Intelligent measurement method and device for complex environment ship lock foundation pit

Abstract

The application provides a complex environment ship lock foundation pit intelligent measurement method and device, relates to the image processing and engineering measurement technical field, and the registration weight is determined according to the laser reflection intensity of each point in the target point cloud set, the weighted distance is constructed by introducing the weight when establishing the point pair corresponding relation, and the registration process is preferentially dependent on the high-quality point with obvious structural features;Based on the weighted corresponding relationship, the optimal space transformation parameter is solved and the displacement field is calculated, further, the displacement meter monitoring data is used as an external reference, the displacement field is calibrated by calculating the deviation and constructing the system deviation field, the cumulative error in the registration process is eliminated, the technical problems that the low reflection intensity area interference registration accuracy under the complex environment and the lack of external reference lead to the systematic drift of the displacement field are effectively solved, and the accuracy and reliability of the deformation area identification are improved.

Description

Technical Field

[0001] This invention relates to the fields of image processing and engineering measurement technology, specifically to an intelligent measurement method and device for lock foundation pits in complex environments. Background Technology

[0002] As a key structure in waterway engineering, the construction of lock pits typically faces complex hydrogeological conditions and high standards of engineering safety. During the excavation and support of the pit, the stability of the slope directly affects construction safety and project progress. Traditional pit measurement methods mainly rely on manual measurement tools such as total stations and levels, which obtain discrete displacement data by periodically observing the slope through the establishment of monitoring points. However, this point-based measurement method has obvious limitations. On the one hand, it is difficult to fully reflect the overall deformation of the slope surface. On the other hand, when encountering complex environments such as frequent water level changes, water accumulation on the slope surface, or mud cover, manual measurement operations themselves face significant safety risks. With the popularization of 3D laser scanning technology, reconstructing the 3D morphology of foundation pits by acquiring high-density point cloud data, and then calculating displacement fields and deformation areas, has become an important development direction in the field of foundation pit monitoring. Ship lock foundation pits are usually located along riverbanks. During construction, multiple environmental factors such as upstream water flow, groundwater level changes, and wave impacts from ship passage need to be addressed simultaneously. These factors often result in low-reflection areas such as local water accumulation, mud covering, and vegetation attachment on the surface of the foundation pit slope, posing a severe challenge to the measurement accuracy based on laser scanning.

[0003] In existing technologies, methods for monitoring foundation pit deformation based on laser point clouds typically employ an iterative nearest-point algorithm to register point clouds from two adjacent periods. The displacement vector after registration is then used to obtain slope deformation information. This type of method achieves relatively ideal registration results when processing point cloud data with uniform surface reflectivity acquired under ideal conditions. However, in actual lock pit environments, due to the presence of low-reflectivity areas on the slope surface, the point cloud data from these areas often contains significant noise and positioning errors. When using the traditional Euclidean nearest-point search strategy, these low-quality points will participate in the registration process. The process leads to a decrease in overall registration accuracy. Some improvement schemes attempt to improve registration quality by pre-removing low-reflection-intensity points, but simple removal strategies result in data loss, causing gaps in deformation analysis results. More importantly, existing technologies lack effective calibration methods for systematic errors in point cloud registration. Point cloud registration results rely entirely on the algorithm's own optimization process, lacking external benchmark verification and correction. When cumulative errors exist in the registration process, the displacement field calculation results will experience systematic drift, leading to deviations or even misjudgments in the subsequent identification of deformation areas.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent measurement method and device for lock pits in complex environments, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A smart measurement method for lock pits in complex environments, comprising the following steps: Step 1: Obtain the reference point cloud of the lock pit slope at the previous moment and the target point cloud at the current moment. Each point in the point cloud contains information on its three-dimensional coordinates and laser reflection intensity. Determine the registration weight of each target point in the target point cloud based on its laser reflection intensity. Step 2: For each reference point in the reference point cloud, based on the weighted distance calculation with registration weights, determine the target point with the smallest weighted distance as its corresponding point to establish the correspondence between point pairs; Step 3: Construct an objective function based on the correspondence and registration weights, solve the objective function using the singular value decomposition method to obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, the displacement vector of each reference point is its transformed coordinates minus its original coordinates, decompose the displacement vector of each reference point into displacement feature vectors including horizontal and vertical displacements to construct the displacement field; Step 4: For each displacement gauge installed on the slope of the foundation pit, extract the displacement characteristic vector at its installation location from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. Analyze the deviation vector at each displacement gauge using the Kriging space interpolation method to construct a systematic deviation field covering the entire area. Subtract the systematic deviation field from the displacement field to obtain the calibrated displacement field. Analyze the displacement value and displacement rate in the calibrated displacement field to determine the deformation area.

[0007] Furthermore, the set of reference points for the lock pit slope at the previous moment and the set of target points at the current moment are obtained, specifically including: Control points were set up within a predetermined area around the slope of the lock pit. The geodetic coordinates of the control points were obtained by static measurement using a global navigation satellite system. A drone equipped with a lidar scanning system was used to perform a full-coverage scan of the lock pit along a set route, acquiring laser point cloud data at the previous and current times. Each point in the laser point cloud data contains three-dimensional coordinates in the laser point cloud coordinate system and laser reflection intensity information. Using the geodetic coordinates of the control points and the three-dimensional coordinates of their corresponding points in the laser point cloud coordinate system, a seven-parameter transformation model was solved using the least squares method to transform the laser point cloud data at the two times to the same geodetic coordinate system, which were then used as the reference point cloud set and the target point cloud set, respectively. Based on the laser reflection intensity of each target point in the target point cloud, the registration weight of each target point is determined, specifically including: The laser reflection intensity of all target points in the target point cloud is obtained, and the maximum reflection intensity value is determined. Based on the positive correlation between laser reflection intensity and registration weight, for each target point in the target point cloud, the registration weight of the target point is calculated through nonlinear mapping based on the difference between the laser reflection intensity of the target point and the maximum reflection intensity value, as well as the preset adjustment coefficient, so that the target point with higher laser reflection intensity gets a larger registration weight.

[0008] Furthermore, the logic for establishing a point-to-point correspondence is as follows: For each reference point in the reference point cloud, calculate the Euclidean distance between the reference point and each target point in the target point cloud. Divide the calculated Euclidean distance by the registration weight of the corresponding target point in the target point cloud to obtain the weighted distance between the reference point and the target point. Select the target point in the target point cloud that has the smallest weighted distance to the reference point as the corresponding point of the reference point, thereby establishing a point-to-point correspondence between the reference point and the target point, which is called a corresponding point pair. The method for determining the optimal rotation matrix and translation vector is as follows: Based on the established correspondence between reference points and target points, an objective function is constructed with rotation matrix and translation vector as independent variables. The value of the objective function is calculated as follows: For each pair of corresponding points, the reference point in the pair is spatially transformed using rotation matrix and translation vector to obtain the coordinates of the transformed reference point. The spatial distance between the transformed reference point coordinates and the corresponding target point coordinates is then calculated, and the square of this spatial distance is calculated. The squared value is then multiplied by the registration weight of the target point in the pair to obtain the weighted squared distance of the pair. The sum of the weighted squared distances of all pairs of corresponding points is the value of the objective function. The rotation matrix and translation vector that minimize the objective function are found using singular value decomposition. The obtained rotation matrix and translation vector are taken as the optimal rotation matrix and translation vector. The optimal rotation matrix and translation vector are applied to each reference point in the reference point cloud to obtain the transformed reference point cloud.

[0009] Furthermore, the displacement vector is decomposed into displacement eigenvectors of horizontal and vertical displacements to obtain the displacement field, which specifically includes: For each reference point in the reference point cloud, its coordinates after optimal rotation matrix and translation vector transformation are subtracted from its original coordinates before transformation to obtain the displacement vector of that reference point. Two horizontal components are extracted from the displacement vector, and the square root of the sum of the squares of these two components is taken to obtain the horizontal displacement characteristic vector of that reference point, defined as the horizontal displacement. The vertical component of the displacement vector is extracted as the vertical displacement characteristic vector of that reference point, defined as the vertical displacement. The horizontal and vertical displacements of all reference points are organized according to their spatial positions to form a displacement field reflecting the magnitude and direction of the deformation of the lock pit slope.

