Method for detecting uneven thickness of tunnel fault zone lining by fusing three-dimensional laser point cloud and ls

Abstract

The present application belongs to the technical field of tunnel engineering detection, and provides a tunnel fault zone lining uneven thickness detection method based on three-dimensional laser point cloud fusion LS, which comprises the following steps: collecting tunnel fault zone point cloud data through a three-dimensional laser scanner; processing the tunnel fault zone point cloud data based on three-dimensional LS (least squares method) to construct a geometric model of the tunnel fault zone; and detecting the tunnel fault zone lining uneven thickness based on the geometric model of the tunnel fault zone and in combination with actual point cloud data. The present application can realize accurate measurement of the tunnel fault zone lining uneven thickness, improve the accuracy and efficiency of detection, has the advantages of high automation degree and fast detection efficiency, and provides strong technical support for safety monitoring and maintenance of tunnel engineering.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering inspection technology, and in particular to a method for detecting uneven thickness of tunnel fault zone lining using three-dimensional laser point cloud fusion (LS). Background Technology

[0002] With the rapid development of tunnel engineering, the requirements for the safety and stability of tunnel structures are increasing. The uneven thickness of the tunnel fault zone lining is one of the important factors affecting the safety of the tunnel structure. Traditional detection methods often rely on manual measurement, which is not only time-consuming and labor-intensive, but also difficult to guarantee the accuracy and comprehensiveness of the measurement. Therefore, it is particularly important to develop an efficient, accurate and comprehensive detection method for the uneven thickness of the tunnel fault zone lining.

[0003] Three-dimensional laser scanning technology, as a non-contact measurement method, boasts advantages such as high precision, high efficiency, and comprehensiveness, and has been widely used in tunnel engineering inspection. However, single three-dimensional laser scanning data is often affected by factors such as noise and obstruction, leading to incomplete data or errors. Therefore, how to combine three-dimensional laser point cloud data with other technologies to improve the accuracy and reliability of detecting uneven thickness of tunnel fault zone lining has become a current research hotspot and challenge.

[0004] To more effectively leverage the large volume of 3D laser scanning data, a certain step size is typically set during cross-section segmentation to extract point cloud data from the tunnel fault zone. Each cross-section is then projected onto a 2D plane for fitting analysis. This process requires transforming the tunnel point cloud coordinate system so that the central axis of the tunnel point cloud data coincides with the Z-axis. Methods for calculating the tunnel central axis direction vector include the minimum bounding region method, Least Squares (LS), and the two-endpoint method. Among these three methods, the minimum bounding region method is the most accurate, but its solution process is complex and time-consuming. In general engineering applications, the two-endpoint method and LS are typically used. The two-endpoint method has poor robustness, while the conventional LS method, when fitting a straight line in 3D space, requires independently fitting lines on two separate planes before synthesizing a single line. Therefore, it is not a true 3D straight line fitting. When the coordinate values ​​of the measured points are of different orders of magnitude, the evaluation results of the least squares algorithm cannot meet the requirements of practical engineering projects with high accuracy requirements.

[0005] In point cloud data preprocessing, only relatively obvious outliers can be removed. Noise points attached to the inner diameter of the pipe wall are difficult to detect and remove with the naked eye. The Ransac algorithm removes outliers and retains inliers for fitting calculation by setting thresholds and iterative calculations, thereby improving accuracy. According to the law of large numbers, random simulation can approximately obtain the optimal model, which is a very robust fitting algorithm.

[0006] Therefore, it is necessary to provide a method for detecting uneven thickness of tunnel fault lining by three-dimensional laser point cloud fusion (LS). Summary of the Invention

[0007] This invention provides a method for detecting uneven thickness of tunnel fault zone lining based on three-dimensional laser point cloud fusion (LS). It can accurately measure the uneven thickness of tunnel fault zone lining, improve the accuracy and efficiency of detection, and has the advantages of high automation and fast detection efficiency, providing strong technical support for the safety monitoring and maintenance of tunnel engineering.

[0008] This invention provides a method for detecting uneven thickness of tunnel fault lining by three-dimensional laser point cloud fusion (LS), including:

[0009] Data on point cloud patterns in tunnel fault zones were acquired using a 3D laser scanner.

[0010] Based on the three-dimensional LS (least squares) method, the point cloud data of the tunnel fault zone is processed to construct a geometric model of the tunnel fault zone.

