal area to obtain the corresponding fine three-dimensional point cloud data;

[0010] Step 4: Construct a regional structure corresponding to the abnormal area using the refined 3D point cloud data, determine the defect appearance of the abnormal area based on the presentation characteristics of the regional structure, and mark it for display in the 3D lining structure.

[0011] In one feasible approach

[0012] Step 1 includes:

[0013] Step 11: Control the three-dimensional laser scanning device to perform laser scanning on the tunnel lining at different scanning angles, obtain the laser scanning results of the three-dimensional laser scanning device at each scanning angle, and obtain the scanning feedback results of the tunnel lining at different scanning angles;

[0014] Step 12: Pair the laser scanning results and scanning feedback results corresponding to the same scanning angle to obtain several scanning reflection points of the tunnel lining. Construct a three-dimensional mapping space based on the scanning intersection features between different scanning angles, and locate each scanning reflection point in the three-dimensional mapping space.

[0015] Step 13: Based on the positioning structure, map the laser scanning results and the scanning feedback results into the three-dimensional mapping space to generate the laser scanning process and obtain the geometric coordinate information and reflection intensity information of the tunnel lining;

[0016] Step 14: Merge the geometric coordinate information and the reflection intensity information, and generate the initial three-dimensional point cloud data of the tunnel lining based on the geometric coordinates and reflection intensity corresponding to each laser point.

[0017] In one feasible approach

[0018] Also includes:

[0019] The three-dimensional laser scanning device is controlled to determine several scanning angles based on a preset angle difference;

[0020] The three-dimensional laser scanning device is controlled to perform laser scanning on the tunnel lining at each of the scanning angles.

[0021] In one feasible approach

[0022] Constructing the three-dimensional lining structure of the tunnel lining based on the initial three-dimensional point cloud data includes:

[0023] Step 21: Calculate the first data density of the initial three-dimensional point cloud data. When the first data density is higher than the specified density, convert each initial three-dimensional point cloud data into voxels and obtain the voxel center point, voxel average point and voxel dense point corresponding to each voxel. Convert each initial three-dimensional point cloud data into the corresponding center three-dimensional point cloud data, average point three-dimensional point cloud data and dense three-dimensional point cloud data to generate a candidate point cloud dataset.

[0024] Step 22: Randomly extract candidate point cloud data from each of the candidate point cloud datasets to construct several low-dimensional three-dimensional point cloud data for the tunnel lining, and select the target low-dimensional three-dimensional point cloud data with the second data density lower than the specified density and the highest similarity to the initial three-dimensional point cloud data.

[0025] Step 23: Identify the geometric appearance features corresponding to each of the target low-dimensional three-dimensional point cloud data, and identify the geometric overlap features between different target low-dimensional three-dimensional point cloud data, and construct the three-dimensional lining structure of the tunnel lining based on the geometric appearance features and the geometric overlap features.

[0026] In one feasible approach

[0027] Identifying the normal vector and curvature value corresponding to each initial three-dimensional point cloud data in the three-dimensional lining structure includes:

[0028] Step 24: Identify the structural points corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure, and perform planar projection on the three-dimensional lining structure from a preset projection perspective to obtain the corresponding two-dimensional lining image, thereby obtaining the structural point distribution characteristics corresponding to each of the two-dimensional lining images.

[0029] Step 25: Determine the structural weights of the initial three-dimensional point cloud data under different projection views based on the structural point distribution characteristics, determine the number of two-dimensional nearest neighbors of each initial three-dimensional point cloud data under each projection view, and identify several two-dimensional nearest neighbors of each initial three-dimensional point cloud data in the three-dimensional lining structure.

[0030] Step 26: Locally fit the two-dimensional nearest neighbor points corresponding to the same projection viewpoint to obtain the local tangent plane corresponding to each of the initial three-dimensional point cloud data. Obtain the normal vector corresponding to each of the local tangent planes as the normal vector of the initial three-dimensional point cloud data. Perform second derivative on each of the local tangent planes and determine the curvature value of the initial three-dimensional point cloud data in combination with the corresponding normal vector direction.

