Concrete spreading process blank layer thickness real-time monitoring method based on improved hausdorff-DBSCAN clustering

Abstract

The invention relates to the field of water conservancy and hydropower construction, in particular to a concrete spreading process blank layer thickness real-time monitoring method based on improved hausdorff-DBSCAN clustering, and the method comprises the steps: collecting concrete spreading process data in real time; preprocessing the collected data, and converting the data into actual leveling process data according to the mutual relation between the coordinates of the warehouse surface construction monitoring equipment and the elevation of the concrete surface; performing elevation change characteristic analysis and dividing into a working section and a moving section; combining data of all the working sections, clustering tracks in the working sections by adopting an improved hausdorff-DBSCAN clustering method, distinguishing different blank layers by different clusters, and visually displaying a time-elevation map and a plane track map; and comparing the elevation of the clustering center point of the current working blank layer with the design center elevation of the blank layer, and sending out alarm information. According to the invention, monitoring and early warning of the thickness of the blank layer for spreading the silo surface can be completed, so that the concrete thickness of the blank layer is kept within a reasonable interval, the quality of concrete spreading construction of the silo surface is ensured, and the safety and stability of a concrete dam are improved.

Description

Technical Field

[0001] This invention relates to the field of water conservancy and hydropower construction, and in particular to a method for real-time monitoring of the thickness of concrete slabs during the leveling process based on improved hausdorff-DBSCAN clustering. Background Technology

[0002] The quality of concrete pouring and compaction directly affects the operational safety of dams, and leveling is a crucial step. However, the construction of arch dams is subject to various complex constraints, making quality control extremely difficult. Conventional methods for controlling the quality of arch dam pouring and compaction primarily rely on manual inspections and on-site monitoring. Currently, there is a lack of high-precision, automated real-time quality control methods. This results in the arch dam leveling construction quality control being heavily influenced by human factors, characterized by lax management and low precision. Consequently, problems such as substituting vibration for leveling and failing to meet standards for the number of compaction passes and thickness exist in actual construction, seriously affecting the dam's construction quality and safety. Therefore, adopting real-time monitoring technology for arch dam leveling, characterized by real-time performance, all-weather capability, and high precision, is of great significance for achieving refined management of engineering construction and ensuring construction quality.

[0003] Currently, Zhong Denghua et al. proposed a real-time monitoring method for the construction quality of core-wall rockfill dams in 2009. This method achieves online monitoring and feedback control of the filling, compaction, and material transportation processes of core-wall rockfill dams. However, this method was developed specifically for the construction characteristics of core-wall rockfill dams, which differ fundamentally from arch dams in terms of construction technology and quality control, and is not suitable for real-time control of the construction quality of arch dam pouring and compaction. Peng Hua et al. proposed a real-time monitoring method and system for the construction quality of concrete dams. This method monitors the construction of concrete dams by focusing on raw materials, concrete, and pouring sites. However, this method focuses on collecting data such as concrete temperature, pouring time, and pouring volume during normal concrete dam construction, and does not achieve real-time, precise, and automated monitoring of the site construction process. In roller-compacted concrete dam construction, site construction is crucial to the construction quality; therefore, this method is also not suitable for real-time monitoring of the construction quality of roller-compacted concrete dam pouring and compaction. In 2014, Zhong Denghua et al. proposed a real-time monitoring method for the construction quality of roller-compacted concrete dam pouring surfaces and a real-time monitoring method for the bonding time between layers, which improved the quality control level and efficiency of arch dam construction. However, the monitoring method adopted in this study has a certain lag, and can only issue an alarm after construction problems occur, thus allowing for remedial measures to be taken.

[0004] Currently, most engineering practices employ rather crude monitoring methods, relying solely on comparing real-time monitoring indicators with preset control standards to determine whether to issue an alarm, which has a certain degree of lag. However, with technological advancements, research on clustering methods both domestically and internationally is quite comprehensive, and many improvements have been made to traditional clustering methods. Combining clustering with construction progress control can enable the monitoring and early warning of the thickness of the concrete layer during leveling, ensuring that the thickness of the concrete layer remains within a reasonable range, guaranteeing the quality of concrete leveling construction, and improving the safety and stability of concrete dams. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution: According to a first aspect of the present invention, the present invention claims protection for a method for real-time monitoring of the thickness of the concrete layer during the leveling process based on improved hausdorff-DBSCAN clustering, comprising the following steps: S1, real-time acquisition of initial process data of the concrete leveling to be monitored; S2, preprocess the collected initial process data, and convert the preprocessed initial process data into actual leveling process data based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface. S3, perform elevation change characteristic analysis on the actual closing process data, and divide the actual closing process data into working segment and moving segment; S4. Merge all working segment data and use the improved hausdorff-DBSCAN clustering method to cluster the trajectories within the working segment. Different clusters are used to distinguish different billet layers, and the time-elevation map and plane trajectory map are displayed through visualization.

[0006] S5. Compare the elevation of the cluster center point of the current working blank layer with the design center elevation of the current working blank layer. If the elevation deviation is greater than the set threshold, an alarm message is issued.

