A displacement monitoring device and method for auxiliary supervision of deep foundation pit construction.

Abstract

The application relates to the technical field of foundation pit displacement measurement, in particular to a displacement monitoring device and method for assisting supervision of a building deep foundation pit, the method comprising the following steps: scanning the building deep foundation pit by using a three-dimensional laser scanner, and obtaining all point cloud data after each scanning; classifying different point cloud data in the local neighborhood of each point cloud data, and determining local dispersion and distribution abnormality of the point cloud data; obtaining a voxel edge length, and performing down-sampling on the point cloud data; and calculating an offset, and evaluating and monitoring displacement of the deep foundation pit. The application can effectively reduce the data volume and improve the processing efficiency by using the dynamically adjusted voxel edge length for down-sampling processing, meanwhile, the basic features and structural information of the point cloud are reserved, the processing effect of the down-sampling on the point cloud data is improved, and the reliability and precision of the deep foundation pit displacement monitoring are improved.

Description

Technical Field

[0001] This application relates to the field of foundation pit displacement measurement technology, specifically to a displacement monitoring device and method for auxiliary supervision of deep foundation pits in construction. Background Technology

[0002] As an underground construction project, the main function of a deep foundation pit is to transfer the weight of the building to the foundation. Therefore, the standardization of deep foundation pit construction determines the safety of the building. During the construction of deep foundation pits, due to the complex underground environment, uneven soil density distribution, and even the presence of cavities and underground rivers, the excavated deep foundation pit may deform or shift, potentially failing to adequately support the weight of the building above, causing the building to tilt or collapse. Therefore, it is necessary to monitor the displacement of the deep foundation pit.

[0003] In the process of using 3D laser technology to monitor the displacement of deep foundation pits, the laser scanning of the deep foundation pit generates massive amounts of point cloud data, which consumes a lot of computing resources. Traditional methods generally use fixed voxel side lengths for downsampling to reduce the amount of data. However, since the displacement of deep foundation pits may be a local shift, the geometric deformation and shift in different areas will lead to significant differences in the distribution density of the point cloud. Fixed voxel side lengths cannot adapt to the spatial changes in point cloud density, resulting in poor downsampling processing of point cloud data, which causes the loss of detailed information and affects the accuracy of deep foundation pit displacement monitoring. Summary of the Invention

[0004] To address the aforementioned technical problems, a displacement monitoring device and method for auxiliary supervision of deep foundation pits in construction are provided to solve the existing issues.

[0005] The solution to the technical problem of this application is to provide a displacement monitoring device and method for auxiliary supervision of deep foundation pits in construction, including the following steps:

[0006] In a first aspect, embodiments of this application provide a displacement monitoring method for auxiliary supervision of deep foundation pits in construction, the method comprising the following steps:

[0007] A 3D laser scanner was used to scan deep foundation pits of buildings to obtain all point cloud data after each scan.

[0008] For each scan, the difference in distance between any point cloud data and different point cloud data in its local neighborhood is analyzed. Different point cloud data in the local neighborhood are classified. The local dispersion of any point cloud data is determined by the fluctuation of the difference in the number of point cloud data in each category and the dispersion of the distance between all point cloud data in different categories and the any point cloud data.

[0009] The distribution anomaly of any point cloud data is calculated by considering the difference in the number of point cloud data distributed in a local range between any point cloud data and different point cloud data in its local neighborhood, combined with the local dispersion.

[0010] For each scan, based on the spatial distribution information and distribution anomaly of each point cloud data, all point cloud data are clustered. Based on the average level of distribution anomaly of all point cloud data in each cluster, the voxel side length of the spatial region corresponding to each cluster after each scan is obtained. The point cloud voxel downsampling algorithm is used to downsample all point cloud data after each scan.

[0011] The surface of the deep foundation pit is divided into multiple monitoring areas. The offset of the surface where the point cloud data of each monitoring area is located after downsampling is analyzed. The offset of each monitoring area after each scan is calculated, and the displacement of the deep foundation pit is evaluated and monitored.