[0011] Based on the geometric morphology model of the tunnel fault zone and combined with actual point cloud data, the uneven thickness of the tunnel fault zone lining is detected.

[0012] Furthermore, point cloud data of the tunnel fault zone was acquired using a 3D laser scanner, including:

[0013] Configure a 3D laser scanner;

[0014] Cloud data of tunnel fault zones were acquired using a 3D laser scanner.

[0015] Furthermore, based on 3D LS (least squares method), the point cloud data of the tunnel fault zone is processed to construct a geometric morphological model of the tunnel fault zone, including:

[0016] The tunnel fault zone point cloud data is filtered and denoised to obtain preprocessed tunnel fault zone point cloud data.

[0017] By fitting point cloud data using LS, a geometric model of the tunnel fault zone is constructed.

[0018] Furthermore, the tunnel fault zone point cloud data is filtered and denoised to obtain preprocessed tunnel fault zone point cloud data, including:

[0019] The RANSAC algorithm was used to remove noise points from the point cloud data of the tunnel fault zone to obtain the first preprocessed data;

[0020] The first preprocessed data is classified using a dynamic clustering algorithm to identify and remove point cloud data of non-lining structures in the tunnel fault zone, thereby obtaining the second preprocessed data.

[0021] The second preprocessed data is smoothed to obtain preprocessed tunnel fault zone point cloud data.

[0022] Furthermore, by fitting the point cloud data using LS, a geometric morphological model of the tunnel fault zone is constructed, including:

[0023] The preprocessed tunnel fault zone point cloud data is subjected to coordinate transformation so that the direction of the tunnel centerline coincides with the Z-axis direction in the pre-defined global coordinate system; the direction vector of the tunnel centerline is obtained by using LS to establish a fitting mathematical model and solve it.

[0024] Multiple cross sections are cut along the central axis based on a set step size, and the point cloud of each cross section is projected onto a two-dimensional plane to obtain two-dimensional tunnel fault zone point cloud data.

[0025] Based on two-dimensional tunnel fault zone point cloud data and combined with tunnel design parameters, a geometric morphological model of the tunnel fault zone is constructed.

[0026] Furthermore, coordinate transformation is performed on the preprocessed tunnel fault zone point cloud data, including:

[0027] The central axis of the tunnel is determined using the method of connecting the two endpoints.

[0028] Establish a local coordinate system and adjust the central axis to align with the Z-axis;

[0029] By using rotation transformation, the preprocessed tunnel fault zone point cloud data is mapped to the global coordinate system.

[0030] Furthermore, a fitting mathematical model is established using LS, including:

[0031] The dataset consisting of preprocessed tunnel fault zone point cloud data is set as follows: ,

[0032] The equation of the fitted line is:

[0033] ,

[0034] The objective function is to minimize the sum of squared distances from each point to the line:

[0035] ,

[0036] The parameters are solved iteratively using matrix operations or numerical optimization tools. , , (a, b, c), which minimizes the objective function value; where n represents the number of points in the point cloud dataset, ( , , ) represents the coordinates of the reference point on the straight line, indicating the geometric center of the tunnel's central axis, used to locate the spatial reference of the fitted straight line; a, b, and c represent the direction vector components of the straight line, indicating the extension of the straight line in the X, Y, and Z directions in three-dimensional space.

[0037] Furthermore, based on the geometric morphology model of the tunnel fault zone and combined with actual point cloud data, the uneven thickness of the tunnel fault zone lining is detected, including:

[0038] Obtain actual point cloud data;

[0039] By comparing and analyzing geometric morphology models with actual point cloud data, the uneven thickness of tunnel fault zone lining can be detected.

[0040] Furthermore, based on comparative analysis of the geometric morphology model and actual point cloud data, the non-uniform thickness of the tunnel fault zone lining is detected, including:

[0041] By comparing the actual point cloud data with the geometric morphology model in terms of spatial location, the deviation areas of the lining surface are identified.

[0042] Calculate the thickness of the deviation area, compare the thickness with the design thickness, and determine the uneven thickness area of ​​the lining.