[0031] In one feasible approach

[0032] Step 3 includes:

[0033] Step 31: Construct several standard identification conditions for the tunnel lining according to the tunnel lining standard, divide the three-dimensional lining structure into several lining planes, determine the planar continuity feature of the lining plane according to the normal vector direction contained in each lining plane, and identify the slope undulation feature of the lining plane according to the curvature value contained in each lining plane.

[0034] Step 32: Using the standard identification conditions, identify the non-standard sub-features contained in the planar continuous features and the slope undulation features, identify the feature regions corresponding to each non-standard sub-feature in the three-dimensional lining structure, and determine several abnormal regions of the tunnel lining;

[0035] Step 33: Identify the abnormality type corresponding to each abnormal region, set the corresponding scanning power for the abnormal region according to the abnormality type, and control the three-dimensional laser scanning device to perform fine scanning on each abnormal region with the corresponding scanning power.

[0036] Step 34: Obtain the refined scanning results corresponding to each of the abnormal regions, and perform data unification processing on the refined scanning results in combination with the corresponding scanning power to obtain the refined 3D point cloud data corresponding to each of the abnormal regions.

[0037] In one feasible approach

[0038] Also includes:

[0039] Each of the aforementioned abnormal regions is labeled;

[0040] A refined scan shortest path is constructed based on the straight-line distance between different abnormal regions, and a scan guidance path is generated by combining the corresponding labels and scan power.

[0041] In one feasible approach

[0042] Step 4 includes:

[0043] Step 41: In the three-dimensional lining structure, determine the structural key points contained in the original structure corresponding to each abnormal region, and determine the structural association between each structural key point and the three-dimensional lining structure.

[0044] Step 42: Construct a region structure corresponding to the abnormal region based on the fine 3D point cloud data, locate the corresponding structural key points in the region structure, and replace the corresponding original structure with the region structure in combination with the corresponding association method to generate a defect display structure;

[0045] Step 43: Identify several defect appearances contained in the defect display structure, use big data to identify the defect type and defect level corresponding to each defect appearance, and mark them for display in the three-dimensional lining structure.

[0046] The beneficial effects of the above technical solution are as follows: To improve the accuracy and quality of identifying internal defects in the tunnel lining and reduce external interference in the detection process, a three-dimensional laser scanning device is first used to perform an initial scan of the tunnel lining. Based on the collected geometric coordinate system information and reflection intensity information, initial three-dimensional point cloud data of the tunnel lining is constructed, providing high-quality data support for subsequent three-dimensional structure construction and defect detection, avoiding detection omissions due to insufficient data. Then, based on the initial three-dimensional point cloud data, the three-dimensional lining structure of the tunnel lining is drawn, and the normal vector and curvature value corresponding to each point cloud data are identified. Finally, by analyzing the normal vector and curvature value, it is determined whether there are defects on the lining surface. Anomaly detection accurately captures minute deformations, protrusions, or depressions that may exist on the lining surface. This allows for refined scanning of the anomaly area, utilizing the obtained detailed 3D point cloud data to analyze the regional structure of the anomaly and displaying it within the 3D lining structure. This visual presentation enables staff to intuitively and clearly see the location, shape, and size of defects on the lining, eliminating the need for tedious data interpretation and facilitating efficient decision-making and processing. Furthermore, secondary scanning of the anomaly area not only reveals the anomaly but also avoids the time-consuming process of repeated scanning, thus improving the efficiency and quality of tunnel lining inspection.

[0047] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0048] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0049] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0050] Figure 1 This is a schematic diagram of the workflow of the lining internal defect detection method based on three-dimensional point cloud in an embodiment of the present invention;

[0051] Figure 2 This is a schematic diagram of the workflow of step 3 of the method for detecting internal defects in lining based on three-dimensional point clouds in an embodiment of the present invention. Detailed Implementation

[0052] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0053] Example 1:

[0054] This embodiment provides a method for detecting internal defects in linings based on three-dimensional point clouds, such as... Figure 1 As shown, it includes:

[0055] Step 1: Use a 3D laser scanning device to scan the tunnel lining, and use the collected geometric coordinate information and reflection intensity information to construct the initial 3D point cloud data of the tunnel lining;