[0010] Furthermore, the system deviation field is obtained through spatial interpolation, and the system deviation field is subtracted from the displacement field to obtain the calibrated displacement field, which specifically includes: The monitoring data of multiple displacement gauges installed at predetermined locations on the slope of the lock pit are obtained at the current moment. The monitoring data consists of the measured vectors of each displacement gauge's installation location from the previous moment to the current moment. The measured vector of each displacement gauge includes the actual displacement in the horizontal direction and the actual displacement in the vertical direction. For each displacement gauge, the point cloud displacement value at the corresponding three-dimensional coordinates in the displacement field is extracted from the displacement field. The point cloud displacement value includes the horizontal displacement and the vertical displacement. The difference between the point cloud displacement value and the actual displacement of each displacement gauge installation location is calculated to obtain the system deviation value in the horizontal direction and the system deviation value in the vertical direction at that location. Based on the horizontal systematic deviation values ​​at all displacement gauge locations, the Kriging space interpolation method is used to calculate the horizontal systematic deviation values ​​at any three-dimensional coordinate of the lock pit slope, forming a horizontal systematic deviation field. Similarly, based on the vertical systematic deviation values ​​at all displacement gauge locations, the Kriging space interpolation method is used to form a vertical systematic deviation field. The horizontal and vertical systematic deviation fields together constitute a continuous systematic deviation field covering the entire displacement field region. For each reference point in the displacement field, according to the three-dimensional coordinates of the reference point, the horizontal and vertical systematic deviation values ​​at the corresponding locations are obtained from the horizontal and vertical systematic deviation fields, respectively. The horizontal displacement of the reference point is subtracted from the horizontal systematic deviation value, and the vertical displacement is subtracted from the vertical systematic deviation value to obtain the calibrated horizontal and vertical displacement values ​​of the reference point. The calibrated horizontal and vertical displacement values ​​of all reference points are organized according to the three-dimensional coordinates of each reference point to form a calibrated displacement field.

[0011] Furthermore, the regions in the calibrated displacement field where the displacement value and displacement rate exceed a preset displacement threshold are defined as deformation regions, specifically including: A displacement rate threshold and a cumulative displacement threshold are set. The displacement rate threshold is used to evaluate the short-term deformation rate of the lock pit slope, and the cumulative displacement threshold is used to evaluate the total long-term deformation of the lock pit slope. For each reference point in the calibrated displacement field, the calibrated horizontal displacement and calibrated vertical displacement carried by that reference point from the previous moment to the current moment are obtained. The calibrated horizontal displacement is divided by the time interval between the previous moment and the current moment to obtain the horizontal displacement rate of the reference point. The calibrated vertical displacement is divided by the time interval to obtain the vertical displacement rate of the reference point. At the same time, from the initial monitoring moment to the current moment, the calibrated horizontal displacement of the reference point in each time interval is accumulated to obtain the cumulative horizontal displacement of the reference point. The calibrated vertical displacement of the reference point in each time interval is accumulated to obtain the cumulative vertical displacement of the reference point. If the horizontal displacement rate of a reference point exceeds the horizontal displacement rate threshold, or the vertical displacement rate exceeds the vertical displacement rate threshold, or the cumulative horizontal displacement exceeds the cumulative horizontal displacement threshold, or the cumulative vertical displacement exceeds the cumulative vertical displacement threshold, then the reference point is marked as a deformation point; the continuous spatial region formed by all adjacent deformation points is marked as a deformation region, and the spatial location, geometric range, and the horizontal displacement, vertical displacement, and resultant displacement of each reference point within the region are output. The resultant displacement is calculated by taking the square root of the sum of the squares of the horizontal and vertical displacements.

[0012] Furthermore, obtaining the reference point cloud set and the target point cloud set also includes preprocessing the raw point cloud data: Using a terrestrial 3D laser scanner or an unmanned aerial vehicle (UAV) oblique photography system, the slope of the lock pit is scanned at fixed time intervals to obtain the original laser point cloud data of the previous and current moments. The acquired original laser point cloud data is then denoised and filtered to remove outliers and noise points. The point clouds from each period are then uniformly registered to the same geodetic coordinate system through coordinate transformation to ensure the spatial consistency of the point cloud data at different times.