[0007] Furthermore, S1 also includes: Real-time positioning data of the current billet construction machinery is obtained through the surface construction monitoring equipment: Equipment number, time, XY plane position, elevation, speed, direction, number of positioning satellites; During the start-up of construction machinery, the surface construction monitoring equipment automatically records one data point per second of the leveling process, forming a sequence of leveling process data points (A1, A2, …, A…). i , …, A m-1 A m ), where i=1,2…(m 1), m, where m represents the number of data points.

[0008] Furthermore, S2 also includes: S21, preprocess the initial process data of the concrete leveling to be monitored collected in real time, deleting data with a rate and direction of 0, a number of positioning satellites below the normal value, and abnormal elevation values, forming a preprocessed sequence of leveling process data points (P1, P2, …, P…). j , …, P n-1 , P n ), where j=1,2…(n 1), n, where n represents the number of data points after preprocessing; S22, Based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface, the preprocessed data point sequence of the leveling process is converted into the actual leveling process data point sequence (p1, p2, …, p…). j , …, p n-1 , p n ), that is, the coordinates of the j-th point in the preprocessed closing process data point sequence are (P jx , P jy , P jz The coordinates of point j in the actual closing process data point sequence are (p jx , p jy , p jz ), where p jx =P jx p jy =P jy p jx =f(P jx ).

[0009] Furthermore, S3 also includes: S31, Set elevation change threshold H limit1 ; S32, set the window size to W, indicating that the sliding window monitors the elevation changes of the latest W actual closing process data points; use the sliding window to monitor the elevation changes of the actual closing process data point sequence, the elevation change is ΔH1=p maxz -p minz In the formula, p maxz p minz These are the highest and lowest elevation points within the window, respectively. S33, if the elevation change ΔH within the current window exceeds the preset warehouse surface elevation change threshold H limit Then mark the last data point before the sliding window as the end point G of the previous working segment. bk The first data point within the sliding window is marked as the starting point Y of the next moving segment. al And mark the current working state as moving; S34, if the elevation change ΔH within the current window falls below the preset warehouse elevation change threshold H again. limit Then mark the last data point before the sliding window as the end point Y of the previous moving segment. bl The first data point within the sliding window is marked as the starting point G of the next working segment. ak And mark the current working status as working; S35, based on the starting point (G) of each working segment and moving segment a1 G a2 , …, G ak , …, G a(o-1) Y ao Y a1 ,Y a2 , …, Y aK , …, Y a(O-1) Y aO ) and endpoint (G) b1 G b2 , …, G bk , …, G b(o-1) Y bo Y b1 Y b2 ,…, Y bK , …, Y b(O-1) Y bO The data is divided into different working segments and moving segments, i.e., G. ak To G bk The set of points between them is the kth working segment, Y aK To Y bK The set of points between them is the Kth moving segment, where o represents the number of working segments and O represents the number of moving segments.

[0010] Furthermore, in S4, the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data.

[0011] Furthermore, S4 also includes: S41, merge all identified working segments; S42, extract time and elevation as features, and standardize using standard deviation; S43, data points within the working segment with a speed set to 0 and a direction change amplitude greater than a preset amplitude threshold are considered as nodes. Data points between two nodes are considered as a trajectory segment. The set of points T within this trajectory segment is used as a clustered data point, thereby generating a clustered dataset D = (T1, T2, …, T…). J , …, T M , T M ,), where M is the number of all trajectories within the working segment.

[0012] S44 uses the improved hausdorff-DBSCAN clustering algorithm to perform clustering, assigns cluster labels to the corresponding working segments, and marks the noise points in S43 as moving segments; S45. Generate a visualization view containing time-elevation map and plane trajectory map based on the clustering results. The working segment is a different billet layer, which is represented in different forms according to the clustering results. The moving segment is represented by a unified symbol.

[0013] Furthermore, in S42, standard deviation standardization is used to remove the mean and normalize the variance of the extracted time and elevation data to improve clustering accuracy.

[0014] Furthermore, in S43, the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data by replacing the Euclidean distance in DBSCAN clustering with the hausdorff distance, which is suitable for trajectory clustering, as the similarity measure.

[0015] The improved hausdorff-DBSCAN clustering method is a density-based method; The parameter eps is specified as the neighborhood radius. The similarity and whether two data points belong to the same cluster are determined based on whether the hausdorf distance between two clustered data points is less than or equal to eps. Specify another parameter, minPts. When the number of points in the neighborhood of a cluster data point reaches or exceeds minPts, the point is considered a core point. Iterate through all points T in the clustering dataset D, if T J If it is not visited, mark it as visited and calculate T. J The Hausdorff distance h to all other points within D J ; (1) In the formula, T x For any other clustered data point within D, p a p b Trajectory T J T x The actual closing process data points within, d(p) a , p b ) is p a p b The Euclidean distance between them; (2) In the formula, p is... a p b The coordinates are respectively (p ax , p ay , paz ) and (p bx , p by , p bz ); Calculate h J If the number of points ≤ eps is less than MinPts, then the point is temporarily marked as noise; if it reaches or exceeds MinPts, it is considered a core point, and a new cluster C is created with this point as the core. L and T J All points within the neighborhood are added to the Seed set to be processed as the initial seed. J ; The seed set is processed in a loop. Each seed point is taken out in turn. If it has not been visited, its neighborhood is visited and its neighborhood is calculated. If the number of its neighborhood points also reaches or exceeds MinPts, the new points in its neighborhood are added to the seed set. Regardless of whether a seed point is a core point, as long as it has not yet belonged to any cluster, it is added to the current cluster. This loop continues until the seed set is empty, completing cluster C. L The expansion of the cluster is such that points temporarily marked as noise are added as boundary points if they are connected. After processing all points, the points that are not assigned to any cluster are considered noise points.