[0056] Step 2: Construct the three-dimensional lining structure of the tunnel lining based on the initial three-dimensional point cloud data, and identify the normal vector and curvature value corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure;

[0057] Step 3: Identify the abnormal areas of the tunnel lining based on the normal vector and the curvature value, and use the three-dimensional laser scanning device to perform a fine scan on each abnormal area to obtain the corresponding fine three-dimensional point cloud data;

[0058] Step 4: Construct a regional structure corresponding to the abnormal area using the refined 3D point cloud data, determine the defect appearance of the abnormal area based on the presentation characteristics of the regional structure, and mark it for display in the 3D lining structure.

[0059] In this example, the geometric coordinate information represents the three-dimensional geometry of the tunnel lining obtained by scanning the tunnel lining with a laser. The geometric coordinate information can accurately reflect the spatial morphology of the lining.

[0060] In this example, the reflection intensity information represents the intensity of the laser light reflected back after it hits the tunnel lining. The reflection intensity information can help determine the uniformity and other characteristics of the lining material.

[0061] In this example, the normal vector represents a vector perpendicular to the tunnel lining cross-section, and the normal vector can reflect the orientation change of the lining surface;

[0062] In this example, the curvature value represents the degree of bending of the tunnel lining;

[0063] In this example, the abnormal area represents an area where there are cavities, cracks, collapses, or leaks that affect the normal use of the tunnel;

[0064] In this example, fine scanning refers to the process of repeatedly scanning abnormal areas with high precision.

[0065] In this example, the region structure represents the structure presented by the abnormal region, and the presentation feature represents the appearance of the region structure.

[0066] The working principle and beneficial effects of the above technical solution are as follows: To improve the accuracy and quality of identifying internal defects in the tunnel lining and reduce external interference in the detection process, a three-dimensional laser scanning device is first used to perform an initial scan of the tunnel lining. Based on the collected geometric coordinate system information and reflection intensity information, initial three-dimensional point cloud data of the tunnel lining is constructed, providing high-quality data support for subsequent three-dimensional structure construction and defect detection, avoiding detection omissions due to insufficient data. Then, based on the initial three-dimensional point cloud data, the three-dimensional lining structure of the tunnel lining is drawn, and the normal vector and curvature value corresponding to each point cloud data are identified. Finally, by analyzing the normal vector and curvature value, it is determined whether there are defects on the lining surface. In anomalies, the system accurately captures minute deformations, protrusions, or depressions that may exist on the lining surface. It then performs a detailed scan of the anomaly area, using the obtained detailed 3D point cloud data to analyze the regional structure of the anomaly area and display it in the 3D lining structure. This visual presentation allows staff to intuitively and clearly see the location, shape, and size of defects on the lining, eliminating the need for tedious interpretation of complex data. This helps staff make efficient decisions and handle issues. Furthermore, by performing a secondary scan of the anomaly area, the system can obtain information about the anomaly while avoiding repeated scans that consume significant time, thus improving the efficiency and quality of tunnel lining inspection.

[0067] Example 2:

[0068] Based on Example 1, the method for detecting internal defects in linings based on three-dimensional point clouds, step 1 includes:

[0069] Step 11: Control the three-dimensional laser scanning device to perform laser scanning on the tunnel lining at different scanning angles, obtain the laser scanning results of the three-dimensional laser scanning device at each scanning angle, and obtain the scanning feedback results of the tunnel lining at different scanning angles;

[0070] Step 12: Pair the laser scanning results and scanning feedback results corresponding to the same scanning angle to obtain several scanning reflection points of the tunnel lining. Construct a three-dimensional mapping space based on the scanning intersection features between different scanning angles, and locate each scanning reflection point in the three-dimensional mapping space.

[0071] Step 13: Based on the positioning structure, map the laser scanning results and the scanning feedback results into the three-dimensional mapping space to generate the laser scanning process and obtain the geometric coordinate information and reflection intensity information of the tunnel lining;

[0072] Step 14: Merge the geometric coordinate information and the reflection intensity information, and generate the initial three-dimensional point cloud data of the tunnel lining based on the geometric coordinates and reflection intensity corresponding to each laser point.