[0012] Preferably, the classification of different point cloud data within a local neighborhood includes:

[0013] Arrange the distances between any point cloud data and all point cloud data in its local neighborhood in ascending order to form a distance sequence; record the difference between each element in the distance sequence and its previous element as the relative difference.

[0014] Cluster the relative differences of all elements in the distance sequence, calculate the average of the relative differences of all elements in each cluster, and divide the distance sequence into multiple subsequences with the element in the cluster corresponding to the largest average as the dividing point, where each subsequence represents a category.

[0015] Preferably, determining the local dispersion of any point cloud data includes:

[0016] Count the number of all elements in each subsequence, calculate the difference between the number of elements in each subsequence and the previous subsequence, and denote it as the quantity difference; calculate the dispersion of the quantity difference of all subsequences corresponding to the distance sequence, and denote it as the first dispersion.

[0017] Calculate the degree of dispersion of all elements in each subsequence, denoted as the second degree of dispersion. Multiply the mean of the second degree of dispersion of all subsequences corresponding to the distance sequence by the first degree of dispersion, and take the local dispersion of any point cloud data.

[0018] Preferably, calculating the distribution anomaly degree of any point cloud data includes:

[0019] For each scan, the number of all point cloud data within the local neighborhood of each point cloud data is counted and recorded as the neighbor count; the median of the neighbor count of all point cloud data within the local neighborhood of any given point cloud data is obtained.

[0020] Calculate the difference between the number of neighbors of any point cloud data and the median, denoted as the relative difference, and perform a positive mapping on the relative difference;

[0021] The distribution anomaly is the product of the result of the positive mapping and the local dispersion.

[0022] Preferably, the step of clustering all point cloud data includes: calculating the distance from each point cloud data point to the origin after each scan, and recording it as the relative distance; forming a two-dimensional array by combining the relative distance of each point cloud data point after each scan with the distribution anomaly degree; and clustering the two-dimensional array of all point cloud data point after each scan to obtain multiple clusters.

[0023] Preferably, the spatial region is the spatial region containing the smallest cuboid containing all point cloud data within each cluster.

[0024] Preferred, the first The second scan voxel side length of each spatial region The calculation formula is: ,in, For the first The second scan The normalized result of the mean of the distribution anomaly of all point cloud data within each spatial region corresponding to a cluster. For preset adjustment coefficients, The minimum voxel side length is preset. This represents the function for rounding up.

[0025] Preferably, the calculation of the offset of each monitoring area after each scan includes:

[0026] For all point cloud data of each monitoring area after downsampling after the first scan, a surface is fitted, and the fitted surface is used as a reference surface.

[0027] For all point cloud data of each monitoring area that have undergone downsampling after each scan, a surface is fitted, and the fitted surface is recorded as a local surface;

[0028] Calculate the normal distance between the local surface of each monitoring area and the reference plane after each scan, and use it as the offset of each monitoring area after each scan.

[0029] Preferably, the assessment and monitoring of the displacement of the deep foundation pit includes: if the offset is greater than a preset threshold, the deep foundation pit has shifted in the monitoring area; otherwise, the deep foundation pit has not shifted in the monitoring area.

[0030] Secondly, this application also provides a displacement monitoring device for auxiliary supervision of deep foundation pits in construction, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the displacement monitoring method for auxiliary supervision of deep foundation pits in construction described in any one of the above-mentioned embodiments.