[0016] The improvement of the hausdorff-DBSCAN clustering method depends on the values ​​of the two parameters eps and minPts. Different parameter values ​​will lead to significant changes in the clustering effect. The optimal values ​​are determined based on the actual closing process and multiple trials.

[0017] Furthermore, S5 also includes: S51, Set elevation deviation threshold H limit2 ; S52, if the current working state is moving, no judgment is made; If the current working status is working, then calculate the deviation value ΔH2=Cz-Hz between the current working billet layer cluster center point elevation and the design center elevation of the billet layer. In the formula, C is the cluster closest to the current time, i.e. the current working billet layer, Cz is the average elevation within the trajectory of the cluster center point TC of the cluster, and Hz is the design center elevation of the billet layer. S53, when the deviation value ΔH2 is greater than the set threshold Hlimit2, an alarm is issued.

[0018] The advantages and positive effects of this invention are: By analyzing the characteristics of the concrete leveling process, the leveling trajectory of construction machinery is pre-divided into working segments and moving segments. Cluster analysis is performed only on the working segment trajectory to improve computational efficiency and effectiveness.

[0019] To address the requirements of trajectory clustering, the hausdorff distance, which is more suitable for trajectory clustering, is used instead of the traditional Euclidean distance in DBSCAN clustering as a similarity metric, significantly improving the ability to handle complex shape clusters (flattening trajectories).

[0020] The improved hausdorff-DBSCAN clustering method relies on density rather than distance for clustering, which overcomes the limitation of distance-based algorithms that can only discover spherical clusters. It is more suitable for clustering irregularly shaped clearing trajectories with obvious spatiotemporal continuity. Attached Figure Description

[0021] Figure 1 The present invention requests protection for a flowchart of a method for real-time monitoring of concrete layer thickness during leveling based on improved hausdorff-DBSCAN clustering. Figure 2 This invention requests protection for a second workflow diagram of a method for real-time monitoring of the thickness of concrete slabs during the leveling process based on improved hausdorff-DBSCAN clustering; Figure 3 The present invention requests protection for a third workflow diagram of a method for real-time monitoring of the thickness of the concrete layer during the leveling process based on improved hausdorff-DBSCAN clustering. Figure 4 The present invention requests protection for the fourth flowchart of a method for real-time monitoring of the thickness of concrete slabs during the leveling process based on improved hausdorff-DBSCAN clustering. Figure 5 The present invention requests protection for the fifth workflow diagram of a method for real-time monitoring of the thickness of concrete slabs during the leveling process based on improved hausdorff-DBSCAN clustering. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0023] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0024] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0025] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for a method for real-time monitoring of the thickness of the concrete layer during the leveling process based on improved Hausdorff-DBSCAN clustering, comprising the following steps: S1, real-time acquisition of initial process data of the concrete leveling to be monitored; S2, preprocess the collected initial process data, and convert the preprocessed initial process data into actual leveling process data based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface. S3, perform elevation change characteristic analysis on the actual closing process data, and divide the actual closing process data into working segment and moving segment; S4. Merge all working segment data and use the improved hausdorff-DBSCAN clustering method to cluster the trajectories within the working segment. Different clusters are used to distinguish different billet layers, and the time-elevation map and plane trajectory map are displayed through visualization.

[0026] S5. Compare the elevation of the cluster center point of the current working blank layer with the design center elevation of the current working blank layer. If the elevation deviation is greater than the set threshold, an alarm message is issued.

[0027] The following embodiments detail the implementation process of a method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering. These embodiments cover all aspects of claims 1 to 9, including explanations of steps, experimental test data, and logical derivations.

[0028] This example is based on a simulated concrete leveling construction scenario. It assumes that the monitoring equipment for the leveling surface is installed on the leveling machinery, data collection continues for 2 hours, recording one data point per second, with an initial total of 7200 data points. All descriptions are presented in text form and do not include code, tables, or diagrams.

[0029] This example simulates a concrete dam leveling construction process. The leveling machinery is a bulldozer, equipment number EQ001. The design elevation of the foundation layer center is 10.0 meters, and the elevation deviation threshold is set to 0.05 meters. The method steps are as follows: S1: Real-time acquisition of initial process data; the EQ001 device records data every second, including timestamp, X coordinate, Y coordinate, Z elevation, velocity, azimuth angle, and number of positioning satellites. The data point sequence is (A1, A2, …, A7200).