[0073] In this example, the scanning perspective refers to the angle presented when a 3D laser scanning device scans from multiple angles to avoid blind spots in the scanning results.

[0074] In this example, the laser scanning result represents the result obtained when the three-dimensional laser scanning device scans at a scanning angle, and the scanning feedback structure represents the result when the tunnel lining reflects the laser.

[0075] In this example, the scan reflection point represents the point in the tunnel lining that reflects the laser beam;

[0076] In this example, the three-dimensional mapping space represents the three-dimensional space used for mapping operations.

[0077] The working principle and beneficial effects of the above technical solution are as follows: To ensure the integrity and comprehensiveness of the generated 3D point cloud data, a 3D laser scanning device first scans the tunnel lining from multiple angles, ensuring that the scanning process covers the entire tunnel lining and captures all surface information. Then, the obtained laser scanning results and scanning feedback results are paired to obtain several scanning reflection points on the tunnel lining. A 3D mapping space is constructed using the scanning intersection features from different perspectives. Each scanning reflection point is precisely located in this space, ensuring that the subsequently derived geometric coordinate information can accurately reflect the actual shape of the tunnel lining. This lays the foundation for the accuracy of the initial 3D point cloud data. Further... The laser scanning process was reconstructed by mapping the laser scanning results and feedback results onto a three-dimensional mapping space, obtaining the geometric coordinates and reflection intensity information of the tunnel lining. This fully utilized the information contained in the scanning data, ensuring that the obtained geometric coordinates and reflection intensity information comprehensively covered the surface features of the tunnel lining. Finally, the geometric coordinates and reflection intensity information were merged to generate the initial three-dimensional point cloud data of the tunnel lining. This initial three-dimensional point cloud data, which integrates multi-dimensional information, is of higher quality and provides reliable and comprehensive data support for subsequent construction of the three-dimensional lining structure, identification of normal vectors and curvature values, and detection of internal defects, thus helping to improve the accuracy and effectiveness of the entire detection process.

[0078] Example 3:

[0079] Based on Example 1, the method for detecting internal defects in linings based on three-dimensional point clouds further includes:

[0080] The three-dimensional laser scanning device is controlled to determine several scanning angles based on a preset angle difference;

[0081] The three-dimensional laser scanning device is controlled to perform laser scanning on the tunnel lining at each of the aforementioned scanning angles.

[0082] In this example, the preset viewing angle difference is a value set in advance by the staff, which is generally 3° between two adjacent viewing angles.

[0083] The working principle and beneficial effects of the above technical solution are as follows: scanning the tunnel lining from multiple perspectives can avoid blind spots that may exist in single-perspective scanning. In particular, for complex parts of the tunnel lining such as corners and depressions, it can fully capture their surface information and reduce the risk of missing defects due to missing data.

[0084] Example 4:

[0085] Based on Example 1, the method for detecting internal defects in tunnel lining based on three-dimensional point cloud data constructs a three-dimensional lining structure of the tunnel lining according to the initial three-dimensional point cloud data, including:

[0086] Step 21: Calculate the first data density of the initial three-dimensional point cloud data. When the first data density is higher than the specified density, convert each initial three-dimensional point cloud data into voxels and obtain the voxel center point, voxel average point and voxel dense point corresponding to each voxel. Convert each initial three-dimensional point cloud data into the corresponding center three-dimensional point cloud data, average point three-dimensional point cloud data and dense three-dimensional point cloud data to generate a candidate point cloud dataset.

[0087] Step 22: Randomly extract candidate point cloud data from each of the candidate point cloud datasets to construct several low-dimensional three-dimensional point cloud data for the tunnel lining, and select the target low-dimensional three-dimensional point cloud data with the second data density lower than the specified density and the highest similarity to the initial three-dimensional point cloud data.

[0088] Step 23: Identify the geometric appearance features corresponding to each of the target low-dimensional three-dimensional point cloud data, and identify the geometric overlap features between different target low-dimensional three-dimensional point cloud data, and construct the three-dimensional lining structure of the tunnel lining based on the geometric appearance features and the geometric overlap features.