[0013] The present invention also provides an intelligent measurement device for complex environment lock pits, the device being used to perform the above-described intelligent measurement method for complex environment lock pits, comprising: The point cloud acquisition module is used to acquire the reference point cloud set of the lock pit slope at the previous moment and the target point cloud set at the current moment. Each point in the point cloud set contains information on its three-dimensional coordinates and laser reflection intensity. The registration weight is determined based on the laser reflection intensity of each target point in the target point cloud set. The relationship registration module is used to determine the target point with the smallest weighted distance as its corresponding point for each reference point in the reference point cloud based on the weighted distance calculation with introduced registration weights, so as to establish the correspondence between point pairs. The data calculation module is used to construct an objective function based on the correspondence and registration weights, solve the objective function by singular value decomposition, obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, and the displacement vector of each reference point is its transformed coordinates minus its original coordinates. The displacement vector of each reference point is decomposed into displacement feature vectors including horizontal and vertical displacements to construct the displacement field. The calibration module is used to extract the displacement feature vector at the installation location of each displacement gauge deployed on the slope of the foundation pit from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. The deviation vector at each displacement gauge is analyzed by Kriging space interpolation to construct a systematic deviation field covering the entire area. The systematic deviation field is then subtracted from the displacement field to obtain the calibrated displacement field. The displacement value and displacement rate in the calibrated displacement field are analyzed to determine the deformation area.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention determines the registration weights based on the laser reflection intensity of each point in the target point cloud, giving smaller weight values ​​to points with lower laser reflection intensity. These low-reflection-intensity points correspond to areas of water accumulation, mud cover, or vegetation on the slope surface, and their point cloud data inherently contains significant noise and positioning errors. By assigning them lower weights, the interference of low-quality data on the subsequent registration process is suppressed from the source. Simultaneously, a weighted distance based on weights is introduced when establishing point pair correspondences, and the determined weight values ​​are directly applied to the nearest neighbor search process, allowing feature points with high reflection intensity to dominate the correspondence establishment process. This weighting mechanism can effectively avoid mismatching noisy points as feature points, thereby improving the initial correspondence quality of point cloud registration. This invention also obtains the optimal spatial transformation parameters by solving the objective function. In this process, the high-weight points contribute more to the optimization influence, making the registration results more biased towards areas with obvious structural features and reliable data quality, thus ensuring the accuracy of the displacement field calculation as a whole. At the same time, by calculating the deviation between the point cloud displacement at the displacement gauge position and the actual monitoring value of the displacement gauge, each displacement gauge is regarded as a precise constraint point in the local area. Using these discrete precise constraint points, a global deviation correction surface is formed through spatial interpolation. Finally, the system deviation value at the corresponding position is subtracted from each point in the displacement field, thereby realizing the organic integration of point cloud displacement data and physical measured data. This invention introduces two judgment indicators simultaneously in the deformation area identification process: displacement rate threshold and cumulative displacement threshold. The displacement rate threshold is used to capture rapid deformation of the slope in a short period of time, while the cumulative displacement threshold is used to assess the total amount of deformation accumulated over a long period of time. By judging the two indicators together, the invention effectively distinguishes between short-term environmental disturbances and long-term trend deformation, avoids false alarms caused by fluctuations in a single measurement, and improves the accuracy and engineering practicality of deformation area identification. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a curve showing the fitting of laser reflection intensity and observation registration weights in this invention. Figure 3 This is a normal probability diagram of the registration weights observed in this invention; Figure 4 This is a flowchart of the overall device structure of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] Example: Please see Figures 1-3 The present invention provides a technical solution: A smart measurement method for lock pits in complex environments, comprising the following steps: Step 1: Obtain the reference point cloud of the lock pit slope at the previous moment and the target point cloud at the current moment. Each point in the point cloud contains information on its three-dimensional coordinates and laser reflection intensity. Determine the registration weight of each target point in the target point cloud based on its laser reflection intensity. In a specific embodiment of the present invention, considering that the surface materials of the lock pit slope are diverse, including exposed rock, concrete support structure, vegetation cover area, damp soil and temporary water accumulation area, different materials have different reflectivity to lidar scanning signals. Point cloud data with high reflectivity corresponds to areas with obvious structural features and stable surfaces, while point cloud data with low reflectivity is easily affected by noise or has insignificant geometric features. If all points are assigned the same weight during the point cloud registration process, low reflectivity points will reduce the registration accuracy, thereby affecting the accuracy of subsequent displacement field calculation. Therefore, it is necessary to assign different registration weights to each target point according to the laser reflection intensity, so that high reflectivity points play a dominant role in the registration. Obtain the set of reference points for the lock pit slope at the previous moment and the set of target points at the current moment, specifically including: Control points were set up within a predetermined area around the slope of the lock pit. The geodetic coordinates of the control points were obtained by static measurement using a global navigation satellite system. A drone equipped with a lidar scanning system was used to perform a full-coverage scan of the lock pit along a set route, acquiring laser point cloud data at the previous and current times. Each point in the laser point cloud data contains three-dimensional coordinates in the laser point cloud coordinate system and laser reflection intensity information. Using the geodetic coordinates of the control points and the three-dimensional coordinates of their corresponding points in the laser point cloud coordinate system, a seven-parameter transformation model was solved using the least squares method to transform the laser point cloud data at the two times to the same geodetic coordinate system, which were then used as the reference point cloud set and the target point cloud set, respectively. To obtain the reference point set and target point set of the lock pit slope at the previous moment, specifically, a geologically stable area at least 10 meters away from the edge of the pit and free from construction machinery disturbance is pre-selected around the lock pit slope. Permanent control points are buried in this area. The control points are in the form of forced centering observation piers or metal targets to facilitate subsequent scanning and identification. A high-precision global navigation satellite system receiver is used to conduct static observation of each control point for at least two hours. The precise three-dimensional coordinates of each control point in the CGCS2000 geodetic coordinate system are obtained through post-differential calculation. The plane positioning accuracy is better than ±5 mm and the elevation accuracy is better than ±10 mm. A drone equipped with a 3D lidar scanning system was used to plan a flight path covering the entire foundation pit and surrounding slopes. The flight altitude was set at 50 to 80 meters, and the scanning speed was more than 500,000 points per second. The lock foundation pit was fully scanned along the same flight path at the previous and current moments. In the raw laser point cloud data obtained from each scan, each point contains two aspects of information: first, the 3D coordinates in the laser scanner's own coordinate system, which are obtained from the laser ranging value, scanning angle, and calculation; second, the laser reflection intensity of the point, which is a dimensionless value reflecting the intensity of the echo energy of the laser beam on the surface of the scanned object. It is related to the target material, surface roughness, and incident angle. To facilitate subsequent coordinate transformation, target-type reflective sheets were placed at each control point position before each scan, so that the control points appear as identifiable points with high reflectivity in the point cloud data. From the raw laser point cloud data of the previous and current times, the three-dimensional coordinates of the target points corresponding to the control points are identified. Using the geodetic coordinates of at least three control points and their corresponding coordinates in the laser point cloud coordinate system, a seven-parameter transformation model is solved using the least squares method. The seven-parameter transformation model includes three translations, three rotation angles, and one scale factor. The solution process is as follows: establish a linearized error equation between the geodetic coordinates and the point cloud coordinates, and iteratively calculate the seven-parameter values ​​that minimize the sum of squared transformation residuals of all control points. Apply the obtained seven parameters to the entire point cloud data at the corresponding time, and transform the three-dimensional coordinates of each point from the laser point cloud coordinate system to the CGCS2000 geodetic coordinate system. After the transformation, the point cloud set of the previous time is called the reference point cloud set, and the point cloud set of the current time is called the target point cloud set. The reference point cloud set and the target point cloud set are located in the same geodetic coordinate system, which ensures the consistency of the reference for subsequent spatial registration. The seven-parameter transformation model is a mathematical method for three-dimensional coordinate transformation, which includes three translation parameters, three rotation parameters, and one scale parameter. The translation parameters represent the offset between the origins of the two coordinate systems, with the same units as the coordinate units; the rotation parameters represent the angular deviation between the coordinate axes, with the units being radians or degrees; and the scale parameter represents the proportional difference in unit length between the two coordinate systems, which is close to 1. These seven parameters are solved by the least squares method to minimize the sum of squares of the transformation residuals of all control points. Control points are permanent targets buried within a predetermined area around the foundation pit slope. Their geodetic coordinates are obtained by static measurement using the Global Navigation Satellite System. The role of control points is to serve as a reference for coordinate transformation, converting the laser point cloud data from the scanner's own coordinate system to a unified geodetic coordinate system. There should be at least three control points, and they should not be distributed on the same straight line to ensure that the solution of the seven-parameter transformation model has sufficient geometric strength. Laser reflection intensity is a dimensionless value obtained by normalizing the energy of the echo signal received by the lidar. It ranges from an integer between 0 and 255, or a decimal between 0 and 1. A higher reflection intensity value indicates a stronger ability of the measured object's surface to reflect the laser beam, corresponding to surfaces such as exposed rock, concrete structures, and metal components. A lower reflection intensity value indicates a weaker reflection ability, corresponding to surfaces such as vegetation, water surfaces, damp soil, and black rubber. The reflection intensity value is affected by various factors such as the target material, surface roughness, incident angle, and measurement distance. However, under the same measurement station and similar distance conditions, the reflection intensity values ​​are comparable. Based on the laser reflection intensity of each target point in the target point cloud, the registration weight of each target point is determined, specifically including: The laser reflection intensity of all target points in the target point cloud is obtained, and the maximum reflection intensity value is determined. Based on the positive correlation between laser reflection intensity and registration weight, for each target point in the target point cloud, the registration weight of the target point is calculated through nonlinear mapping based on the difference between the laser reflection intensity of the target point and the maximum reflection intensity value, as well as the preset adjustment coefficient, so that the target point with higher laser reflection intensity gets a larger registration weight. Traverse the laser reflection intensity of all target points in the target point cloud, find the maximum value, and record it as the maximum reflection intensity value. The calculation principle for the registration weight is set as follows: the higher the laser reflection intensity of the target point, the greater its registration weight; the lower the laser reflection intensity of the target point, the smaller its registration weight. For each target point in the target point cloud, let its laser reflection intensity be , and use the following nonlinear mapping function to calculate its registration weight: the weight is equal to 1 divided by 1 plus an adjustment coefficient multiplied by the difference between the maximum reflection intensity value and the laser reflection intensity of that point. The adjustment coefficient ranges from 0.01 to 0.1, and the specific value is calibrated according to the overall distribution range of the reflection intensity in the target point cloud. The formula for calculating the registration weight is: in, This represents the registration weight of the i-th target point in the target point cloud, with a value range of... ,when hour, ;when much smaller hour, The value approaches 0, where i is the index of a target point in the target point cloud, ranging from 1 to M, and M is the total number of target points in the target point cloud. The laser reflection intensity at the target point is a dimensionless value, directly output by the lidar scanning system. Different devices have different output ranges, commonly 0~255 (8-bit) or 0~65535 (16-bit). This value reflects the intensity of the laser beam's echo energy from the surface of the object being measured. A higher value indicates stronger reflection, corresponding to a harder material and a drier surface. This represents the maximum value of the laser reflection intensity of all target points in the target point cloud, i.e. This is obtained by traversing the entire target point cloud and is used to convert differences in reflection intensity into relative differences. The introduction of this feature makes the weight calculation insensitive to the device range and gives it good versatility. The preset adjustment coefficient ranges from 0.01 to 0.1. The physical meaning of this formula is: when the reflection intensity of the target point is equal to the maximum value, the weight is 1; when the reflection intensity is much less than the maximum value, the weight approaches 0. The value of determines the sensitivity of the weight to differences in reflection intensity. The larger the value, the faster the weight decays as the reflection intensity decreases; in this invention; The statistical maximum value comes directly from the target point cluster and does not require manual setting. The value needs to be calibrated based on the reflection intensity distribution of the actual point cloud data to ensure that the weights can effectively distinguish between high-quality points and low-quality points. This invention provides the following calibration method: randomly select 1000 points from the target point cloud set and calculate the standard deviation of the reflection intensity of these points. Then set ,like Falling Then take ;like Then take ;like Then take The physical meaning of this calibration method is to ensure that the weight of points with reflection intensity below two standard deviations of the mean is less than 0.2, thereby suppressing the impact of these low-quality points on subsequent registration to an acceptable level. The specific data of some target point numbers and registration weights are shown in Table 1.