[0043] Furthermore, based on comparative analysis of geometric morphological models and actual point cloud data, the detection of uneven thickness in tunnel fault zone lining is achieved, which also includes:

[0044] A dynamic threshold range is set, and a thickness non-uniformity judgment standard is generated based on the dynamic threshold range; the steps for setting the dynamic threshold range are as follows:

[0045] Based on the tunnel geological classification map and historical construction and maintenance records, the historical threshold range is obtained, and a benchmark area threshold library is set up based on the historical threshold range.

[0046] Local features are extracted from the acquired point cloud data of the tunnel fault zone to obtain a local feature dataset;

[0047] Based on a convolutional neural network model, the optimal threshold for the tunnel fault zone is predicted according to the local feature dataset.

[0048] Based on the optimal threshold, the threshold library for the benchmark region is dynamically adjusted to generate a dynamic threshold range.

[0049] When the calculated thickness deviation exceeds the dynamic threshold range, the deviation area of ​​the lining surface is determined to be an uneven thickness area.

[0050] Based on the tunnel geological classification map, historical construction and maintenance records, and thickness deviation data, a Bayesian network model is used to predict the probability of causes leading to thickness deviation.

[0051] Based on the causal probability, and using a database of matching relationships between the set causal probabilities and repair cases, repair cases are matched to obtain the optimal repair solution matching result.

[0052] Compared with the prior art, the present invention has the following advantages and beneficial effects: it can accurately measure the uneven thickness of the lining in the fault zone of the tunnel, improve the accuracy and efficiency of the detection, and has the advantages of high automation and fast detection efficiency, providing strong technical support for the safety monitoring and maintenance of tunnel engineering.

[0053] 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.

[0054] 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

[0055] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0056] Figure 1 A schematic diagram of the steps for detecting uneven thickness of tunnel fault zone lining by 3D laser point cloud fusion (LS).

[0057] Figure 2 A schematic diagram illustrating the steps of acquiring point cloud data of tunnel fault zones using a 3D laser scanner;

[0058] Figure 3 This diagram illustrates the steps involved in processing point cloud data of a tunnel fault zone based on 3D LS (least squares) to construct a geometric model of the tunnel fault zone. Detailed Implementation

[0059] 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.

[0060] This invention provides a method for detecting uneven thickness of tunnel fault lining by three-dimensional laser point cloud fusion (LS), such as... Figure 1 As shown, it includes:

[0061] Data on point cloud patterns in tunnel fault zones were acquired using a 3D laser scanner.

[0062] Based on the three-dimensional LS (least squares) method, the point cloud data of the tunnel fault zone is processed to construct a geometric model of the tunnel fault zone.

[0063] Based on the geometric morphology model of the tunnel fault zone and combined with actual point cloud data, the uneven thickness of the tunnel fault zone lining is detected.

[0064] The working principle of the above technical solution is as follows: To realize the method for detecting uneven thickness of tunnel fault zone lining using 3D laser point cloud fusion (LS), this invention first uses a high-precision 3D laser scanner to perform a comprehensive scan of the tunnel fault zone to obtain detailed point cloud data. These data points accurately record the surface morphology and spatial location of the tunnel fault zone, providing a solid foundation for subsequent processing and analysis. Next, 3D LS (least squares method) is used to process the acquired point cloud data. Least squares is a mathematical optimization technique that finds the best function match for the data by minimizing the sum of squared errors. In this step, the LS algorithm is used to fit and smooth the point cloud data, thereby constructing a geometric model of the tunnel fault zone. This model can intuitively display the spatial structure and morphological characteristics of the tunnel fault zone, providing an important reference for subsequent detection work. Finally, based on the constructed geometric model of the tunnel fault zone and combined with actual point cloud data, the uneven thickness of the tunnel fault zone lining is detected. By comparing and analyzing the geometric model with the actual point cloud data, the variation area and degree of lining thickness can be accurately identified, providing strong support for tunnel maintenance and management.

[0065] The beneficial effects of the above technical solution are as follows: The solution provided in this embodiment can significantly improve the accuracy and efficiency of detecting uneven thickness of tunnel fault zone lining; detailed point cloud data is obtained through a high-precision 3D laser scanner, ensuring the comprehensiveness and accuracy of the data; the point cloud data is processed using a 3D LS algorithm to construct a geometric model of the tunnel fault zone, intuitively demonstrating the spatial structure and morphological characteristics of the tunnel fault zone; and by comparing and analyzing the geometric model with the actual point cloud data, the variation areas and degrees of lining thickness can be accurately identified, providing a scientific basis for tunnel maintenance and management.