[0089] In this example, the first data density represents the density of the initial 3D point cloud data;

[0090] In this example, the density is specified as between 5,000 and 10,000 points per cubic meter;

[0091] In this example, a voxel represents the result of converting the initial 3D point cloud data into basic elements in 3D space;

[0092] In this example, the voxel center point represents the geometric center of the voxel, the voxel average point represents the average position of all points within the voxel, and the voxel dense point represents the position with the largest number of points within the voxel.

[0093] In this example, the central 3D point cloud data represents the point cloud data generated based on the center point of the voxel, the average point 3D point cloud data represents the point cloud data generated based on the average voxel point, and the voxel dense point represents the point cloud data generated based on the voxel dense point.

[0094] In this example, the central 3D point cloud data, the average point 3D point cloud data, and the dense 3D point cloud data are all low-dimensional 3D point cloud data.

[0095] In this example, the geometric coincidence feature represents the characteristics that appear when different target low-dimensional 3D point cloud data overlap;

[0096] In this example, geometric appearance refers to the appearance of the target's low-dimensional 3D point cloud data.

[0097] The working principle and beneficial effects of the above technical solution are as follows: Since the actual tunnel lining may have a complex structure, in order to generate an accurate and effective tunnel lining, the initial three-dimensional point cloud data is first converted into different types of voxel point cloud data to generate candidate point cloud datasets. This simplifies the high-density point cloud data, reduces the amount of data, and avoids the waste of computing resources and low processing efficiency caused by data redundancy when constructing the three-dimensional structure. Then, the low-dimensional three-dimensional point cloud data that meets the target are selected for geometric appearance construction, and the geometric overlap features between different target low-dimensional three-dimensional point cloud data are combined to construct the three-dimensional lining structure of the tunnel lining. This makes the constructed three-dimensional lining structure not only conform to the actual geometric shape of the tunnel lining, but also has good integrity and coherence. In this way, it can be ensured that even when the lining has complex shapes such as bending and protrusion, an accurate three-dimensional structure can be constructed, which improves the adaptability of the method to complex engineering scenarios.

[0098] Example 5:

[0099] Based on Example 1, the method for detecting internal defects in linings based on three-dimensional point clouds identifies the normal vector and curvature value corresponding to each initial three-dimensional point cloud data in the three-dimensional lining structure, including:

[0100] Step 24: Identify the structural points corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure, and perform planar projection on the three-dimensional lining structure from a preset projection perspective to obtain the corresponding two-dimensional lining image, thereby obtaining the structural point distribution characteristics corresponding to each of the two-dimensional lining images.

[0101] Step 25: Determine the structural weights of the initial three-dimensional point cloud data under different projection views based on the structural point distribution characteristics, determine the number of two-dimensional nearest neighbors of each initial three-dimensional point cloud data under each projection view, and identify several two-dimensional nearest neighbors of each initial three-dimensional point cloud data in the three-dimensional lining structure.

[0102] Step 26: Locally fit the two-dimensional nearest neighbor points corresponding to the same projection viewpoint to obtain the local tangent plane corresponding to each of the initial three-dimensional point cloud data. Obtain the normal vector corresponding to each of the local tangent planes as the normal vector of the initial three-dimensional point cloud data. Perform second derivative on each of the local tangent planes and determine the curvature value of the initial three-dimensional point cloud data in combination with the corresponding normal vector direction.

[0103] In this example, an initial 3D point cloud data corresponds to a structural point, and the structural point represents a part of the 3D lining structure.

[0104] In this example, the preset projection angle is in the same direction as the x-axis, y-axis, and z-axis.

[0105] In this example, the structural weights represent the importance of the structure presented by the initial 3D point cloud data from a projection viewpoint.