[0019] Table 1 Data Statistics Table Analysis of the first fifteen sets of data reveals a clear positive correlation between laser reflection intensity and registration weight. As laser reflection intensity increases, so does the registration weight. For example, when the reflection intensity is 55, the observation weight is only 0.084; however, when the reflection intensity rises to 234, the observation weight reaches 0.495. This indicates that target points with higher reflection intensity have greater registration weights and play a more significant role in subsequent registration processes. Further observation of the relationship between reflection intensity and weight reveals that this positive correlation is not linear, but rather exhibits a gradual saturation trend. Specifically, in the low reflection intensity region, as the reflection intensity increases from 55 to 78, the observation weight rises from 0.084 to 0.106, showing a significant increase. In the high reflection intensity region, as the reflection intensity increases from 234 to 251, the observation weight rises from 0.495 to 0.841. Although the absolute increase is larger, the relative growth rate has slowed down. This reflects the nonlinear characteristic of the denominator in the registration weight calculation formula decreasing with increasing reflection intensity. Furthermore, comparing the theoretical weights and the observed weights reveals that the actual data exhibits slight random fluctuations around the theoretical values, but the overall trend is highly consistent with theoretical expectations. For example, when the reflection intensity is 251, the theoretical weight is 0.833 and the observed weight is 0.841, which are very close. When the reflection intensity is 55, the theoretical weight is 0.091 and the observed weight is 0.084, showing only a slight deviation. This fluctuation is consistent with the unavoidable random errors in actual measurements, but it does not change the fundamental mathematical relationship between reflection intensity and weight. Therefore, the registration weight calculation formula can fit the actual observed data well, providing a reliable basis for subsequent weighted distance calculation and spatial transformation parameter solution. When the laser reflection intensity of a target point is close to the maximum reflection intensity value, the weight approaches 1; when the laser reflection intensity of the target point is much less than the maximum reflection intensity value, the weight approaches 0. This calculation method gives high reflection intensity areas such as exposed rock and concrete structures a larger registration weight, while low reflection intensity areas such as vegetation, water surfaces, and damp soil receive a smaller registration weight. The physical meaning of the registration weight is that in the subsequent process of establishing the correspondence between point pairs and solving the spatial transformation parameters, the higher weight will contribute more to the final result, thereby enhancing the constraint effect of areas with obvious structural features and suppressing the interference of noise or weak texture areas.

[0020] Step 2: For each reference point in the reference point cloud, based on the weighted distance calculation with registration weights, determine the target point with the smallest weighted distance as its corresponding point to establish the correspondence between point pairs; The logic for establishing a point-to-point correspondence is as follows: For each reference point in the reference point cloud, calculate the Euclidean distance between the reference point and each target point in the target point cloud. Divide each calculated Euclidean distance by the registration weight of the corresponding target point in the target point cloud to obtain the weighted distance between the reference point and the target point. Select the target point in the target point cloud that has the smallest weighted distance to the reference point as the corresponding point of the reference point, thereby establishing a point-to-point correspondence between the reference point and the target point, which is called a corresponding point pair. The reference point is a single data point in the reference point cloud. Each reference point contains the three-dimensional coordinates in the geodetic coordinate system acquired at the previous moment, without carrying registration weights. The relationship between the reference point and the reference point cloud is that of a whole to an individual. The target is a single data point in the target point cloud. Each target point contains the three-dimensional coordinates in the geodetic coordinate system acquired at the current moment, as well as the registration weights calculated in step 1. The relationship between the target point and the target point cloud is also that of a whole to an individual. The corresponding point pair is a binary tuple consisting of a reference point and a target point. The target point is the corresponding point selected from the target point cloud for the reference point by minimizing the weighted distance criterion. The corresponding point pair records the mapping relationship between point cloud data of the same spatial location at two different moments and is the basic unit for subsequent solution of spatial transformation parameters. The establishment of this correspondence is the foundation for subsequent solutions to spatial transformation parameters. Its core idea is: for each reference point, find a target point in the target point cloud that is spatially closest to it and has a high registration weight as its corresponding point, thus forming a one-to-one point pair mapping; specifically, the reference point cloud contains several reference points collected at the previous moment, each reference point having three-dimensional coordinates in the geodetic coordinate system, and the target point cloud contains several target points collected at the current moment, each target point also having three-dimensional coordinates in the geodetic coordinate system, and each target point has obtained a registration weight through step 1, which reflects the importance of the target point in subsequent calculations; For each reference point in the reference point cloud, the first step is to calculate the spatial straight-line distance between the reference point and each target point in the target point cloud, i.e., the Euclidean distance. Let the three-dimensional coordinates of a reference point be and the three-dimensional coordinates of a target point be . The Euclidean distance between them is defined as the square root of the sum of the squares of the differences between the three coordinate components of the reference point and the three coordinate components of the target point. The smaller the Euclidean distance, the closer the reference point and the target point are in spatial position. However, selecting the corresponding point solely based on Euclidean distance presents a problem because the target point cloud contains a large number of points with low reflection intensity. These points correspond to areas with insignificant geometric features or are easily affected by noise, such as vegetation, water surfaces, and damp soil. If only Euclidean distance is considered, these low-quality points will be incorrectly selected as the corresponding points, thereby reducing the registration accuracy. Therefore, it is necessary to introduce the registration weights of each target point calculated in step 1 to correct the Euclidean distance. The specific method is as follows: Divide the Euclidean distance between the reference point and each target point by the registration weight of the target point to obtain a new metric called the weighted distance. The formula for calculating the weighted distance is: weighted distance equals Euclidean distance divided by registration weight. Since the registration weight ranges from greater than zero to less than or equal to 1, the registration weight of a target point with high reflectivity is close to 1, while the registration weight of a target point with low reflectivity is much less than 1. Therefore, for two target points with similar Euclidean distances, the target point with a larger registration weight will have a smaller weighted distance, and the target point with a smaller registration weight will have a larger weighted distance. In other words, the weighted distance reflects both the proximity of the spatial locations and the quality of the target points, making it easier to select target points with high reflectivity and high weight as the corresponding points. The formula for calculating the weighted distance is: ,in This refers to a reference point (three-dimensional coordinate vector) in the reference point cloud. Let i be the i-th target point (three-dimensional coordinate vector) in the target point cloud. For reference point With the target point The Euclidean distance between them Weighted distance; For each reference point p, the target point that minimizes the weighted distance is selected as its corresponding point, thus obtaining a series of corresponding point pairs, denoted as p. ,in , For the total number of corresponding point pairs, each It is a selected target point in the target point cloud, and its corresponding weight is... (i.e., the original weight of the target point); For each reference point in the reference point cloud, the weighted distance between the reference point and all target points in the target point cloud is calculated sequentially. Among all the calculated weighted distances, the minimum value is found. The target point corresponding to the minimum value is the corresponding point of the reference point. In other words, the corresponding point is selected from the target point cloud by minimizing the weighted distance criterion. This selection method ensures that each reference point is established in a one-to-one correspondence with a target point that is spatially close and of high quality. Once each reference point in the reference point cloud has found its corresponding target point, a series of point pairs are formed. Each point pair contains a reference point and a target point corresponding to that reference point. Such point pairs are called corresponding point pairs. The set of all corresponding point pairs constitutes the correspondence between the reference point cloud and the target point cloud. This correspondence will be used to solve the spatial transformation parameters in step 3. It should be noted that during the process of establishing the correspondence, there may be cases where multiple reference points correspond to the same target point, or some target points may not be corresponding to any reference point. These situations will be gradually improved in subsequent iterative optimizations.