[0031] This application has at least the following beneficial effects:

[0032] This application classifies the point cloud data within its neighborhood by analyzing the distance between each point cloud data point and its neighboring point cloud data point after each scan. Its beneficial effect lies in dividing the spherical space represented by the neighborhood into multiple concentric spherical layers according to distance, facilitating subsequent evaluation of the uniformity of the point cloud distribution. It also determines the local dispersion of each point cloud data point, which is beneficial because it considers the differences and fluctuations in the number of point clouds within the concentric spherical layers represented by different categories, as well as the fluctuations in the distance of the point clouds within the concentric spherical layers from the center of the sphere. This reflects the non-uniformity of the point cloud distribution within the local neighborhood of the point cloud data point, thus enhancing the evaluation... The significance of unevenness or other irregular features in the local neighborhood is estimated, thus indicating the possibility of geometric deformation and displacement in the local neighborhood; the distribution anomaly degree of each point cloud data is calculated, which is beneficial because it considers the changes in the geometric shape within the local neighborhood of the point cloud data, reflects the difference between the local distribution density of the point cloud data and the local distribution density of the surrounding point clouds, assesses the degree of point cloud distribution anomaly caused by changes in local geometry within the local neighborhood of the point cloud data, and indicates the possibility of displacement in the local neighborhood of the point cloud data; all point cloud data are then processed. Clustering is used to obtain the voxel edge lengths of the spatial regions corresponding to each cluster after each scan. Its beneficial effect lies in dividing the point cloud into different spatial regions based on its spatial location information and anomalies. This allows for dynamic adjustment of voxel edge lengths based on the distribution anomalies of the point cloud within these regions. Smaller voxel edge lengths are set for regions prone to deformation and displacement to retain more detailed information, while larger voxel edge lengths are set for regions with uniform point cloud distribution. This reduces data volume and saves computational resources while preserving the overall geometric state of the region. A point cloud voxel downsampling algorithm is used to downsample the point cloud data distributed across all spatial regions after each scan, calculating the offset of each monitoring region after each scan. This assesses and monitors the displacement of deep foundation pits. Its beneficial effect lies in dynamically setting the voxel edge lengths based on the point cloud distribution, effectively reducing data volume and improving processing efficiency while preserving the basic characteristics and structural information of the point cloud. This improves the downsampling effect of the point cloud data and enhances the reliability and accuracy of deep foundation pit displacement monitoring. Attached Figure Description

[0033] The following description, in conjunction with the accompanying drawings, provides a more detailed explanation of a displacement monitoring method for auxiliary supervision of deep foundation pits in this application.

[0034] Figure 1 A flowchart illustrating the steps of a displacement monitoring method for auxiliary supervision of deep foundation pits in construction, provided in this application embodiment;

[0035] Figure 2 A flowchart illustrating the steps of the method for obtaining the local dispersion of point cloud data provided in the embodiments of this application. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this application clearer, the displacement monitoring device and method for auxiliary supervision of deep foundation pits proposed in this application will be further described in detail below with reference to the accompanying drawings and implementation examples. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit the scope of this application.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0038] Please see Figure 1 The diagram illustrates a flowchart of a displacement monitoring method for auxiliary supervision of deep foundation pits in construction, according to an embodiment of this application. The method includes the following steps:

[0039] Step 1: Use a 3D laser scanner to scan the deep foundation pit of the building and obtain all point cloud data after each scan.

[0040] Foundation pit engineering is a high-risk project, mainly due to the variability, complexity and uncertainty of geological conditions. Various safety hazards may exist during construction. Deformation of deep foundation pits during construction can lead to safety accidents. Therefore, it is necessary to monitor and report the deformation or displacement of deep foundation pits in a timely manner during construction.

[0041] Based on the above analysis, a 3D laser scanner was used to scan the deep foundation pit of the building to obtain all point cloud data of the deep foundation pit in each scan. Since the surface of the deep foundation pit is an irregular surface, the collected point cloud data will contain noise. Therefore, the point cloud data is filtered.

[0042] It should be noted that the accuracy range of the 3D laser scanner in measurement is 20um~100um. In this embodiment, the measurement accuracy is 20um. As other implementation methods, the implementer can set it according to the actual situation. Secondly, a bilateral filtering algorithm is used to filter and denoise the point cloud data. The bilateral filtering algorithm is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can use other methods of existing technology, such as Gaussian filtering algorithm, etc. This embodiment does not impose any special restrictions on this.

[0043] At this point, all point cloud data from each scan is obtained.