[0106] The working principle and beneficial effects of the above technical solution are as follows: By projecting the three-dimensional lining structure into a two-dimensional lining image through a preset projection viewpoint, the distribution characteristics of structural points are obtained. The two-dimensional images under different projection viewpoints can present the distribution of structural points from multiple dimensions, avoiding the limitations of single-viewpoint analysis. Then, based on the distribution characteristics of structural points, the structural weights and the number of two-dimensional nearest neighbors are determined, thereby identifying the two-dimensional nearest neighbors corresponding to the initial three-dimensional point cloud data in the three-dimensional lining structure. This reduces the interference of outliers on the fitting results and ensures the reliability of local tangent plane fitting. Finally, by combining local fitting with differentiation, the normal vector and curvature value of each initial three-dimensional point cloud data are determined, which sharply captures the small anomalies on the lining surface and provides a reliable basis for subsequent accurate identification of abnormal areas.

[0107] Example 6:

[0108] Based on Example 1, the method for detecting internal defects in linings based on three-dimensional point clouds, such as... Figure 2 As shown, step 3 includes:

[0109] Step 31: Construct several standard identification conditions for the tunnel lining according to the tunnel lining standard, divide the three-dimensional lining structure into several lining planes, determine the planar continuity feature of the lining plane according to the normal vector direction contained in each lining plane, and identify the slope undulation feature of the lining plane according to the curvature value contained in each lining plane.

[0110] Step 32: Using the standard identification conditions, identify the non-standard sub-features contained in the planar continuous features and the slope undulation features, identify the feature regions corresponding to each non-standard sub-feature in the three-dimensional lining structure, and determine several abnormal regions of the tunnel lining;

[0111] Step 33: Identify the abnormality type corresponding to each abnormal region, set the corresponding scanning power for the abnormal region according to the abnormality type, and control the three-dimensional laser scanning device to perform fine scanning on each abnormal region with the corresponding scanning power.

[0112] Step 34: Obtain the refined scanning results corresponding to each of the abnormal regions, and perform data unification processing on the refined scanning results in combination with the corresponding scanning power to obtain the refined 3D point cloud data corresponding to each of the abnormal regions.

[0113] In this example, the planar continuity feature represents the planar continuity of the tunnel lining, while the slope bullying feature represents the undulation of the tunnel lining surface.

[0114] In this example, non-standard sub-features represent features that do not meet the standard recognition criteria, such as planar continuity features and slope undulation features.

[0115] In this example, the anomaly types include: voids, cracks, seepage, and leaks;

[0116] In this example, since different abnormal areas have different shapes, using different scanning frequencies can improve the accuracy of the scanning results.

[0117] In this example, data unification refers to the process of standardizing the numerical units of measurement for the detailed scan results of different anomaly areas.

[0118] The working principle and beneficial effects of the above technical solution are as follows: By dividing the three-dimensional lining structure into lining planes, determining the continuous features of the plane through the normal vector direction, identifying the slope undulation features by using curvature values, and then identifying them using standard identification conditions to determine the corresponding abnormal areas, this feature-based abnormal area identification method can accurately locate the problematic areas, avoiding indiscriminate inspection of the entire lining structure, reducing unnecessary workload, and improving inspection efficiency. Then, based on the abnormality type of each abnormal area, the corresponding scanning power is set for the three-dimensional laser scanning equipment to perform fine scanning of the abnormal areas. Targeted scanning power settings can ensure clear and accurate scanning results while avoiding data distortion or resource waste caused by improper power. Finally, the fine scanning results are processed to obtain corresponding fine three-dimensional point cloud data, making the fine three-dimensional point cloud data of different abnormal areas comparable and consistent. This provides a reliable data foundation for subsequent operations such as constructing regional structures and determining defect appearances, ensuring the effectiveness and consistency of data throughout the entire inspection process.

[0119] Example 7:

[0120] Based on Example 6, the method for detecting internal defects in linings based on three-dimensional point clouds further includes:

[0121] Each of the aforementioned abnormal regions is labeled;

[0122] A refined scan shortest path is constructed based on the straight-line distance between different abnormal regions, and a scan guidance path is generated by combining the corresponding labels and scan power.

[0123] The working principle and beneficial effects of the above technical solution are as follows: In order to avoid scanning omissions and improve scanning efficiency, the scanning path is planned in advance to guide the 3D laser scanning equipment to perform the corresponding scanning work.