[0066] In one embodiment, such as Figure 2 As shown, point cloud data of the tunnel fault zone was acquired using a 3D laser scanner, including:

[0067] Configure a 3D laser scanner;

[0068] Cloud data of tunnel fault zones were acquired using a 3D laser scanner.

[0069] The working principle of the above technical solution is as follows: To acquire point cloud data of tunnel fault zones using a 3D laser scanner, this invention first selects a suitable 3D laser scanner based on the actual conditions of the tunnel and the detection requirements, and configures it accordingly, including setting parameters such as scanning speed, scanning angle, and resolution, to ensure that the acquired point cloud data has sufficient accuracy and coverage. The working principle of the 3D laser scanner is based on the principle of laser ranging. By emitting a laser beam and receiving the reflected laser signal, the distance between the laser beam and the tunnel surface is calculated, thereby obtaining the 3D coordinate information of the tunnel surface. During the scanning process, the 3D laser scanner scans at a certain speed and angle, forming a series of point cloud data. Then, the 3D laser scanner is activated to scan the tunnel fault zone. Through the interaction between the laser beam and the tunnel surface, the 3D coordinate information of the tunnel fault zone surface is obtained, forming point cloud data. This point cloud data contains information such as the geometric shape and surface features of the tunnel fault zone, providing basic data for subsequent processing and analysis.

[0070] The beneficial effects of the above technical solution are as follows: the solution provided in this embodiment can achieve accurate detection of uneven thickness of tunnel fault zone lining; the point cloud data collected by the three-dimensional laser scanner has the characteristics of high precision and wide coverage, which can truly reflect the geometric shape and surface features of the tunnel fault zone, avoiding the measurement error and data loss problems that may exist in traditional detection methods; the point cloud data has three-dimensionality, which can intuitively display the spatial distribution and morphological characteristics of the tunnel fault zone, and help to conduct more in-depth analysis and research on the tunnel fault zone.

[0071] In one embodiment, such as Figure 3 As shown, based on 3D LS (least squares) method, point cloud data of the tunnel fault zone is processed to construct a geometric model of the tunnel fault zone, including:

[0072] The tunnel fault zone point cloud data is filtered and denoised to obtain preprocessed tunnel fault zone point cloud data.

[0073] By fitting point cloud data using LS, a geometric model of the tunnel fault zone is constructed.

[0074] The working principle of the above technical solution is as follows: By filtering and denoising, noise points in point cloud data can be effectively removed, improving the accuracy and reliability of the data. This step is the foundation for building a geometric morphology model and ensuring the quality of point cloud data in subsequent processing. Next, the least squares method is used to fit the preprocessed point cloud data, which is a key step in building the geometric morphology model. The least squares method finds the best function match for the data by minimizing the sum of squared errors, thereby accurately describing the geometric morphology of the tunnel fault zone. In this process, the algorithm automatically identifies the distribution characteristics of the point cloud data and constructs a geometric morphology model that is highly consistent with the actual situation.

[0075] The beneficial effects of the above technical solution are as follows: By adopting the solution provided in this embodiment, based on the high-precision point cloud data collected by the three-dimensional laser scanner, and through scientific algorithm processing, an accurate description of the geometric morphology of the tunnel fault zone can be achieved, providing reliable basic data for subsequent analysis and research.

[0076] In one embodiment, filtering and denoising the point cloud data of the tunnel fault zone to obtain preprocessed point cloud data of the tunnel fault zone includes:

[0077] The RANSAC algorithm was used to remove noise points from the point cloud data of the tunnel fault zone to obtain the first preprocessed data;

[0078] The first preprocessed data is classified using a dynamic clustering algorithm to identify and remove point cloud data of non-lining structures in the tunnel fault zone, thereby obtaining the second preprocessed data.

[0079] The second preprocessed data is smoothed to obtain preprocessed tunnel fault zone point cloud data.