[0021] Step 3: Construct an objective function based on the correspondence and registration weights, solve the objective function using the singular value decomposition method to obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, the displacement vector of each reference point is its transformed coordinates minus its original coordinates, decompose the displacement vector of each reference point into displacement feature vectors including horizontal and vertical displacements to construct the displacement field; Since the reference point cloud and the target point cloud were collected at the previous and current times respectively, displacement and deformation of the lock pit slope occurred between the two times, resulting in positional deviations between reference points and corresponding target points at the same spatial location. To accurately calculate the displacement of each reference point from the previous time to the current time, it is necessary to find an optimal spatial transformation parameter so that the reference point cloud can be aligned with the target point cloud to the maximum extent after the transformation. This spatial transformation parameter includes a rotation matrix and a translation vector. The rotation matrix is ​​a 3x3 orthogonal matrix used to describe the rotational motion of the reference point cloud in three-dimensional space; the translation vector is a column vector with three components used to describe the translational motion of the reference point cloud in three-dimensional space. Applying the rotation matrix and translation vector together to each reference point in the reference point cloud yields the transformed coordinates of that reference point, which reflect the position that the reference point should be in after alignment with the target point cloud. The method for determining the optimal rotation matrix and translation vector is as follows: Based on the established correspondence between reference points and target points, an objective function is constructed with rotation matrix and translation vector as independent variables. The value of the objective function is calculated as follows: For each pair of corresponding points, the reference point in the pair is spatially transformed using rotation matrix and translation vector to obtain the coordinates of the transformed reference point. The spatial distance between the transformed reference point coordinates and the corresponding target point coordinates is then calculated, and the square of this spatial distance is calculated. The squared value is then multiplied by the registration weight of the target point in the pair to obtain the weighted squared distance of the pair. The sum of the weighted squared distances of all pairs of corresponding points is the value of the objective function. The rotation matrix and translation vector that minimize the objective function are found using singular value decomposition. The obtained rotation matrix and translation vector are taken as the optimal rotation matrix and translation vector. The optimal rotation matrix and translation vector are applied to each reference point in the reference point cloud to obtain the transformed reference point cloud. The value of the objective function is calculated using the following formula: in, This is the reference point (three-dimensional coordinate vector) in the k-th corresponding point pair. This is the target point (three-dimensional coordinate vector) in the kth corresponding point pair. For the kth corresponding point, the target point is... Registration weights, Let be the rotation matrix (3×3 orthogonal matrix) to be solved, satisfying (Identity matrix), and its determinant is (This represents pure rotation, without reflection). The rotation matrix contains three independent parameters (represented by rotation angles about the X, Y, and Z axes). Let be the translation vector to be solved, which is The column vectors represent the translation along the X, Y, and Z axes in the geodetic coordinate system. The units of the translation vectors are the same as the units of the three-dimensional coordinates. Reference point via rotation matrix Translation vector Transformed coordinates and target point The square of the Euclidean distance between the coordinates, this value follows and It changes with the change, and its range of values ​​is non-negative real numbers; The objective function value is denoted by , which is the sum of the squared weighted distances between all corresponding point pairs, and its value range is a non-negative real number. Find Minimized and The objective function is the optimal rotation matrix and translation vector, which is solved using the weighted singular value decomposition method. This objective function is a weighted least squares error function, with equal weight to all point pairs in the traditional iterative nearest point algorithm (i.e., ...). Unlike other formulas, this one introduces registration weights. The error terms of high-weight corresponding point pairs (corresponding to target points with high reflection intensity) account for a larger proportion in the summation process, thereby driving the optimization process to prioritize aligning these high-confidence feature points. The error terms of low-weight corresponding point pairs are suppressed, minimizing the impact of noise or weak texture regions on the final transformation parameters. This improves the accuracy and robustness of point cloud registration, especially in complex environments such as lock pits where there are many low-reflection intensity interference areas (water surface, wet mud, vegetation). To find the optimal rotation matrix and translation vector, an objective function needs to be constructed to quantitatively evaluate the quality of the transformation parameters. The objective function takes the rotation matrix and translation vector as independent variables. The smaller the function value, the higher the alignment between the reference point set and the target point set after transformation. Specifically, based on each corresponding point pair established in step 2, each corresponding point pair contains a reference point and a target point corresponding to the reference point, and the target point carries the registration weight calculated in step 1. For each corresponding point pair, the three-dimensional coordinates of the reference point in the corresponding point pair are multiplied by the rotation matrix and then added to the translation vector to obtain the new coordinates of the reference point after spatial transformation, which are called the transformed reference point coordinates. Then, the spatial distance between the coordinates of the reference point after the transformation and the coordinates of its corresponding target point is calculated, i.e., the Euclidean distance. This spatial distance is squared to obtain the squared spatial distance of the corresponding point pair. Then, the squared spatial distance is multiplied by the registration weight of the target point in the corresponding point pair to obtain the squared weighted distance of the corresponding point pair. The role of the registration weight is to make the corresponding point pairs with high reflectivity and obvious structural features occupy a larger proportion in the objective function, thereby improving the accuracy of solving the transformation parameters. The squared weighted distances of all current corresponding point pairs are added together, and the sum is the function value of the objective function under the current rotation matrix and translation vector. The mathematical essence of this objective function is an error function in the form of weighted least squares. To find the rotation matrix and translation vector that minimizes the objective function, singular value decomposition (SVD) is used. SVD is a matrix factorization algorithm that decomposes any real matrix into the product of three matrices: the left singular matrix... diagonal matrix And right singular matrix The specific steps for solving the optimal rotation matrix and translation vector in this invention are as follows: Calculate the weighted centroid of the reference point set, which is the weighted average of all reference point coordinates multiplied by the corresponding target point weights; similarly, calculate the weighted centroid of the target point set; subtract the weighted centroid of each reference point to obtain the decentralized reference point coordinates; subtract the weighted centroid of each target point to obtain the decentralized target point coordinates, and construct the weighted covariance matrix. : ,in and Let the weighted centroids of the reference point cloud and the target point cloud be respectively. Calculate the weighted centroid of the reference point cloud: Calculate the weighted centroid of the target point cloud: Its elements are equal to the product of the coordinates of the decentralized reference point and the coordinates of the decentralized target point in all corresponding point pairs, multiplied by the registration weight of the corresponding target point, and summed over all point pairs; Subtracting the weighted centroid of each reference point from the reference point's coordinates yields the decentralized reference point's coordinates: Subtract the weighted centroid of each target point to obtain the decentralized target point coordinates: ; For covariance matrix Perform singular value decomposition to obtain , Given the transpose sign, calculate the candidate rotation matrix. Verify the determinant value of the candidate rotation matrix. If the determinant value is positive 1, then... This is the optimal rotation matrix; if the determinant is negative 1, then it needs to be... Invert the last column and recalculate to make the determinant a positive 1; this is the optimal translation vector. It equals the weighted centroid of the target point set minus the optimal rotation matrix multiplied by the weighted centroid of the reference point set; specifically, if ,but If the determinant value is Then it is necessary to Recalculate after inverting the last column Make the determinant value Optimal translation vector ; Singular value decomposition is performed on the covariance matrix to obtain three orthogonal matrices. The optimal solution of the rotation matrix is ​​calculated based on the decomposition results, which is the product of the two orthogonal matrices obtained by decomposition. The determinant of the rotation matrix is ​​ensured to be positive 1 to ensure that it represents pure rotation rather than reflection. After obtaining the optimal rotation matrix, the optimal solution of the translation vector is equal to the weighted centroid of the target point set minus the optimal rotation matrix multiplied by the weighted centroid of the reference point set. Thus, the obtained rotation matrix and translation vector are the optimal rotation matrix and translation vector that minimize the objective function. A rotation matrix is ​​a 3x3 orthogonal matrix used to describe the rotational motion of a rigid body about coordinate axes in three-dimensional space. The rotation matrix satisfies the following properties: the magnitude of each row and column is 1; the dot product between different rows or columns is 0; and the determinant of the matrix is ​​positive 1 (indicating pure rotation, excluding reflection). The rotation matrix contains three independent parameters, represented by rotation angles about the X, Y, and Z axes. A translation vector is a column vector containing three components, representing the translation along the X, Y, and Z axes in the geodetic coordinate system. The units of the translation vector are the same as those of the three-dimensional coordinates. The displacement vector is decomposed into displacement eigenvectors of horizontal and vertical displacements to obtain the displacement field, which specifically includes: For each reference point in the reference point cloud, its coordinates after optimal rotation matrix and translation vector transformation are subtracted from its original coordinates before transformation to obtain the displacement vector of the reference point. Two horizontal components are extracted from the displacement vector, and the square root of the sum of the squares of these two components is taken to obtain the horizontal displacement characteristic vector of the reference point, defined as the horizontal displacement. The vertical component of the displacement vector is extracted as the vertical displacement characteristic vector of the reference point, defined as the vertical displacement. The horizontal and vertical displacements of all reference points are organized according to their spatial positions to form a displacement field reflecting the magnitude and direction of the deformation of the lock pit slope. The optimal rotation matrix and optimal translation vector obtained from the solution are applied to each reference point in the reference point cloud in turn. For each reference point, its three-dimensional coordinates are first multiplied by the optimal rotation matrix and then added to the optimal translation vector to obtain the transformed coordinates of the reference point. The transformed coordinates of all reference points constitute a point cloud, which is called the transformed reference point cloud. The transformed reference point cloud is aligned with the target point cloud to the maximum extent in space, but there will still be residual deviations between the two caused by the actual displacement of the slope. These residual deviations are the basis for subsequent calculation of the displacement vector. The displacement vector is defined as follows: the transformed coordinates of the reference point minus the original coordinates of the reference point before the transformation. The transformed coordinates reflect the expected position of the reference point when it is aligned with the target point at the current moment, and the original coordinates before the transformation reflect the actual position of the reference point at the previous moment. The difference between the two is the three-dimensional spatial displacement of the reference point from the previous moment to the current moment. This difference is a vector containing three components, called the displacement vector. The three components of the displacement vector correspond to the eastward displacement, northward displacement, and vertical displacement in the geodetic coordinate system, respectively. To facilitate engineering analysis and safety assessment, the three-dimensional displacement vector needs to be decomposed into two independent components: horizontal displacement and vertical displacement. Horizontal displacement reflects the sliding or lateral deformation of the slope in the horizontal plane, while vertical displacement reflects the settlement or uplift of the slope. The specific decomposition method is as follows: extract the two horizontal components from the displacement vector, namely the eastward displacement component and the northward displacement component. Add the squares of these two components, and then take the square root of the sum to obtain the displacement characteristic vector of the horizontal displacement of the reference point, which is defined as the horizontal displacement quantity. The horizontal displacement quantity is a scalar representing the total distance actually moved by the reference point in the horizontal plane. Extract the vertical displacement characteristic vector from the displacement vector and directly use it as the vertical displacement quantity of the reference point. A positive value of the vertical displacement quantity indicates uplift, and a negative value indicates settlement. After all reference points have completed the calculation of horizontal and vertical displacements, these displacements need to be organized according to the spatial location of each reference point. Specifically, each reference point has a unique three-dimensional coordinate in the geodetic coordinate system, which identifies the specific location where the displacement occurred. The spatial coordinates of all reference points are associated with the corresponding horizontal and vertical displacements to form a spatially distributed dataset, which is called the displacement field.