[0030] For example, the data for point A1 is: time = 0 seconds, X = 100.0 meters, Y = 200.0 meters, Z = 10.5 meters, speed = 0.5 meters / second, direction = 90°, satellites = 6; the data for point A2 is: time = 1 second, X = 100.5 meters, Y = 200.0 meters, Z = 10.5 meters, speed = 0.5 meters / second, direction = 90°, satellites = 6. Continuous data collection forms a sequence.

[0031] S2: Preprocess the initial data and convert it into actual closing process data. First, delete invalid data: points with both velocity and direction of 0, points with fewer than 5 satellites, and points with elevation anomalies. After deleting 200 points, the remaining 7000 points form a sequence (P1, P2, …, P7000). Then, transform the coordinates: the actual elevation is calculated using the function f. Assuming the equipment elevation is 0.5 meters higher than the actual elevation, therefore pjz = Pjz - 0.5. For example, the actual coordinates of point P1 are (100.0, 200.0, 10.0), and point P2 is (100.5, 200.0, 10.0). The final actual data point sequence is (p1, p2, …, p7000).

[0032] S3: Elevation change characteristic analysis divides the system into working segments and moving segments. The elevation change threshold Hlimit1 = 0.1 meters is set, and the sliding window size W = 10. The actual data sequence is monitored: for example, the elevation from window points p1 to p10 is 10.0 meters to 10.1 meters, ΔH1 = 0.1 meters; the elevation from points p11 to p20 is 10.1 meters to 10.25 meters, ΔH1 = 0.15 meters. Point p20 is marked as the end point of the working segment Gb1, and point p21 is marked as the start point of the moving segment Ya1, with the status "moving"; the elevation from points p30 to p39 is 10.3 meters to 10.35 meters, ΔH1 = 0.05 meters. Point p39 is marked as the end point of the moving segment Yb1, and point p40 is marked as the start point of the working segment Ga2, with the status "working". Ultimately, 5 working segments and 4 moving segments are divided.

[0033] S4: Merge and cluster work segment data. Merge all work segment data, totaling 3000 points. Extract time and elevation features: convert time to seconds starting from 0 seconds, and use elevation directly. Standard deviation standardization: the mean of time features is 1800 seconds and the standard deviation is 600 seconds; the mean of elevation features is 10.2 meters and the standard deviation is 0.1 meters. Subtract the mean from each feature value and divide by the standard deviation. Then, segment the trajectory: set a threshold of 30° for the magnitude of directional change; when the directional change exceeds 30°, it is considered a node, and the set of points between two nodes is the trajectory. Generate a clustered dataset D=(T1, T2, …, T50). Improve Hausdorff-DBSCAN clustering: parameters eps=0.5 meters, minPts=3. Calculate the Hausdorff distance between trajectories; for example, the Hausdorff distance between T1 and T2 is 0.3 meters, indicating they are similar. T1 has 4 neighborhood points (≥minPts), making it a core point, and a cluster CL1 is created. After expansion, 3 clusters (CL1, CL2, CL3) and 5 noisy trajectories are formed. Visualization: In the time-elevation plot, cluster CL1 is represented by red dots, CL2 by blue dots, and CL3 by green dots, with moving segments represented by gray dots; in the planar trajectory plot, clusters are represented by different line types.

[0034] S5: Elevation Deviation Monitoring and Alarm. Set the elevation deviation threshold Hlimit2=0.05 meters. When the current state is working, calculate the elevation of the cluster center point of CL3 as Cz=10.08 meters, the design center elevation as Hz=10.0 meters, the deviation ΔH2=0.08 meters, and issue an alarm message "Bill layer elevation exceeds limit, please adjust".

[0035] Furthermore, S1 also includes: Real-time positioning data of the current billet construction machinery is obtained through the surface construction monitoring equipment: Equipment number, time, XY plane position, elevation, speed, direction, number of positioning satellites; During the start-up of construction machinery, the surface construction monitoring equipment automatically records one data point per second of the leveling process, forming a sequence of leveling process data points (A1, A2, …, A…). i , …, A m-1 A m ), where i=1,2…(m 1), m, where m represents the number of data points.

[0036] In this embodiment, device EQ001 records one data point per second during the start-up of the construction machinery, forming a sequence (A1, A2, …, A7200). Each data point includes: device number = EQ001, time, X coordinate, Y coordinate, Z elevation, speed, direction, and number of positioning satellites. For example, the detailed data for point A1 is as above; the data for point A3600 is: time = 3600 seconds, X = 125.0 meters, Y = 225.0 meters, Z = 10.3 meters, speed = 0.8 meters / second, direction = 180°, and number of satellites = 7. The data point sequence completely covers the 2-hour construction process.

[0037] Furthermore, referring to Figure 2 S2 also includes: S21, preprocess the initial process data of the concrete leveling to be monitored collected in real time, deleting data with a rate and direction of 0, a number of positioning satellites below the normal value, and abnormal elevation values, forming a preprocessed sequence of leveling process data points (P1, P2, …, P…). j , …, P n-1 , P n ), where j=1,2…(n 1), n, where n represents the number of data points after preprocessing; S22, Based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface, the preprocessed data point sequence of the leveling process is converted into the actual leveling process data point sequence (p1, p2, …, p…). j , …, p n-1 , p n ), that is, the coordinates of the j-th point in the preprocessed closing process data point sequence are (P jx , P jy , P jz The coordinates of the j-th point in the actual closing process data point sequence are (pjx, pjy, pjz), where pjx=Pjx, pjy=Pjy, and pjx=f(Pjx).