[0080] The working principle of the above technical solution is as follows: The RANSAC algorithm is an iterative method used to estimate the parameters of a mathematical model from a set of data containing a large amount of noise. In the point cloud data of the tunnel fault zone, noise points may be caused by factors such as scanning errors and environmental interference. The RANSAC algorithm can effectively identify and remove these noise points, improving the purity of the data. The dynamic clustering algorithm automatically divides the data into different categories according to the spatial distribution characteristics of the point cloud data. Among them, the point cloud data of non-lining structure and the point cloud data of lining structure differ in geometric shape and distribution. The dynamic clustering algorithm can accurately identify and remove the point cloud data of non-lining structure, ensuring the accuracy of subsequent analysis. The smoothing process further reduces the small fluctuations and errors in the data, making the preprocessed point cloud data of the tunnel fault zone smoother and more continuous, providing a high-quality data foundation for subsequent construction of geometric morphology models and non-uniform thickness detection.

[0081] The beneficial effects of the above technical solution are as follows: the solution provided in this embodiment can significantly improve the accuracy and efficiency of detecting uneven thickness of tunnel fault lining; the RANSAC algorithm effectively removes noise points, the dynamic clustering algorithm accurately identifies and removes non-lining structure point cloud data, and the smoothing process reduces small fluctuations and errors in the data; the high-quality preprocessed data provides a solid foundation for the subsequent construction of geometric morphology models and the detection of uneven thickness.

[0082] In one embodiment, LS is used to fit point cloud data to construct a geometric morphological model of the tunnel fault zone, including:

[0083] The preprocessed tunnel fault zone point cloud data is subjected to coordinate transformation so that the direction of the tunnel centerline coincides with the Z-axis direction in the pre-defined global coordinate system; the direction vector of the tunnel centerline is obtained by using LS to establish a fitting mathematical model and solve it.

[0084] Multiple cross sections are cut along the central axis based on a set step size, and the point cloud of each cross section is projected onto a two-dimensional plane to obtain two-dimensional tunnel fault zone point cloud data.

[0085] Based on two-dimensional tunnel fault zone point cloud data and combined with tunnel design parameters, a geometric morphological model of the tunnel fault zone is constructed.

[0086] The working principle of the above technical solution is as follows: LS (Least Squares) is a mathematical optimization technique that finds the best function match for data by minimizing the sum of squares of errors. In this invention, LS not only helps to accurately determine the direction of the tunnel's central axis and ensure the accuracy of coordinate transformation, but also provides a solid mathematical foundation for subsequent steps. After coordinate transformation, the central axis of the point cloud data is aligned with the Z-axis of the global coordinate system, greatly simplifying subsequent processing steps. Taking multiple sections along the central axis with a set step size is a key step in constructing the geometric morphology model. This step ensures that the point cloud data of each key part of the tunnel is fully considered without omission. Projecting the point cloud of each section onto a two-dimensional plane not only reduces the complexity of data processing but also helps to more intuitively analyze and understand the distribution characteristics of the point cloud data. Combining the tunnel's design parameters, constructing a geometric morphology model based on the two-dimensional tunnel fault zone point cloud data is a prerequisite for detecting uneven thickness. This step makes full use of the tunnel's known information and improves the accuracy and reliability of the model.

[0087] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, a solid foundation is laid for subsequent non-uniform thickness detection through preprocessing and model building steps.

[0088] In one embodiment, performing coordinate transformation on preprocessed tunnel fault zone point cloud data includes:

[0089] The central axis of the tunnel is determined using the method of connecting the two endpoints.

[0090] Establish a local coordinate system and adjust the central axis to align with the Z-axis;

[0091] By using rotation transformation, the preprocessed tunnel fault zone point cloud data is mapped to the global coordinate system.

[0092] The working principle of the above technical solution is as follows: First, using the two-endpoint connection method, the two ends of the tunnel are selected as reference points. By connecting these two points, the preliminary central axis of the tunnel is determined. This method is simple and intuitive and can roughly depict the direction of the tunnel. Subsequently, in order to more accurately describe the geometry of the tunnel, a local coordinate system is established, and the initially determined central axis is adjusted to be aligned with the Z-axis of the local coordinate system through transformation operations such as translation and rotation. This step ensures that the direction of the central axis is consistent with the global coordinate system used for subsequent analysis, which facilitates subsequent data processing and analysis. After the local coordinate system is adjusted, the point cloud data is mapped from the local coordinate system to the global coordinate system through rotation transformation. The parameters of the rotation transformation are determined by calculating the rotation matrix between the local coordinate system and the global coordinate system, ensuring that the position and orientation of the point cloud data in the global coordinate system are accurate.