[0022] Step 4: For each displacement gauge installed on the slope of the foundation pit, extract the displacement feature vector at its installation location from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. Analyze the deviation vector at each displacement gauge using the Kriging space interpolation method to construct a systematic deviation field covering the entire area. Subtract the systematic deviation field from the displacement field to obtain the calibrated displacement field. Analyze the displacement value and displacement rate in the calibrated displacement field to determine the deformation area. Because of systematic errors in the lidar point cloud acquisition process, including laser ranging errors, coordinate transformation residuals, and differences in scanning angles at different times, the displacement field contains certain systematic deviations. In order to eliminate these deviations and improve the accuracy of the displacement field in reflecting the real slope deformation, it is necessary to use the monitoring data of displacement gauges deployed on the foundation pit slope to calibrate the displacement field. The displacement gauge is a contact deformation monitoring instrument consisting of a displacement sensor, a fixed end, and a measuring rod. Its installation position is selected on the key section or potential slip zone of the slope, which can continuously and in real time measure the measured vector at that point and has high local accuracy. By comparing the high-precision measured vector of the displacement gauge with the calculated value at the corresponding position in the point cloud displacement field, the distribution of systematic deviations can be obtained. Then, the discrete deviation information is extended to the entire monitoring area through spatial interpolation. Finally, the systematic deviation is subtracted from the displacement field to obtain the calibrated displacement field. The regions in the displacement field after analysis and calibration where the displacement value and displacement rate exceed the preset displacement threshold are defined as deformation regions, specifically including: A displacement rate threshold and a cumulative displacement threshold are set. The displacement rate threshold is used to evaluate the short-term deformation rate of the lock pit slope, and the cumulative displacement threshold is used to evaluate the total long-term deformation of the lock pit slope. For each reference point in the calibrated displacement field, the calibrated horizontal displacement and calibrated vertical displacement carried by that reference point from the previous moment to the current moment are obtained. The calibrated horizontal displacement is divided by the time interval between the previous moment and the current moment to obtain the horizontal displacement rate of the reference point. The calibrated vertical displacement is divided by the time interval to obtain the vertical displacement rate of the reference point. At the same time, from the initial monitoring moment to the current moment, the calibrated horizontal displacement of the reference point in each time interval is accumulated to obtain the cumulative horizontal displacement of the reference point. The calibrated vertical displacement of the reference point in each time interval is accumulated to obtain the cumulative vertical displacement of the reference point. The displacement field is a dataset consisting of the spatial coordinates of all reference points and their corresponding horizontal and vertical displacements. Since the installation position of the displacement meter will not coincide exactly with the spatial coordinates of a certain reference point, an interpolation method is needed to extract the displacement value at that position from the displacement field. Specifically, the three reference points closest to the installation position of the displacement meter are searched in the displacement field. The horizontal and vertical displacements of the three reference points are weighted and averaged using the reciprocal of the distance between the three reference points and the installation position of the displacement meter as weights to obtain the point cloud displacement value at the displacement meter position. This point cloud displacement value reflects the displacement estimate calculated by the lidar measurement system at that position from the previous moment to the current moment. The deviation vector is defined as the difference between the point cloud displacement value and the measured vector of the displacement gauge, i.e., the point cloud displacement value minus the measured vector of the displacement gauge. The deviation vector is a three-dimensional vector containing horizontal and vertical deviations. If the measured vector of the displacement gauge only provides a component in a certain direction, then only the deviation in that direction is calculated, and the deviations in other directions are not considered for the time being. The physical meaning of the deviation vector is: the systematic error generated by the point cloud measurement system relative to the high-precision displacement gauge at this location. Since the number of displacement gauges is limited, generally dozens, the deviation information only exists on a limited number of discrete points. In order to obtain the systematic deviation at any location on the entire slope surface, it is necessary to use the deviation vectors of these discrete points for spatial interpolation to construct a continuous deviation distribution covering the entire monitoring area, which is called the systematic deviation field. Point cloud displacement values ​​are displacement estimates extracted from the displacement field and corresponding to the installation location of the displacement gauge. Since the coordinates of the displacement gauge installation location will not exactly coincide with the coordinates of a reference point in the displacement field, an interpolation method is required. The specific interpolation method is as follows: search for the three reference points closest to the displacement gauge installation location in the displacement field, calculate the Euclidean distance between these three reference points and the displacement gauge installation location, take the reciprocal of the distance as the weight, and perform a weighted average of the horizontal and vertical displacements of the three reference points to obtain the point cloud displacement value at the displacement gauge installation location. The mathematical expression of this interpolation method is: the point cloud displacement value is equal to the sum of the displacements of the three reference points multiplied by the reciprocal of their respective distances, and then divided by the sum of the reciprocals of the three distances. This invention employs the Kriging spatial interpolation method to construct a systematic deviation field. Kriging interpolation is a statistical spatial interpolation technique whose core idea is to use the spatial autocorrelation between known points to predict the values ​​of unknown points. This method considers not only the distance between known points and the interpolation point but also the spatial distribution structure between known points, providing optimal unbiased estimates. The specific operation steps are as follows: First, a semi-variogram analysis is performed on the deviation vectors at all displacement gauge locations to calculate the spatial variation of the deviation vectors at different distances, fitting a semi-variogram model, such as a spherical model, exponential model, or Gaussian model. Then, for any interpolation point on the slope, the influence weights of each displacement gauge on that point are determined according to the semi-variogram model, minimizing the variance of the interpolation result. The systematic deviation value of the interpolation point is equal to the sum of the products of the deviation vectors at each displacement gauge and their corresponding weights. By traversing all spatial locations within the entire monitoring area, the systematic deviation value at each location is obtained, thus forming a continuous systematic deviation field covering the entire displacement field region. Each deviation vector in the systematic deviation field has the same component structure as the displacement feature vector in the displacement field. If the horizontal displacement rate of a reference point exceeds the horizontal displacement rate threshold, or the vertical displacement rate exceeds the vertical displacement rate threshold, or the cumulative horizontal displacement exceeds the cumulative horizontal displacement threshold, or the cumulative vertical displacement exceeds the cumulative vertical displacement threshold, then the reference point is marked as a deformation point; the continuous spatial region formed by all adjacent deformation points is marked as a deformation region, and the spatial location, geometric range, and the horizontal displacement, vertical displacement, and resultant displacement of each reference point within the region are output. The resultant displacement is calculated by taking the square root of the sum of the squares of the horizontal and vertical displacements. The calibration method involves extracting the system deviation value at the corresponding position from the system deviation field based on the spatial coordinates of each reference point in the displacement field, subtracting the system deviation value from the displacement vector of the reference point to obtain the calibrated displacement feature vector of the reference point, and reorganizing the calibrated displacement feature vectors of all reference points according to their spatial positions to form a calibrated displacement field. The calibrated displacement field eliminates the systematic errors introduced by point cloud acquisition and registration algorithms, and more realistically reflects the actual deformation of the lock pit slope. For the deformation zone, which refers to the continuous spatial area where the slope surface has shifted and endangers construction safety, in order to quantify the significance of the deformation, two thresholds need to be set: displacement rate threshold and cumulative displacement threshold. The displacement rate threshold is used to assess the short-term deformation rate of the slope, reflecting the intensity of the slope's activity per unit time. It is set according to engineering geological conditions and construction specifications. For example, the horizontal displacement rate threshold is set to 5 mm per day and the vertical displacement rate threshold is set to 3 mm per day. The cumulative displacement threshold is used to assess the total long-term deformation of the slope, reflecting the cumulative deformation of the slope from the initial monitoring time to the current time. For example, the horizontal cumulative displacement threshold is set to 50 mm and the vertical cumulative displacement threshold is set to 30 mm. These thresholds can be dynamically adjusted according to the support design level of the foundation pit, the sensitivity of the surrounding environment, and the monitoring cycle. For each reference point in the calibrated displacement field, it is necessary to obtain the horizontal and vertical displacement of the reference point from the previous moment to the current moment. These displacements have been calculated in step 3, but have been corrected after calibration. Therefore, the calibrated horizontal and vertical displacements are used. The time interval between the previous moment and the current moment is recorded as the time interval, which is generally the interval between two scans, such as one day or one week. The horizontal displacement of the reference point is divided by the time interval to obtain the horizontal displacement rate of the reference point, in millimeters per day. The vertical displacement of the reference point is divided by the time interval to obtain the vertical displacement rate of the reference point. It is necessary to calculate the cumulative displacement of the reference point from the initial monitoring time to the current time. The initial monitoring time refers to the first moment when monitoring of the foundation pit slope begins, which is the benchmark measurement time before the foundation pit is excavated or at the start of construction. From the initial monitoring time to the current time, several time intervals have passed. Within each time interval, the reference point corresponds to a horizontal displacement and a vertical displacement. The horizontal displacements within these time intervals are algebraically summed to obtain the cumulative horizontal displacement of the reference point. The vertical displacements within these time intervals are algebraically summed to obtain the cumulative vertical displacement of the reference point. When summing, attention should be paid to the sign of the displacement: the horizontal displacement is taken as the absolute value or according to the sign of the specified direction; a positive value for the vertical displacement indicates uplift, and a negative value indicates settlement. The judgment criteria are set as follows: if the horizontal displacement rate of a reference point exceeds the horizontal displacement rate threshold, or the vertical displacement rate exceeds the vertical displacement rate threshold, or the cumulative horizontal displacement exceeds the cumulative horizontal displacement threshold, or the cumulative vertical displacement exceeds the cumulative vertical displacement threshold, then the reference point is determined to be a deformation point. As long as any one of the above four conditions is met, the reference point is considered to have deformed and needs to be marked. The mark of the deformation point is a binary attribute, that is, whether the point is a deformation point. After determining the deformation points of all reference points, adjacent deformation points need to be merged into continuous regions. The basis for determining adjacency is the spatial distance between reference points. A connection distance threshold is set, for example, 0.5 meters. If the Euclidean distance between two deformation points is less than this threshold, they are considered to be adjacent. All deformation points that meet the adjacency condition are connected to form several connected regions. The spatial range formed by the deformation points in each connected region is called the deformation region. There are one or more deformation regions, depending on the deformation distribution of the slope. Finally, the output includes detailed information for each deformation region, including: the spatial location of the deformation region, i.e., the geodetic coordinate range covered by the region, which can be represented by the corner coordinates of the minimum bounding box; the geometric range of the deformation region, i.e., the area, longest length, average width, and other geometric parameters of the region; and the horizontal displacement, vertical displacement, and combined displacement of each reference point within the region. The combined displacement is the synthesis of the horizontal and vertical displacements, and its calculation method is: add the squares of the horizontal and vertical displacements, and then take the square root of the sum. The combined displacement reflects the total distance the reference point moves in three-dimensional space and is a comprehensive indicator for assessing the severity of the overall slope deformation. This output information can be used to generate deformation cloud maps, displacement contour maps, and deformation alarm reports for engineering management personnel to conduct safety assessments and construction decisions. Obtaining the reference point cloud and the target point cloud also includes preprocessing the raw point cloud data: Using a ground-based 3D laser scanner or an unmanned aerial vehicle (UAV) oblique photography system, the slope of the lock pit is scanned at fixed time intervals to obtain the original laser point cloud data of the previous and current times. The original laser point cloud data is then denoised and filtered to remove outliers and noise points. The point clouds from each period are then uniformly registered to the same geodetic coordinate system through coordinate transformation to ensure the spatial consistency of the point cloud data at different times. In the process of acquiring reference point clouds and target point clouds, it is also necessary to preprocess the raw point cloud data to ensure data quality and spatial consistency. Specifically, a terrestrial 3D laser scanner or UAV oblique photography system is used to scan the slope of the lock pit at a fixed time cycle. This time cycle is set according to the construction progress and deformation rate, such as once a day or once a week. The raw laser point cloud data obtained from each scan contains outliers and noise points, which are generated by factors such as birds, dust, and equipment vibration, and do not reflect the true slope surface geometry. Therefore, the original point cloud data needs to be denoised and filtered. The denoising process uses a statistical filtering method. For each point, the number of neighboring points within a radius of 0.5 meters is calculated. If the number of neighboring points is less than three, the point is identified as an outlier and removed. The voxel grid filtering method is used for downsampling, and the point cloud space is divided into small cubic grids with a side length of 0.05 meters. The point closest to the grid center is retained in each grid. After preprocessing, the point clouds of each period are uniformly registered to the CGCS2000 geodetic coordinate system through a seven-parameter transformation model.