[0124] Example 8:

[0125] Based on Example 1, the method for detecting internal defects in linings based on three-dimensional point clouds, step 4 includes:

[0126] Step 41: In the three-dimensional lining structure, determine the structural key points contained in the original structure corresponding to each abnormal region, and determine the structural association between each structural key point and the three-dimensional lining structure.

[0127] Step 42: Construct a region structure corresponding to the abnormal region based on the fine 3D point cloud data, locate the corresponding structural key points in the region structure, and replace the corresponding original structure with the region structure in combination with the corresponding association method to generate a defect display structure;

[0128] Step 43: Identify several defect appearances contained in the defect display structure, use big data to identify the defect type and defect level corresponding to each defect appearance, and mark them for display in the three-dimensional lining structure.

[0129] In this example, the original structure represents the structure of the abnormal region in the three-dimensional lining structure;

[0130] In this example, the structural association method represents the connection method between different structural key points;

[0131] In this example, structural key points refer to structural points in the original structure that connect with other structures.

[0132] The working principle and beneficial effects of the above technical solution are as follows: In order to reduce the disorder phenomenon in the three-dimensional lining structure, the replacement process of the regional structure is guided by analyzing the key points and connection methods of the original structure in the abnormal area. This generates a defect display structure, identifies the types and levels of defects contained in the structure, and displays them in a standard manner. From an overall perspective, staff can analyze the interaction between defects and the surrounding structure and assess the potential threat of defects to the stability of the entire lining structure.

[0133] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting internal defects in linings based on three-dimensional point clouds, characterized in that, include: Step 1: Use a 3D laser scanning device to scan the tunnel lining, and use the collected geometric coordinate information and reflection intensity information to construct the initial 3D point cloud data of the tunnel lining; Step 2: Construct the three-dimensional lining structure of the tunnel lining based on the initial three-dimensional point cloud data, and identify the normal vector and curvature value corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure; Step 3: Identify the abnormal areas of the tunnel lining based on the normal vector and the curvature value, and use the three-dimensional laser scanning device to perform a fine scan on each abnormal area to obtain the corresponding fine three-dimensional point cloud data; Step 4: Construct a regional structure corresponding to the abnormal area using the refined 3D point cloud data, determine the defect appearance of the abnormal area based on the presentation characteristics of the regional structure, and mark it for display in the 3D lining structure; The process of constructing the three-dimensional lining structure of the tunnel lining based on the initial three-dimensional point cloud data includes: Step 21: Calculate the first data density of the initial three-dimensional point cloud data. When the first data density is higher than the specified density, convert each initial three-dimensional point cloud data into voxels and obtain the voxel center point, voxel average point and voxel dense point corresponding to each voxel. Convert each initial three-dimensional point cloud data into the corresponding center three-dimensional point cloud data, average point three-dimensional point cloud data and dense three-dimensional point cloud data to generate a candidate point cloud dataset. Step 22: Randomly extract candidate point cloud data from each of the candidate point cloud datasets to construct several low-dimensional three-dimensional point cloud data for the tunnel lining, and select the target low-dimensional three-dimensional point cloud data with the second data density lower than the specified density and the highest similarity to the initial three-dimensional point cloud data. Step 23: Identify the geometric appearance features corresponding to each of the target low-dimensional three-dimensional point cloud data, and identify the geometric overlap features between different target low-dimensional three-dimensional point cloud data, and construct the three-dimensional lining structure of the tunnel lining based on the geometric appearance features and the geometric overlap features.

2. The three-dimensional point cloud-based lining internal defect detection method of claim 1, wherein, Step 1 includes: Step 11: Control the three-dimensional laser scanning device to perform laser scanning on the tunnel lining at different scanning angles, obtain the laser scanning results of the three-dimensional laser scanning device at each scanning angle, and obtain the scanning feedback results of the tunnel lining at different scanning angles; Step 12: Pair the laser scanning results and scanning feedback results corresponding to the same scanning angle to obtain several scanning reflection points of the tunnel lining. Construct a three-dimensional mapping space based on the scanning intersection features between different scanning angles, and locate each scanning reflection point in the three-dimensional mapping space. Step 13: Based on the positioning structure, map the laser scanning results and the scanning feedback results into the three-dimensional mapping space to generate the laser scanning process and obtain the geometric coordinate information and reflection intensity information of the tunnel lining; Step 14: Merge the geometric coordinate information and the reflection intensity information, and generate the initial three-dimensional point cloud data of the tunnel lining based on the geometric coordinates and reflection intensity corresponding to each laser point.