[0044] Step 2: For each scan, analyze the difference in distance between any point cloud data and different point cloud data in its preset local neighborhood. Classify the different point cloud data in the local neighborhood. Determine the local dispersion of any point cloud data by the fluctuation of the number of point cloud data in each category and the dispersion of the distance between all point cloud data in different categories and the any point cloud data.

[0045] Furthermore, since the area covered by deep foundation pits is large, the point cloud data generated by 3D laser scanners is very dense, usually containing millions or even hundreds of millions of point clouds. These data consume a lot of computing resources and time when stored, processed and analyzed. Point cloud voxel downsampling algorithm is a method to reduce computational complexity by reducing the amount of point cloud data, while preserving the geometric structure and main features of the point cloud as much as possible.

[0046] The point cloud voxel downsampling algorithm divides the point cloud into multiple 3D meshes, each called a voxel. A representative point is selected in each voxel to replace all points within that voxel. This method can significantly reduce the amount of data while preserving the overall geometric features. The side length of the voxel mesh is a key parameter, which determines the size of each voxel. If the original point cloud data is very dense, a larger voxel side length can be selected to significantly reduce the amount of data, thereby saving computational resources. If the original point cloud data is relatively sparse, a smaller voxel side length can be selected to retain more geometric details.

[0047] Secondly, in deep foundation pit monitoring, when the point cloud distribution in a certain area is dense and uniform, it indicates that the point cloud data presents a regular geometric shape in space. It can usually be approximated as a symmetrical structure centered on the center point. The spatial distribution characteristics of this area can be characterized by the center point. Therefore, a larger voxel side length can be set for this area to significantly reduce the amount of data.

[0048] Furthermore, the flowchart of the method for obtaining the local dispersion of point cloud data provided in the embodiments of this application is as follows: Figure 2 As shown.

[0049] First, we analyze the distribution of the number of point cloud data within the neighborhood of each point cloud data point, specifically:

[0050] For each scan, a spherical space with a preset radius is centered on any point cloud data and recorded as the local neighborhood;

[0051] In this embodiment, the preset neighborhood radius is set to 100um. In other implementation methods, the implementer can set it according to the actual situation.

[0052] Arrange the distances between any point cloud data and all point cloud data in its local neighborhood in ascending order to form a distance sequence;

[0053] In this embodiment, the distance is measured by calculating the Euclidean distance between any point cloud data and each point cloud data in its local neighborhood. The calculation of the Euclidean distance is a well-known technique and will not be described in detail here.

[0054] The difference between each element in the distance sequence and its preceding element is denoted as the relative difference;

[0055] It should be noted that the relative difference is not calculated for the first element in the distance sequence.

[0056] Cluster the relative differences of all elements in the distance sequence to obtain two clusters;

[0057] In this embodiment, the K-means clustering algorithm is used for clustering. The K-means clustering algorithm is a well-known technology and will not be described in detail here. As other implementation methods, implementers may use other methods of existing technology, such as the DBSCAN clustering algorithm. This embodiment does not impose any special restrictions on this.

[0058] Calculate the average of the relative differences of all elements within each cluster, and use the element within the cluster corresponding to the largest average as the split point to divide the distance sequence into multiple subsequences;

[0059] It should be noted that each split point is assigned to the next subsequence; for ease of understanding, let's assume the distance sequence is... ,in, , , If is the dividing point, then the subsequences are respectively , , , .

[0060] It should be noted that the local neighborhood is a spherical space with the point cloud data as the center and a preset neighborhood radius. This spherical space is divided into multiple concentric spherical layers by dividing points. If the point cloud data is evenly distributed in the local neighborhood, the number of point cloud data contained in the concentric spherical layers will increase with the increase of distance, that is, the number of elements in the subsequence will gradually increase.

[0061] Count the number of all elements in each subsequence, and calculate the difference between the number of elements in each subsequence and the number of elements in the previous subsequence. This difference is denoted as the quantity difference.