[0093] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, the preprocessed tunnel fault zone point cloud data is uniformly and standardized in the global coordinate system, providing a reliable data foundation for subsequent construction of geometric morphology models and non-uniform thickness detection.

[0094] In one embodiment, LS is used to establish a fitting mathematical model, including:

[0095] The dataset consisting of preprocessed tunnel fault zone point cloud data is set as follows: ,

[0096] The equation of the fitted line is:

[0097] ,

[0098] The objective function is to minimize the sum of squared distances from each point to the line:

[0099] ,

[0100] The parameters are solved iteratively using matrix operations or numerical optimization tools. , , (a, b, c), which minimizes the objective function value; where n represents the number of points in the point cloud dataset, ( , , ) represents the coordinates of the reference point on the straight line, indicating the geometric center of the tunnel's central axis, used to locate the spatial reference of the fitted straight line; a, b, and c represent the direction vector components of the straight line, indicating the extension of the straight line in the X, Y, and Z directions in three-dimensional space.

[0101] The working principle of the above technical solution is as follows: using the LS algorithm, the tunnel point cloud dataset, i.e., a series of three-dimensional coordinate points, is utilized. To fit a straight line that best represents the tunnel's central axis, the equation of this straight line is given by the parameters ( , , (a, b, c) are determined, where n represents the number of points in the point cloud dataset. , , The reference point, a specific point on the straight line, represents the geometric center of the tunnel's central axis, providing a benchmark for the spatial positioning of the line. Parameters a, b, and c constitute the direction vector of the line, reflecting its extension trend in the X, Y, and Z directions, respectively. In the solution process, a strategy of minimizing the objective function is adopted, namely, minimizing the sum of squared distances from all points to the fitted line. This objective function reflects the requirement for the accuracy of the fitted line. By iteratively adjusting the parameter values, the objective function value gradually decreases until it reaches its minimum, thus obtaining the optimal fitted line equation. By comparing the difference between the actual lining thickness and the ideal thickness represented by the fitted line, uneven areas of the lining can be effectively identified, allowing for corresponding repair measures to ensure the safety and stability of the tunnel structure.

[0102] The beneficial effects of the above technical solution are as follows: the solution provided in this embodiment can accurately detect the uneven thickness of the tunnel fault lining. The straight line of the tunnel centerline fitted by the LS algorithm provides an accurate reference benchmark for the measurement of the lining thickness. The strategy of minimizing the objective function ensures the accuracy of the fitted straight line, thereby improving the accuracy of the lining thickness measurement. By comparing the difference between the actual lining thickness and the ideal thickness, the uneven area of ​​the lining can be detected in a timely manner, providing strong support for the maintenance and repair of the tunnel.

[0103] In one embodiment, based on the geometric morphology model of the tunnel fault zone and combined with actual point cloud data, the non-uniform thickness of the tunnel fault zone lining is detected, including:

[0104] Obtain actual point cloud data;

[0105] By comparing and analyzing geometric morphology models with actual point cloud data, the uneven thickness of tunnel fault zone lining can be detected.

[0106] The working principle of the above technical solution is as follows: In order to realize the detection of uneven thickness of tunnel fault zone lining based on the geometric morphology model of the tunnel fault zone and combined with actual point cloud data, this invention first uses a high-precision three-dimensional laser scanner to scan the inside of the tunnel to obtain actual point cloud data inside the tunnel. This point cloud data contains three-dimensional coordinate information of various locations inside the tunnel, which can truly reflect the geometric morphology of the tunnel. Then, the obtained actual point cloud data is compared and analyzed with the pre-established geometric morphology model of the tunnel fault zone. The geometric morphology model is established based on the tunnel design drawings and known tunnel geometric features, and it describes the ideal geometric morphology of the tunnel fault zone. Through comparative analysis, the differences between the actual point cloud data and the geometric morphology model can be found. These differences reflect the uneven thickness of the tunnel fault zone lining. Specifically, this uneven thickness can be quantified by calculating the distance or deviation between the actual point cloud data and the geometric morphology model at various locations. If the distance or deviation at a certain location exceeds a preset threshold, the lining thickness at that location is considered uneven, which may pose a safety hazard.