[0038] In this embodiment, S21: Preprocessing to delete invalid data. There are 150 points with both velocity and direction of 0; 40 points with fewer than 5 satellites; and 10 elevation anomaly points. In the deleted sequence (P1, P2, …, P7000), point P1 corresponds to the original A1, point P2 corresponds to the original A2, and so on.

[0039] S22: Coordinate Transformation. The actual elevation is calculated as pjz = Pjz - 0.5. For example, the actual coordinates of point P1 are (100.0, 200.0, 10.0), and the actual coordinates of point P7000 are (150.0, 250.0, 10.2). The transformed sequence (p1, p2, …, p7000) is used for subsequent analysis.

[0040] Furthermore, referring to Figure 3 S3 also includes: S31, Set the elevation change threshold Hlimit1; S32, set the window size to W, indicating that the sliding window monitors the elevation changes of the latest W actual closing process data points; use the sliding window to monitor the elevation changes of the actual closing process data point sequence, the elevation change is ΔH1=p maxz -p minz In the formula, p maxz p minz These are the highest and lowest elevation points within the window, respectively. S33, if the elevation change ΔH within the current window exceeds the preset warehouse surface elevation change threshold H limit Then mark the last data point before the sliding window as the end point G of the previous working segment. bk The first data point within the sliding window is marked as the starting point Y of the next moving segment. al And mark the current working state as moving; S34, if the elevation change ΔH within the current window falls below the preset warehouse elevation change threshold H again. limit Then mark the last data point before the sliding window as the end point Y of the previous moving segment. bl The first data point within the sliding window is marked as the starting point G of the next working segment. ak And mark the current working status as working; S35, based on the starting point (G) of each working segment and moving segment a1 G a2 , …, G ak , …, G a(o-1) Y ao Y a1 ,Y a2 , …, Y aK , …, Y a(O-1) YaO ) and endpoint (G) b1 G b2 , …, G bk , …, G b(o-1) Y bo Y b1 Y b2 ,…, Y bK , …, Y b(O-1) Y bO The data is divided into different working segments and moving segments, i.e., G. ak To G bk The set of points between them is the kth working segment, Y aK To Y bK The set of points between them is the Kth moving segment, where o represents the number of working segments and O represents the number of moving segments.

[0041] In this embodiment, S31: Set the elevation change threshold Hlimit1 = 0.1 meters.

[0042] S32: Sliding window size W = 10. Calculate ΔH1 = pmaxz - pminz for each window. For example, for window points p1 to p10: pmaxz = 10.1 meters, pminz = 10.0 meters, ΔH1 = 0.1 meters; for window points p11 to p20: pmaxz = 10.25 meters, pminz = 10.1 meters, ΔH1 = 0.15 meters.

[0043] S33: Mark the segment point when ΔH1>Hlimit1. For example, at point p20, ΔH1=0.15 meters, mark p20 as Gb1 and p21 as Ya1, and the state moves.

[0044] S34: Mark the segment point when ΔH1≤Hlimit1. For example, at point p39, ΔH1=0.05 meters, mark p39 as Yb1 and p40 as Ga2, and the state is working.

[0045] S35: Segmentation Results: Working segment start points (Ga1, Ga2, Ga3, Ga4, Ga5) and end points (Gb1, Gb2, Gb3, Gb4, Gb5); moving segment start points (Ya1, Ya2, Ya3, Ya4) and end points (Yb1, Yb2, Yb3, Yb4). For example, working segment 1 runs from Ga1 to Gb1 and contains 20 points; moving segment 1 runs from Ya1 to Yb1 and contains 19 points. There are a total of 5 working segments and 4 moving segments.

[0046] Furthermore, in S4, the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data.

[0047] Furthermore, referring to Figure 4 S4 also includes: S41, merge all identified working segments; S42, extract time and elevation as features, and standardize using standard deviation; S43, data points within the working segment with a speed set to 0 and a direction change amplitude greater than a preset amplitude threshold are considered as nodes. Data points between two nodes are considered as a trajectory segment. The set of points T within this trajectory segment is used as a clustered data point, thereby generating a clustered dataset D = (T1, T2, …, T…). J , …, T M , T M ,), where M is the number of all trajectories within the working segment.

[0048] S44 uses the improved hausdorff-DBSCAN clustering algorithm to perform clustering, assigns cluster labels to the corresponding working segments, and marks the noise points in S43 as moving segments; S45. Generate a visualization view containing time-elevation map and plane trajectory map based on the clustering results. The working segment is a different billet layer, which is represented in different forms according to the clustering results. The moving segment is represented by a unified symbol.

[0049] In this embodiment, S41: merge all working segment data, totaling 3000 points.

[0050] S42: Extract time and elevation features. Time values ​​range from 0 to 7200 seconds, and elevation values ​​range from 10.0 to 10.5 meters. Standardization: The mean of the time feature is 1800 seconds, and the standard deviation is 600 seconds. For example, the standardization of time 0 seconds for point p1 is (0-1800) / 600=-3.0. The mean of the elevation feature is 10.2 meters, and the standard deviation is 0.1 meters. For example, the standardization of elevation 10.0 meters for point p1 is (10.0-10.2) / 0.1=-2.0.