[0062] Calculate the degree of dispersion of the quantity difference of all subsequences corresponding to the distance sequence, and denote it as the first degree of dispersion;

[0063] In this embodiment, the degree of dispersion is measured by calculating the coefficient of variation of the quantity difference of all subsequences corresponding to the distance sequence. The calculation of the coefficient of variation is a well-known technique and will not be described in detail here. As other implementation methods, implementers may use other methods of the prior art, such as variance, standard deviation, etc. This embodiment does not impose any special restrictions on this.

[0064] It should be noted that the smaller the first dispersion, the more uniform the distribution of point cloud data in the local neighborhood; conversely, the larger the dispersion, the less uniform the distribution of point cloud data in the local neighborhood, and there may be local changes or displacements.

[0065] Secondly, due to the uneven surface of the deep foundation pit, there are subtle differences in the distance from the center of the point cloud data contained within a concentric spherical layer. By analyzing the fluctuation of elements within each subsequence and combining it with the first degree of dispersion, the local dispersion is calculated, specifically as follows:

[0066] Calculate the degree of dispersion of all elements in each subsequence, denoted as the second degree of dispersion. Multiply the mean of the second degree of dispersion of all subsequences corresponding to the distance sequence by the first degree of dispersion, and take the local dispersion of any point cloud data.

[0067] In this embodiment, the degree of dispersion is measured by calculating the standard deviation of all elements within each subsequence. As in other implementations, implementers may use other methods of the prior art, such as variance, etc. This embodiment does not impose any special restrictions on this.

[0068] It should be noted that the larger the second dispersion, the greater the difference in distance from the center of the point cloud data contained in the concentric sphere layer represented by the subsequence, reflecting the more uneven distribution of the point cloud data, and the possible existence of local clustering or sparse phenomena; the larger the obtained local dispersion, the more uneven the distribution of the point cloud data in the local neighborhood, reflecting the more significant the unevenness or other irregular features in the local neighborhood, and the greater the possibility that the location of the point cloud data in the deep foundation pit has undergone significant displacement.

[0069] Thus, the local dispersion of each point cloud data after each scan is obtained.

[0070] Step 3: Calculate the distribution anomaly of any point cloud data by considering the difference in the number of point cloud data distributed within a local range between any point cloud data and different point cloud data in its local neighborhood, combined with the local dispersion. For each scan, cluster all point cloud data based on the spatial distribution information and distribution anomaly of each point cloud data. Based on the average level of the distribution anomaly of all point cloud data within each cluster, obtain the voxel side length of the spatial region corresponding to each cluster after each scan.

[0071] Furthermore, when a point cloud location in a deep foundation pit shifts, it causes changes in the local geometry, such as local depressions, bulges, or tilts. These geometric changes result in uneven distribution of point cloud data, manifesting as localized density or sparseness. In such cases, the likelihood of displacement within that local area increases significantly. To more accurately capture these changes and retain more detailed information, the voxel side length of the point cloud voxel downsampling algorithm should be appropriately set. For example, in local areas where displacement is possible, a smaller voxel side length should be selected to sample more point cloud data. A smaller voxel side length can more accurately reflect changes in local geometry, thereby improving the accuracy of subsequent deep foundation pit displacement measurements. Conversely, in areas where no displacement has occurred, a larger voxel side length can be used to reduce the amount of data and improve processing efficiency.

[0072] Based on the above analysis, the distribution anomaly is calculated by combining the distribution of point cloud data within a local neighborhood with the local dispersion. Specifically:

[0073] The number of all point cloud data within the local neighborhood of each point cloud data is counted and denoted as the neighbor count.

[0074] Obtain the median of the number of neighbors of all point cloud data within the local neighborhood of any given point cloud data;

[0075] Calculate the difference between the number of neighbors of any point cloud data and the median, denoted as the relative difference, and perform a positive mapping on the relative difference;

[0076] In this embodiment, the absolute value of the difference between the number of neighbors of any point cloud data and the median is denoted as the relative difference; secondly, the positive mapping process is as follows: positive mapping is performed through an exponential function, assuming the relative difference is denoted as... ,but The result is taken as the result of the positive mapping, where, It is an exponential function with the natural constant as the base; through the positive mapping process, the result of the positive mapping is made to be greater than 0.