3. The three-dimensional point cloud-based lining internal defect detection method of claim 1, wherein, Also includes: The three-dimensional laser scanning device is controlled to determine several scanning angles based on a preset angle difference; The three-dimensional laser scanning device is controlled to perform laser scanning on the tunnel lining at each of the scanning angles.

4. The three-dimensional point cloud-based lining internal defect detection method of claim 1, wherein, Identifying the normal vector and curvature value corresponding to each initial three-dimensional point cloud data in the three-dimensional lining structure includes: Step 24: Identify the structural points corresponding to each of the initial three-dimensional point cloud data in the three-dimensional lining structure, and perform planar projection on the three-dimensional lining structure from a preset projection perspective to obtain the corresponding two-dimensional lining image, thereby obtaining the structural point distribution characteristics corresponding to each of the two-dimensional lining images. Step 25: Determine the structural weights of the initial three-dimensional point cloud data under different projection views based on the structural point distribution characteristics, determine the number of two-dimensional nearest neighbors of each initial three-dimensional point cloud data under each projection view, and identify several two-dimensional nearest neighbors of each initial three-dimensional point cloud data in the three-dimensional lining structure. Step 26: Locally fit the two-dimensional nearest neighbor points corresponding to the same projection viewpoint to obtain the local tangent plane corresponding to each of the initial three-dimensional point cloud data. Obtain the normal vector corresponding to each of the local tangent planes as the normal vector of the initial three-dimensional point cloud data. Perform second derivative on each of the local tangent planes and determine the curvature value of the initial three-dimensional point cloud data in combination with the corresponding normal vector direction. 5.The three-dimensional point cloud-based lining internal defect detection method of claim 1, wherein, Step 3 includes: Step 31: Construct several standard identification conditions for the tunnel lining according to the tunnel lining standard, divide the three-dimensional lining structure into several lining planes, determine the planar continuity feature of the lining plane according to the normal vector direction contained in each lining plane, and identify the slope undulation feature of the lining plane according to the curvature value contained in each lining plane. Step 32: Using the standard identification conditions, identify the non-standard sub-features contained in the planar continuous features and the slope undulation features, identify the feature regions corresponding to each non-standard sub-feature in the three-dimensional lining structure, and determine several abnormal regions of the tunnel lining; Step 33: Identify the abnormal type corresponding to each abnormal region, set the corresponding scanning power for the abnormal region according to the abnormal type, and control the three-dimensional laser scanning device to perform fine scanning on each abnormal region with the corresponding scanning power. Step 34: Obtain the refined scanning results corresponding to each of the abnormal regions, and perform data unification processing on the refined scanning results in combination with the corresponding scanning power to obtain the refined 3D point cloud data corresponding to each of the abnormal regions. 6.The three-dimensional point cloud-based lining internal defect detection method of claim 5, wherein, Also includes: Each of the aforementioned abnormal regions is labeled; A refined scanning shortest path is constructed based on the straight-line distance between different abnormal regions, and a scanning guidance path is generated by combining the corresponding labels and scanning power.

7. The method for detecting internal defects in linings based on three-dimensional point clouds as described in claim 1, characterized in that, Step 4 includes: Step 41: In the three-dimensional lining structure, determine the structural key points contained in the original structure corresponding to each abnormal region, and determine the structural association between each structural key point and the three-dimensional lining structure. Step 42: Construct a region structure corresponding to the abnormal region based on the fine 3D point cloud data, locate the corresponding structural key points in the region structure, and replace the corresponding original structure with the region structure in combination with the corresponding association method to generate a defect display structure; Step 43: Identify several defect appearances contained in the defect display structure, use big data to identify the defect type and defect level corresponding to each defect appearance, and mark them for display in the three-dimensional lining structure.

Images