[0107] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, combined with the high precision of three-dimensional laser scanning technology and the prior knowledge of geometric morphology models, it is possible to accurately detect the uneven thickness of the lining in tunnel fault zones.

[0108] In one embodiment, the detection of uneven thickness of tunnel fault zone lining is achieved by comparing and analyzing geometric morphological models with actual point cloud data, including:

[0109] By comparing the actual point cloud data with the geometric morphology model in terms of spatial location, the deviation areas of the lining surface are identified.

[0110] Calculate the thickness of the deviation area, compare the thickness with the design thickness, and determine the uneven thickness area of ​​the lining.

[0111] The working principle of the above technical solution is as follows: the geometric morphology model, as the ideal morphology of the tunnel fault zone, provides the theoretical thickness and shape that the lining should achieve, while the actual point cloud data reflects the real situation inside the tunnel, including the actual position and shape of the lining. By comparing the two one by one in spatial position, it is possible to accurately identify which areas of the lining surface deviate from the ideal shape, i.e., deviation areas. These deviation areas mean that the thickness of the lining is uneven at these locations, or there are other forms of geometric defects. After identifying the deviation areas, the thickness of these areas is further calculated. This usually involves processing the point cloud data, extracting the specific shape of the lining surface, and calculating the thickness value of each point accordingly. Then, these thickness values ​​are compared with the design thickness. The design thickness is determined according to the tunnel design requirements and engineering specifications, representing the ideal thickness that the lining should achieve. By comparing the actual thickness with the design thickness, it is possible to determine which areas of the lining thickness do not meet the requirements, i.e., uneven thickness areas.

[0112] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, it is possible to accurately detect the uneven thickness of the lining in the tunnel fault zone, which not only improves the accuracy and efficiency of the detection, but also helps to discover potential tunnel safety hazards in a timely manner.

[0113] In one embodiment, the detection of uneven thickness of tunnel fault zone lining is achieved by comparing and analyzing geometric morphological models with actual point cloud data, and further includes:

[0114] A dynamic threshold range is set, and a thickness non-uniformity judgment standard is generated based on the dynamic threshold range; the steps for setting the dynamic threshold range are as follows:

[0115] Based on the tunnel geological classification map and historical construction and maintenance records, the historical threshold range is obtained, and a benchmark area threshold library is set up based on the historical threshold range.

[0116] Local features are extracted from the acquired point cloud data of the tunnel fault zone to obtain a local feature dataset;

[0117] Based on a convolutional neural network model, the optimal threshold for the tunnel fault zone is predicted according to the local feature dataset.

[0118] Based on the optimal threshold, the threshold library for the benchmark region is dynamically adjusted to generate a dynamic threshold range.

[0119] When the calculated thickness deviation exceeds the dynamic threshold range, the deviation area of ​​the lining surface is determined to be an uneven thickness area.

[0120] Based on the tunnel geological classification map, historical construction and maintenance records, and thickness deviation data, a Bayesian network model is used to predict the probability of causes leading to thickness deviation.

[0121] Based on the causal probability, and using a database of matching relationships between the set causal probabilities and repair cases, repair cases are matched to obtain the optimal repair solution matching result.

[0122] The working principle of the above technical solution is as follows: In order to detect and determine the uneven thickness of tunnel fault zone lining according to a preset uneven thickness judgment standard, this invention considers the information provided by the tunnel geological classification map and historical construction and maintenance records when setting the dynamic threshold range. Furthermore, through prediction using local feature datasets and convolutional neural network models, a more accurate and optimal threshold adapted to the current tunnel conditions is obtained. This step ensures that the uneven thickness judgment standard can be flexibly adjusted according to the actual situation, improving the accuracy and practicality of detection. When a thickness deviation exceeding the dynamic threshold range is detected, the uneven thickness area on the lining surface is automatically marked. Further, to improve the pertinence and effectiveness of the repair plan, the recommended repair plan is refined based on the specific uneven thickness distribution of the tunnel fault zone lining, combined with the tunnel's geological characteristics and environmental factors. Additionally, based on feedback data from the historical maintenance case library, the predictive ability of the Bayesian network model is continuously optimized to improve the accuracy of predicting the probability of causes, thereby providing more reliable technical support for subsequent repair work.