[0077] The product of the positive mapping result and the local dispersion is used as the distribution anomaly degree of any point cloud data.

[0078] It should be noted that the larger the positive mapping result, the more significantly the local distribution density of the point cloud data is different from the local distribution density of the surrounding point cloud. The higher the degree of distribution anomaly, the greater the obtained distribution anomaly degree, indicating that there is an anomaly in the distribution of the point cloud data at its location, which may be due to changes in local geometry, and the greater the possibility of displacement.

[0079] Secondly, displacement within deep foundation pits may not be a systemic displacement, but rather a spatial regional displacement, and multiple regions may exist within this displacement area. While the anomaly rate of point cloud data distribution may be the same across different regions, their spatial distribution differs, and they cannot be simply classified as the same type. Therefore, it is necessary to classify the point cloud data by combining its spatial information and anomaly rate, specifically as follows:

[0080] Calculate the distance from each point cloud data point to the origin, and record it as the relative distance;

[0081] It should be noted that the origin is usually the location of the 3D laser scanner, where the coordinates of the origin are... .

[0082] The relative distances and distribution anomalies of each point cloud data after each scan are combined into a two-dimensional array;

[0083] Cluster the two-dimensional array of all point cloud data after each scan to obtain multiple clusters;

[0084] In this embodiment, the DBSCAN clustering algorithm (Density-Based Spatial Clustering of Applications with Noise) is used for clustering. The DBSCAN clustering algorithm is a well-known technology and will not be described in detail here. As other implementation methods, implementers can use other methods of existing technology, such as hierarchical clustering algorithms. This embodiment does not impose any special restrictions on this.

[0085] The spatial region containing the smallest cuboid containing all point cloud data within each cluster is taken as the spatial region corresponding to each cluster.

[0086] It should be noted that, for ease of understanding, the process of determining the minimum cuboid is as follows: obtain the maximum and minimum values ​​of all point cloud data within the cluster in each dimension, that is, obtain the maximum and minimum values ​​of all point cloud data within the cluster on the X-axis, denoted as […]. , The maximum and minimum values ​​on the Y-axis are denoted as follows: , The maximum and minimum values ​​on the Z-axis are denoted as follows: , , with 8 vertices , , , , , , , The area represented by the enclosed cuboid is considered as a spatial region.

[0087] The formula for calculating the voxel side length of each spatial region after each scan is:

[0088]

[0089] in For the first The second scan voxel side length of a spatial region For the first The second scan The normalized result of the mean of the distribution anomaly of all point cloud data within each spatial region corresponding to a cluster. For preset adjustment coefficients, The minimum voxel side length is preset. This represents the floor function;

[0090] In this embodiment, the sigmoid function is used for normalization. The sigmoid function is a well-known technique and will not be described in detail here. As for other implementation methods, implementers can use other methods from the prior art, such as the softmax function, tanh function, etc. This embodiment does not impose any special restrictions on this. Secondly, a preset adjustment coefficient is used. The value is set to 6, which is used to control the adjustment range of the voxel side length, and is the preset minimum voxel side length. The value is set to 4, so that the range of values ​​for the voxel side length is within... As another implementation method, the implementer can set it according to the actual situation.

[0091] It should be noted that the normalization result... The larger the value, the higher the degree of point cloud distribution anomaly in the spatial region. The more likely the spatial region will be displaced due to geometric deformation. In this case, a smaller voxel side length should be set for this spatial region to sample more point cloud data and retain more detailed information. Conversely, the smaller the value, the more uniform the point cloud distribution in the spatial region. The less likely the point cloud will be displaced, the smaller the voxel side length should be set for this spatial region. This will reduce the amount of data while preserving the overall geometric state of the spatial region and saving computational resources.