[0051] S43: Trajectory Segmentation. Set a threshold of 30° for the magnitude of directional change. Mark nodes when the directional change exceeds 30°. The point set between nodes is the trajectory; for example, trajectory T1 contains points p101 to p120, and trajectory T2 contains points p121 to p140. A total of 50 trajectory segments are generated, and the clustered dataset D=(T1, T2, …, T50).

[0052] S44: Improved Hausdorff-DBSCAN clustering. Parameters eps=0.5 meters, minPts=3. Calculate the Hausdorff distance between trajectories: for example, T1 and T2 have a Hausdorff distance hJ=0.3 meters, so they are similar; T1 has 4 neighborhood points, which are the core points, so a cluster CL1 is created; after expansion, CL1 contains trajectories T1, T2, T5, and T6; CL2 contains T3, T4, and T7; CL3 contains T8 and T9; noisy trajectories T10, T11, etc. are marked as moving segments.

[0053] S45: Visualization. Time-Elevation Plot: X-axis is time, Y-axis is elevation; cluster CL1 uses red dots, CL2 uses blue dots, CL3 uses green dots, and moving segments use gray dots. Planar Trajectory Plot: X-axis is X coordinate, Y-axis is Y coordinate; cluster CL1 uses solid lines, CL2 uses dashed lines, CL3 uses dotted lines, and moving segments use gray dashed lines.

[0054] Furthermore, in S42, standard deviation standardization is used to remove the mean and normalize the variance of the extracted time and elevation data to improve clustering accuracy.

[0055] In this embodiment, the standardization formula is: Standardized value = (Original value - Mean) / Standard deviation. The time feature has a mean of 1800 seconds and a standard deviation of 600 seconds. For example, point p1 with a time of 0 seconds is standardized to -3.0, and point p3600 with a time of 3600 seconds is standardized to 3.0. The elevation feature has a mean of 10.2 meters and a standard deviation of 0.1 meters. For example, point p1 with an elevation of 10.0 meters is standardized to -2.0, and point p3600 with an elevation of 10.4 meters is standardized to 2.0. After standardization, the feature values ​​range from approximately -3 to 3, eliminating the influence of dimensions and improving clustering accuracy.

[0056] Furthermore, in S43, the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data by replacing the Euclidean distance in DBSCAN clustering with the hausdorff distance, which is suitable for trajectory clustering, as the similarity measure.

[0057] The improved hausdorff-DBSCAN clustering method is a density-based method; The parameter eps is specified as the neighborhood radius. The similarity and whether two data points belong to the same cluster are determined based on whether the hausdorf distance between two clustered data points is less than or equal to eps. Specify another parameter, minPts. When the number of points in the neighborhood of a cluster data point reaches or exceeds minPts, the point is considered a core point. Iterate through all points T in the clustering dataset D, if T J If it is not visited, mark it as visited and calculate T. JThe Hausdorff distance h to all other points within D J ; (1) In the formula, T x For any other clustered data point within D, p a p b Trajectory T J T x The actual closing process data points within, d(p) a , p b ) is p a p b The Euclidean distance between them; (2) In the formula, p is... a p b The coordinates are respectively (p ax , p ay , p az ) and (p bx , p by , p bz ); Calculate h J If the number of points ≤ eps is less than MinPts, then the point is temporarily marked as noise; if it reaches or exceeds MinPts, it is considered a core point, and a new cluster C is created with this point as the core. L and T J All points within the neighborhood are added to the Seed set to be processed as the initial seed. J ; The seed set is processed in a loop. Each seed point is taken out in turn. If it has not been visited, its neighborhood is visited and its neighborhood is calculated. If the number of its neighborhood points also reaches or exceeds MinPts, the new points in its neighborhood are added to the seed set. Regardless of whether a seed point is a core point, as long as it has not yet belonged to any cluster, it is added to the current cluster. This loop continues until the seed set is empty, completing cluster C. L The expansion of the cluster is such that points temporarily marked as noise are added as boundary points if they are connected. After processing all points, the points that are not assigned to any cluster are considered noise points.

[0058] The improvement of the hausdorff-DBSCAN clustering method depends on the values ​​of the two parameters eps and minPts. Different parameter values ​​will lead to significant changes in the clustering effect. The optimal values ​​are determined based on the actual closing process and multiple trials.

[0059] In this embodiment, the parameters are set as follows: eps = 0.5 meters and minPts = 3.

[0060] Hausdorff distance calculation: For trajectories TJ and Tx, the Hausdorff distance hJ = max( h(TJ, Tx), h(Tx, TJ) ), where h(TJ, Tx) = max_{pa in TJ} min_{pb in Tx} d(pa, pb), and d(pa, pb) is the Euclidean distance. For example, trajectories T1 and T2: coordinates of point p101 in T1 (110.0, 210.0, 10.1), coordinates of point p121 in T2 (110.2, 210.1, 10.1), d(p101, p121)=√((110.0-110.2)²+(210.0-210.1)²+(10.1-10.1)²)=0.224 meters; after calculating all point pairs, h(T1, T2)=0.3 meters, h(T2, T1)=0.3 meters, therefore hJ=0.3 meters.