[0023] Please see Figure 4 The present invention also provides an intelligent measurement device for complex environment lock pits, the device being used to perform the above-described intelligent measurement method for complex environment lock pits, comprising: The point cloud acquisition module is used to acquire the reference point cloud set of the lock pit slope at the previous moment and the target point cloud set at the current moment. Each point in the point cloud set contains information on its three-dimensional coordinates and laser reflection intensity. The registration weight is determined based on the laser reflection intensity of each target point in the target point cloud set. The relationship registration module is used to determine the target point with the smallest weighted distance as its corresponding point for each reference point in the reference point cloud based on the weighted distance calculation with introduced registration weights, so as to establish the correspondence between point pairs. The data calculation module is used to construct an objective function based on the correspondence and registration weights, solve the objective function by singular value decomposition, obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, and the displacement vector of each reference point is its transformed coordinates minus its original coordinates. The displacement vector of each reference point is decomposed into displacement feature vectors including horizontal and vertical displacements to construct the displacement field. The calibration module is used to extract the displacement feature vector at the installation location of each displacement gauge deployed on the slope of the foundation pit from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. The deviation vector at each displacement gauge is analyzed by Kriging space interpolation to construct a systematic deviation field covering the entire area. The systematic deviation field is then subtracted from the displacement field to obtain the calibrated displacement field. The displacement value and displacement rate in the calibrated displacement field are analyzed to determine the deformation area.

[0024] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0025] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0026] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0027] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent measurement of lock pits in complex environments, characterized in that, The specific steps include: Step 1: Obtain the reference point cloud of the lock pit slope at the previous moment and the target point cloud at the current moment. Each point in the point cloud contains information on its three-dimensional coordinates and laser reflection intensity. Determine the registration weight of each target point in the target point cloud based on its laser reflection intensity. Step 2: For each reference point in the reference point cloud, based on the weighted distance calculation with registration weights, determine the target point with the smallest weighted distance as its corresponding point to establish the correspondence between point pairs; Step 3: Construct an objective function based on the correspondence and registration weights, solve the objective function using the singular value decomposition method to obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, the displacement vector of each reference point is its transformed coordinates minus its original coordinates, decompose the displacement vector of each reference point into displacement feature vectors including horizontal and vertical displacements to construct the displacement field; Step 4: For each displacement gauge installed on the slope of the foundation pit, extract the displacement characteristic vector at its installation location from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. Analyze the deviation vector at each displacement gauge using the Kriging space interpolation method to construct a systematic deviation field covering the entire area. Subtract the systematic deviation field from the displacement field to obtain the calibrated displacement field. Analyze the displacement value and displacement rate in the calibrated displacement field to determine the deformation area.

2. The intelligent measurement method for lock pits in complex environments according to claim 1, characterized in that: Obtain the set of reference points for the lock pit slope at the previous moment and the set of target points at the current moment, specifically including: Control points were set up within a predetermined area around the slope of the lock pit. The geodetic coordinates of the control points were obtained by static measurement using a global navigation satellite system. A drone equipped with a lidar scanning system was used to perform a full-coverage scan of the lock pit along a set route, acquiring laser point cloud data at the previous and current times. Each point in the laser point cloud data contains three-dimensional coordinates in the laser point cloud coordinate system and laser reflection intensity information. Using the geodetic coordinates of the control points and the three-dimensional coordinates of their corresponding points in the laser point cloud coordinate system, a seven-parameter transformation model was solved using the least squares method to transform the laser point cloud data at the two times to the same geodetic coordinate system, which were then used as the reference point cloud set and the target point cloud set, respectively. Based on the laser reflection intensity of each target point in the target point cloud, the registration weight of each target point is determined, specifically including: The laser reflection intensity of all target points in the target point cloud is obtained, and the maximum reflection intensity value is determined. Based on the positive correlation between laser reflection intensity and registration weight, for each target point in the target point cloud, the registration weight of the target point is calculated through nonlinear mapping based on the difference between the laser reflection intensity of the target point and the maximum reflection intensity value, as well as the preset adjustment coefficient, so that the target point with higher laser reflection intensity gets a larger registration weight.