[0092] Thus, the voxel side lengths of each spatial region after each scan are obtained.

[0093] Step 4: Use the point cloud voxel downsampling algorithm to downsample all point cloud data after each scan; divide the surface of the deep foundation pit into multiple monitoring areas, analyze the offset of the surface where the point cloud data of each monitoring area is located after each scan, calculate the offset of each monitoring area after each scan, and evaluate and monitor the displacement of the deep foundation pit.

[0094] Furthermore, based on the voxel side length, the point cloud data in each spatial region under each scan is downsampled, specifically as follows:

[0095] For each scan, based on the voxel side length, a point cloud voxel downsampling algorithm is used to downsample the point cloud data in each spatial region;

[0096] It should be noted that the point cloud voxel downsampling algorithm is a well-known technology and will not be described in detail here. The point cloud voxel downsampling process is as follows: by dividing each spatial region into a cubic grid with a side length equal to the side length of the voxel, each cubic grid represents a voxel, and a representative point is retained within each voxel. The representative point can be the centroid of each voxel or the center point of each voxel, thereby realizing the downsampling process of the point cloud data.

[0097] The surface of the deep foundation pit was divided into multiple monitoring zones;

[0098] In this embodiment, the surface of the deep foundation pit is divided into 1000 monitoring areas. As for other implementation methods, the implementer can set them according to the actual situation.

[0099] All point cloud data sampled after the first scan of each monitoring area are fitted with a surface, and the fitted surface is used as a reference surface.

[0100] In this embodiment, the nonlinear least squares method is used for surface fitting. The nonlinear least squares method is a well-known technique and will not be described in detail here.

[0101] For each monitoring area, all point cloud data sampled after each scan are fitted with a surface, which is denoted as a local surface;

[0102] Calculate the normal distance between the local surface of each monitoring area and the reference plane after each scan, and use it as the offset of each monitoring area after each scan;

[0103] It should be noted that the normal distance refers to the perpendicular distance between two planes along the direction of the normal vector. The calculation of the normal distance is a well-known technique and will not be elaborated here.

[0104] If the offset is greater than the preset threshold, the deep foundation pit will be displaced in the monitoring area; otherwise, the deep foundation pit will not be displaced in the monitoring area.

[0105] In this embodiment, the preset threshold value is 0.1 mm. In other implementation methods, the implementer can set it according to the actual situation.

[0106] If a deep foundation pit shifts in a certain monitoring area, an early warning will be issued in a timely manner to ensure construction safety.

[0107] Based on the same inventive concept as the above method, this application embodiment also provides a displacement monitoring device for auxiliary supervision of deep foundation pits in construction, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described displacement monitoring methods for auxiliary supervision of deep foundation pits in construction.

[0108] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0109] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0110] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application. Therefore, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application, without departing from the content of the technical solution of this application, shall fall within the protection scope of the technical solution of this application.