[0061] DBSCAN Steps: Starting with trajectory T1, mark it as visited. Calculate the Hausdorff distance to all other trajectories. Find trajectories T2, T5, and T6 with a distance ≤ eps. Therefore, T1 is the core point, and a cluster CL1 is created with a seed set Seed1 = {T1, T2, T5, T6}. Process T2: Unvisited. After visiting, calculate its neighborhood, finding T1, T5, T6, and T3. However, T3 has only 2 neighborhood points, so it is not expanded, but T2 is added to CL1. Repeat until Seed1 is empty, and CL1 contains T1, T2, T5, and T6. Process other points similarly, forming clusters CL2 and CL3. Noisy trajectories such as T10 are not included in any cluster.

[0062] Parameter determination: After multiple trials, it was found that there were too many clusters when eps=0.3 meters and too few clusters when eps=0.7 meters. Finally, eps=0.5 meters was selected. There was too much noise when minPts=2 and too few clusters when minPts=4. Finally, minPts=3 was selected.

[0063] Furthermore, referring to Figure 5 S5 also includes: S51, Set elevation deviation threshold H limit2 ; S52, if the current working state is moving, no judgment is made; if the current working state is working, the deviation value ΔH2=Cz-Hz between the current working billet layer cluster center point elevation and the design center elevation of the billet layer is calculated, where C is the cluster closest to the current time, i.e. the current working billet layer, Cz is the average elevation within the trajectory of the cluster center point TC of the cluster, and Hz is the design center elevation of the billet layer. S53, when the deviation value ΔH2 is greater than the set threshold Hlimit2, an alarm is issued.

[0064] In this embodiment, S51: Set the elevation deviation threshold Hlimit2 = 0.05 meters.

[0065] S52: Status Judgment: When the status is moving, no judgment is made; when the status is working, calculate the elevation of the cluster center point of the current cluster CL3 as Cz=10.08 meters, the design center elevation as Hz=10.0 meters, and the deviation as ΔH2=0.08 meters.

[0066] S53: ΔH2=0.08m>Hlimit2=0.05m, alarm issued: "Blank layer thickness deviation 0.08m, exceeding the threshold of 0.05m, please check the closing operation."

[0067] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0068] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0069] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for real-time monitoring of the thickness of the concrete layer during the leveling process based on improved Hausdorff-DBSCAN clustering, characterized in that, Includes the following steps: S1, real-time acquisition of initial process data of the concrete leveling to be monitored; S2, preprocess the collected initial process data, and convert the preprocessed initial process data into actual leveling process data based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface. S3, perform elevation change characteristic analysis on the actual closing process data, and divide the actual closing process data into working segment and moving segment; S4. Merge all working segment data, and use the improved hausdorff-DBSCAN clustering method to cluster the trajectories within the working segment. Different clusters are used to distinguish different billet layers, and the time-elevation map and plane trajectory map are displayed through visualization. S5. Compare the elevation of the cluster center point of the current working blank layer with the design center elevation of the current working blank layer. If the elevation deviation is greater than the set threshold, an alarm message is issued.

2. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering as described in claim 1, characterized in that, S1 also includes: Real-time positioning data of the current billet construction machinery is obtained through the surface construction monitoring equipment: Equipment number, time, XY plane position, elevation, speed, direction, number of positioning satellites; During the start-up of construction machinery, the surface construction monitoring equipment automatically records one data point per second of the leveling process, forming a sequence of leveling process data points (A1, A2, …, A…). i , …, A m-1 A m ), where i=1,2…(m 1), m, where m represents the number of data points.

3. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering as described in claim 1, characterized in that, S2 also includes: S21, preprocess the initial process data of the concrete leveling to be monitored collected in real time, deleting data with a rate and direction of 0, a number of positioning satellites below the normal value, and abnormal elevation values, forming a preprocessed sequence of leveling process data points (P1, P2, …, P…). j , …, P n-1 , P n ), where j=1,2…(n 1), n, where n represents the number of data points after preprocessing; S22, Based on the relationship between the coordinates of the construction monitoring equipment and the elevation of the concrete surface, the preprocessed data point sequence of the leveling process is converted into the actual leveling process data point sequence (p1, p2, …, p…). j , …, p n-1 , p n ), that is, the coordinates of the j-th point in the preprocessed closing process data point sequence are (P jx , P jy , P jz The coordinates of point j in the actual closing process data point sequence are (p jx , p jy , p jz ), where p jx =P jx p jy =P jy p jx =f(P jx ).

4. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering as described in claim 1, characterized in that, S3 also includes: S31, Set elevation change threshold H limit1 ; S32, set the window size to W, indicating that the sliding window monitors the elevation changes of the latest W actual closing process data points; use the sliding window to monitor the elevation changes of the actual closing process data point sequence, the elevation change is ΔH1=p maxz -p minz In the formula, p maxz p minz These are the highest and lowest elevation points within the window, respectively. S33, if the elevation change ΔH within the current window exceeds the preset warehouse surface elevation change threshold H limit Then mark the last data point before the sliding window as the end point G of the previous working segment. bk The first data point within the sliding window is marked as the starting point Y of the next moving segment. al And mark the current working state as moving; S34, if the elevation change ΔH within the current window falls below the preset warehouse elevation change threshold H again. limit Then mark the last data point before the sliding window as the end point Y of the previous moving segment. bl The first data point within the sliding window is marked as the starting point G of the next working segment. ak And mark the current working status as working; S35, based on the starting point (G) of each working segment and moving segment a1 G a2 , …, G ak , …, G a(o-1) Y ao Y a1 Y a2 ,…, Y aK , …, Y a(O-1) Y aO ) and endpoint (G) b1 G b2 , …, G bk , …, G b(o-1) Y bo Y b1 Y b2 , …,Y bK , …, Y b(O-1) Y bO The data is divided into different working segments and moving segments, i.e., G. ak To G bk The set of points between them is the kth working segment, Y aK To Y bK The set of points between them is the Kth moving segment, where o represents the number of working segments and O represents the number of moving segments.

5. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering according to claim 1, characterized in that, In S4, the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data.

6. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering according to claim 1, characterized in that, S4 also includes: S41, merge all identified working segments; S42, extract time and elevation as features, and standardize using standard deviation; S43, data points within the working segment with a speed set to 0 and a direction change amplitude greater than a preset amplitude threshold are considered as nodes. Data points between two nodes are considered as a trajectory segment. The set of points T within this trajectory segment is used as a clustered data point, thereby generating a clustered dataset D = (T1, T2, …, T…). J , …, T M , T M ,), where M is the number of all trajectories within the working segment; S44 uses the improved hausdorff-DBSCAN clustering algorithm to perform clustering, assigns cluster labels to the corresponding working segments, and marks the noise points in S43 as moving segments; S45. Generate a visualization view containing time-elevation map and plane trajectory map based on the clustering results. The working segment is a different billet layer, which is represented in different forms according to the clustering results. The moving segment is represented by a unified symbol.

7. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering according to claim 6, characterized in that, In S42, standard deviation standardization is used to remove the mean and normalize the variance of the extracted time and elevation data to improve clustering accuracy.

8. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering according to claim 6, characterized in that, In S43, the hausdorff distance, which is suitable for trajectory clustering, is used instead of the Euclidean distance in DBSCAN clustering as a similarity measure, and the improved hausdorff-DBSCAN clustering method is used to cluster the working segment data. The improved hausdorff-DBSCAN clustering method is a density-based method; The parameter eps is specified as the neighborhood radius. The similarity and whether two data points belong to the same cluster are determined based on whether the hausdorf distance between two clustered data points is less than or equal to eps. Specify another parameter, minPts. When the number of points in the neighborhood of a cluster data point reaches or exceeds minPts, the point is considered a core point. Iterate through all points T in the clustering dataset D. If T J If it is not visited, mark it as visited and calculate T. J The Hausdorff distance h to all other points within D J ; （1） In the formula, T x For any other clustered data point within D, p a p b Trajectory T J T x The actual closing process data points within, d(p) a , p b ) is p a p b The Euclidean distance between them; （2） In the formula, p a p b The coordinates are respectively (p ax , p ay , p az ) and (p bx , p by , p bz ); Calculate h J If the number of points ≤ eps is less than MinPts, then the point is temporarily marked as noise; if it reaches or exceeds MinPts, it is considered a core point, and a new cluster C is created with this point as the core. L and T J All points within the neighborhood are added to the Seed set to be processed as the initial seed. J ; The seed set is processed in a loop. Each seed point is taken out in turn. If it has not been visited, its neighborhood is visited and its neighborhood is calculated. If the number of its neighborhood points also reaches or exceeds MinPts, the new points in its neighborhood are added to the seed set. Regardless of whether a seed point is a core point, as long as it has not yet belonged to any cluster, it is added to the current cluster. This loop continues until the seed set is empty, completing cluster C. L The expansion of the cluster is such that points temporarily marked as noise are added as boundary points if they are connected. After processing all points, the points that are not assigned to any cluster are considered noise points. The improvement of the hausdorff-DBSCAN clustering method depends on the values ​​of the two parameters eps and minPts. Different parameter values ​​will lead to significant changes in the clustering effect. The optimal values ​​are determined based on the actual closing process and multiple trials.

9. The method for real-time monitoring of concrete layer thickness during leveling based on improved Hausdorff-DBSCAN clustering according to claim 1, characterized in that, S5 also includes: S51, Set elevation deviation threshold H limit2 ; S52, if the current working state is moving, no judgment is made; If the current working status is "working", then calculate the deviation value ΔH2=C between the current working layer cluster center point elevation and the design center elevation of the layer. z -H z In the formula, C represents the cluster closest to the current time, i.e., the current working layer. z Let T be the cluster center point of this cluster. C Average elevation within the trajectory, H z The design center elevation of this billet layer; S53, when the deviation value ΔH2 is greater than the set threshold H limit2 An alarm will be triggered at that time.

Images