3. The intelligent measurement method for lock pits in complex environments according to claim 2, characterized in that: The logic for establishing a point-to-point correspondence is as follows: For each reference point in the reference point cloud, calculate the Euclidean distance between the reference point and each target point in the target point cloud. Divide each calculated Euclidean distance by the registration weight of the corresponding target point in the target point cloud to obtain the weighted distance between the reference point and the target point. Select the target point in the target point cloud that has the smallest weighted distance to the reference point as the corresponding point of the reference point, thereby establishing a point-to-point correspondence between the reference point and the target point, which is called a corresponding point pair. The method for determining the optimal rotation matrix and translation vector is as follows: Based on the established correspondence between reference points and target points, an objective function is constructed with rotation matrix and translation vector as independent variables. The value of the objective function is calculated as follows: For each pair of corresponding points, the reference point in the pair is spatially transformed using rotation matrix and translation vector to obtain the coordinates of the transformed reference point. The spatial distance between the transformed reference point coordinates and the corresponding target point coordinates is then calculated, and the square of this spatial distance is calculated. The squared value is then multiplied by the registration weight of the target point in the pair to obtain the weighted squared distance of the pair. The sum of the weighted squared distances of all pairs of corresponding points is the value of the objective function. The rotation matrix and translation vector that minimize the objective function are obtained by using singular value decomposition. The obtained rotation matrix and translation vector are taken as the optimal rotation matrix and translation vector. The optimal rotation matrix and translation vector are applied to each reference point in the reference point cloud to obtain the transformed reference point cloud.

4. The intelligent measurement method for a complex environment lock pit according to claim 3, characterized in that: The displacement vector is decomposed into displacement eigenvectors of horizontal and vertical displacements to obtain the displacement field, which specifically includes: For each reference point in the reference point cloud, its coordinates after optimal rotation matrix and translation vector transformation are subtracted from its original coordinates before transformation to obtain the displacement vector of that reference point. Two horizontal components are extracted from the displacement vector, and the square root of the sum of the squares of these two components is taken to obtain the horizontal displacement characteristic vector of that reference point, defined as the horizontal displacement. The vertical component of the displacement vector is extracted as the vertical displacement characteristic vector of that reference point, defined as the vertical displacement. The horizontal and vertical displacements of all reference points are organized according to their spatial positions to form a displacement field reflecting the magnitude and direction of the deformation of the lock pit slope.

5. The intelligent measurement method for lock pits in complex environments according to claim 4, characterized in that: The system deviation field is obtained through spatial interpolation, and the calibrated displacement field is obtained by subtracting the system deviation field from the displacement field. Specifically, it includes: The monitoring data of multiple displacement gauges installed at predetermined locations on the slope of the lock pit are obtained at the current moment. The monitoring data consists of the measured vectors of each displacement gauge's installation location from the previous moment to the current moment. The measured vector of each displacement gauge includes the actual displacement in the horizontal direction and the actual displacement in the vertical direction. For each displacement gauge, the point cloud displacement value at the corresponding three-dimensional coordinates in the displacement field is extracted from the displacement field. The point cloud displacement value includes the horizontal displacement and the vertical displacement. The difference between the point cloud displacement value and the actual displacement of each displacement gauge installation location is calculated to obtain the system deviation value in the horizontal direction and the system deviation value in the vertical direction at that location. Based on the horizontal systematic deviation values ​​at all displacement gauge locations, the Kriging space interpolation method is used to calculate the horizontal systematic deviation values ​​at any three-dimensional coordinate of the lock pit slope, forming a horizontal systematic deviation field. Similarly, based on the vertical systematic deviation values ​​at all displacement gauge locations, the Kriging space interpolation method is used to form a vertical systematic deviation field. The horizontal and vertical systematic deviation fields together constitute a continuous systematic deviation field covering the entire displacement field region. For each reference point in the displacement field, according to the three-dimensional coordinates of the reference point, the horizontal and vertical systematic deviation values ​​at the corresponding locations are obtained from the horizontal and vertical systematic deviation fields, respectively. The horizontal displacement of the reference point is subtracted from the horizontal systematic deviation value, and the vertical displacement is subtracted from the vertical systematic deviation value to obtain the calibrated horizontal and vertical displacement values ​​of the reference point. The calibrated horizontal and vertical displacement values ​​of all reference points are organized according to the three-dimensional coordinates of each reference point to form a calibrated displacement field.

6. The intelligent measurement method for a complex environment lock pit according to claim 5, characterized in that: The regions in the displacement field after analysis and calibration where the displacement value and displacement rate exceed the preset displacement threshold are defined as deformation regions, specifically including: A displacement rate threshold and a cumulative displacement threshold are set. The displacement rate threshold is used to evaluate the short-term deformation rate of the lock pit slope, and the cumulative displacement threshold is used to evaluate the total long-term deformation of the lock pit slope. For each reference point in the calibrated displacement field, the calibrated horizontal displacement and calibrated vertical displacement carried by that reference point from the previous moment to the current moment are obtained. The calibrated horizontal displacement is divided by the time interval between the previous moment and the current moment to obtain the horizontal displacement rate of the reference point. The calibrated vertical displacement is divided by the time interval to obtain the vertical displacement rate of the reference point. At the same time, from the initial monitoring moment to the current moment, the calibrated horizontal displacement of the reference point in each time interval is accumulated to obtain the cumulative horizontal displacement of the reference point. The calibrated vertical displacement of the reference point in each time interval is accumulated to obtain the cumulative vertical displacement of the reference point. If the horizontal displacement rate of a reference point exceeds the horizontal displacement rate threshold, or the vertical displacement rate exceeds the vertical displacement rate threshold, or the cumulative horizontal displacement exceeds the cumulative horizontal displacement threshold, or the cumulative vertical displacement exceeds the cumulative vertical displacement threshold, then the reference point is marked as a deformation point; the continuous spatial region formed by all adjacent deformation points is marked as a deformation region, and the spatial location, geometric range, and the horizontal displacement, vertical displacement, and resultant displacement of each reference point within the region are output. The resultant displacement is calculated by taking the square root of the sum of the squares of the horizontal and vertical displacements.

7. The intelligent measurement method for lock pits in complex environments according to claim 2, characterized in that: Obtaining the reference point cloud and the target point cloud also includes preprocessing the raw point cloud data: Using a terrestrial 3D laser scanner or an unmanned aerial vehicle (UAV) oblique photography system, the slope of the lock pit is scanned at fixed time intervals to obtain the original laser point cloud data of the previous and current moments. The acquired original laser point cloud data is then denoised and filtered to remove outliers and noise points. The point clouds from each period are then uniformly registered to the same geodetic coordinate system through coordinate transformation to ensure the spatial consistency of the point cloud data at different times.

8. An intelligent measurement device for a complex environment lock pit, characterized in that: The device is used to perform the intelligent measurement method for complex environment lock pits as described in any one of claims 1-7, including: The point cloud acquisition module is used to acquire the reference point cloud set of the lock pit slope at the previous moment and the target point cloud set at the current moment. Each point in the point cloud set contains information on its three-dimensional coordinates and laser reflection intensity. The registration weight is determined based on the laser reflection intensity of each target point in the target point cloud set. The relationship registration module is used to determine the target point with the smallest weighted distance as its corresponding point for each reference point in the reference point cloud based on the weighted distance calculation with introduced registration weights, so as to establish the correspondence between point pairs. The data calculation module is used to construct an objective function based on the correspondence and registration weights, solve the objective function by singular value decomposition, obtain the optimal rotation matrix and translation vector, apply the rotation matrix and translation vector to the reference point set to obtain the transformed reference point set, and the displacement vector of each reference point is its transformed coordinates minus its original coordinates. The displacement vector of each reference point is decomposed into displacement feature vectors including horizontal and vertical displacements to construct the displacement field. The calibration module is used to extract the displacement feature vector at the installation location of each displacement gauge deployed on the slope of the foundation pit from the displacement field and subtract it from the measured vector of the displacement gauge to obtain the deviation vector at the displacement gauge. The deviation vector at each displacement gauge is analyzed by Kriging space interpolation to construct a systematic deviation field covering the entire area. The systematic deviation field is then subtracted from the displacement field to obtain the calibrated displacement field. The displacement value and displacement rate in the calibrated displacement field are analyzed to determine the deformation area.

Images