[0123] The beneficial effects of the above technical solution are as follows: by adopting the solution provided in this embodiment, not only can the uneven thickness of the lining in the tunnel fault zone be accurately detected and effectively repaired, but the cost and risk of tunnel maintenance can also be reduced, providing a strong guarantee for the safe operation of the tunnel.

[0124] 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 uneven thickness of tunnel fault lining by 3D laser point cloud fusion (LS), characterized in that, include: Data on point cloud patterns in tunnel fault zones were acquired using a 3D laser scanner. The RANSAC algorithm was used to remove noise points from the point cloud data of the tunnel fault zone to obtain the first preprocessed data; The first preprocessed data is classified using a dynamic clustering algorithm to identify and remove point cloud data of non-lining structures in the tunnel fault zone, thereby obtaining the second preprocessed data. The second preprocessed data is smoothed to obtain preprocessed point cloud data of the tunnel fault zone. The preprocessed tunnel fault zone point cloud data is subjected to coordinate transformation so that the direction of the tunnel's central axis coincides with the Z-axis direction in a pre-defined global coordinate system. The direction vector of the tunnel's central axis is obtained by solving a fitting mathematical model using LS. Multiple cross-sections are extracted along the central axis based on a set step size, and the point clouds of each cross-section are projected onto a two-dimensional plane to obtain two-dimensional tunnel fault zone point cloud data. Based on the two-dimensional tunnel fault zone point cloud data and combined with the tunnel's design parameters, a geometric morphological model of the tunnel fault zone is constructed. The preprocessed tunnel fault zone point cloud data is compared with the geometric morphology model to identify the deviation area of ​​the lining surface; the thickness of the deviation area is calculated and compared with the design thickness to determine the uneven thickness area of ​​the lining. It also includes: setting a dynamic threshold range and generating a thickness non-uniformity judgment standard based on the dynamic threshold range; wherein, the steps for setting the dynamic threshold range are: obtaining the historical threshold range based on the tunnel geological classification map and historical construction and maintenance records, and setting a benchmark area threshold library based on the historical threshold range. Local features are extracted from the acquired point cloud data of the tunnel fault zone to obtain a local feature dataset; based on the convolutional neural network model, the optimal threshold of the tunnel fault zone is predicted according to the local feature dataset; based on the optimal threshold, the threshold library of the benchmark area is dynamically adjusted to generate a dynamic threshold range. When the calculated thickness deviation exceeds the dynamic threshold range, the deviation area of ​​the lining surface is determined to be an uneven thickness area. Based on the tunnel geological classification map, historical construction and maintenance records, and thickness deviation data, a Bayesian network model is used to predict the probability of the cause of the thickness deviation. Based on the probability of the cause, the maintenance cases are matched according to the set probability of the cause and the corresponding matching relationship library of maintenance cases to obtain the matching result of the optimal maintenance solution.

2. The tunnel fault zone lining uneven thickness detection method of three-dimensional laser point cloud fusion LS according to claim 1, characterized in that, Point cloud data of the tunnel fault zone was acquired using a 3D laser scanner, including: Configure a 3D laser scanner; Cloud data of tunnel fault zones were acquired using a 3D laser scanner.

3. The tunnel fault zone lining uneven thickness detection method of three-dimensional laser point cloud fusion LS according to claim 1, characterized in that, Perform coordinate transformation on the preprocessed tunnel fault zone point cloud data, including: The central axis of the tunnel is determined using the method of connecting the two endpoints. Establish a local coordinate system and adjust the central axis to align with the Z-axis; By using rotation transformation, the preprocessed tunnel fault zone point cloud data is mapped to the global coordinate system.

4. The method for detecting uneven thickness of tunnel fault lining by three-dimensional laser point cloud fusion (LS) according to claim 1, characterized in that, The fitting mathematical model is established using LS, including: The dataset composed of the pretreated tunnel fault zone point cloud data is set as: , The equation of the fitted line is: ， The objective function is to minimize the sum of squared distances from each point to the line: ， The parameters are solved iteratively using matrix operations or numerical optimization tools. , , (a, b, c), which minimizes the objective function value; where n represents the number of points in the point cloud dataset, ( , , ) represents the coordinates of the reference point on the straight line, indicating the geometric center of the tunnel's central axis, used to locate the spatial reference of the fitted straight line; a, b, and c represent the direction vector components of the straight line, indicating the extension of the straight line in the X, Y, and Z directions in three-dimensional space.

Images