Claims

1. A displacement monitoring method for assisting supervision of a building deep foundation pit, characterized in that, The method comprises the following steps: Scanning the deep foundation pit of the building by using a three-dimensional laser scanner to obtain all point cloud data after each scanning; For each scanning, analyzing the difference between the distance of any point cloud data and different point cloud data in its local neighborhood, classifying the different point cloud data in the local neighborhood, and determining the local dispersion of the any point cloud data by the fluctuation of the number of point cloud data in each category and the dispersion of the distance between all point cloud data in different categories and the any point cloud data; Calculating the distribution anomaly degree of the any point cloud data by the number difference of the point cloud data distributed in the local range between the any point cloud data and different point cloud data in its local neighborhood, combined with the local dispersion; For each scanning, clustering all point cloud data based on the spatial distribution information and the distribution anomaly degree of each point cloud data, obtaining the voxel edge length of each clustering cluster corresponding to the spatial region after each scanning based on the average level of the distribution anomaly degree of all point cloud data in each clustering cluster, and performing down-sampling processing on all point cloud data after each scanning by using point cloud voxel down-sampling algorithm; Dividing the surface of the deep foundation pit into multiple monitoring regions, analyzing the offset of the surface of each monitoring region in the point cloud data after each scanning and down-sampling processing, calculating the offset of each monitoring region after each scanning, evaluating and monitoring the displacement of the deep foundation pit; The classification of the different point cloud data in the local neighborhood comprises: Arranging the distance between the any point cloud data and all point cloud data in its local neighborhood in ascending order to form a distance sequence; the difference between each element in the distance sequence and its previous element is recorded as a relative difference; Clustering all elements in the distance sequence, calculating the average value of the relative difference of all elements in each clustering cluster, taking the element in the clustering cluster corresponding to the maximum average value as a segmentation point, and dividing the distance sequence into multiple subsequences, wherein each subsequence represents a category; The determination of the local dispersion of the any point cloud data comprises: Counting the number of all elements in each subsequence, calculating the difference between the number of each subsequence and its previous subsequence, recording the difference as a number difference, calculating the dispersion degree of the number difference of all subsequences corresponding to the distance sequence, recording the dispersion degree as a first dispersion degree; Calculating the dispersion degree of all elements in each subsequence, recording the dispersion degree as a second dispersion degree, and taking the product of the average of the second dispersion degree of all subsequences corresponding to the distance sequence and the first dispersion degree as the local dispersion of the any point cloud data; The calculation of the distribution anomaly degree of the any point cloud data comprises: For each scanning, counting the number of all point cloud data in the local neighborhood of each point cloud data, recording the number as a neighborhood number, and obtaining the median of the neighborhood number of all point cloud data in the local neighborhood of the any point cloud data; Calculating the difference between the neighborhood number of the any point cloud data and the median, recording the difference as a relative difference, and positively mapping the relative difference; The distribution anomaly degree is the product of the positively mapped result and the local dispersion.

2. The displacement monitoring method for assisting the supervision of a building deep foundation pit according to claim 1, wherein, The clustering of all point cloud data comprises: calculating the distance of each point cloud data after each scan to the origin, denoted as a relative distance; combining the relative distance of each point cloud data after each scan with a distribution anomaly degree to form a two-dimensional array; and clustering the two-dimensional array of all point cloud data after each scan to obtain a plurality of clustering clusters.

3. The displacement monitoring method for supervising a deep foundation pit of a building according to claim 1, wherein The spatial region is a spatial region in which a minimum cuboid containing all point cloud data in each clustering cluster is located.

4. The displacement monitoring method for supervising a deep foundation pit of a building according to claim 1, wherein No. The second scan voxel side length of each spatial region The calculation formula is: ,in, For the first The second scan The normalized result of the mean of the distribution anomalies of all point cloud data within each spatial region corresponding to a cluster. For preset adjustment coefficients, The minimum voxel side length is preset. This represents the function for rounding up.

5. The displacement monitoring method for supervising a deep foundation pit of a building according to claim 1, wherein The calculation of the offset of each monitoring region after each scan comprises: performing surface fitting on all point cloud data of each monitoring region after the first scan for down-sampling processing, and taking the fitted surface as a reference datum plane; performing surface fitting on all point cloud data of each monitoring region after each scan for down-sampling processing, and taking the fitted surface as a local surface; calculating the normal distance between the local surface corresponding to each monitoring region after each scan and the reference datum plane as the offset of each monitoring region after each scan.

6. The displacement monitoring method for supervising a deep foundation pit of a building according to claim 1, wherein The evaluation and monitoring of the displacement of the deep foundation pit comprises: if the offset is greater than a preset threshold, the deep foundation pit has displacement in the monitoring region; otherwise, the deep foundation pit has no displacement in the monitoring region.

7. A displacement monitoring device for assisting supervision of a building deep foundation pit, comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that, The processor executes the computer program to implement the steps of the displacement monitoring method for the auxiliary supervision of a building deep foundation pit according to any one of claims 1-6.

Images