Smart construction site construction safety supervision method and system based on artificial intelligence

Abstract

The present application discloses a method and system for supervision of construction safety of a smart construction site based on artificial intelligence, the method comprising: obtaining drone aerial images and point cloud data at preset time intervals; generating a digital elevation model based on the point cloud data; extracting elevation data from the digital elevation model; extracting ground object features and edge features from drone aerial images; integrating digital orthophotos, ground object features and digital elevation models to obtain vector maps; matching edge features in overlapping images in drone aerial images, and inputting them into a machine learning model to generate an aerial triangulation model; creating a stereo pair based on the aerial triangulation model to generate a real-life three-dimensional model; integrating the real-life three-dimensional model and the vector map to obtain a digital map of the construction site; obtaining personnel location data; determining risk factors, and analyzing to obtain a safety hazard analysis warning zone for the construction site, determining the safe construction distance and the construction safety risk factor of the construction area and displaying them.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of digital maps, and in particular to a method and system for smart construction site safety supervision based on artificial intelligence. Background Art

[0002] During construction, the construction site changes rapidly and the processes are frequently interspersed. Therefore, the safety risks of the construction site are also in constant dynamic change. However, the current methods and technical means of actual project safety management are relatively backward and still rely on traditional safety management methods, such as paper records and manual inspections. This will lead to untimely information updates and inability to intuitively identify high-risk areas and potential hidden dangers on the construction site. It can only react passively after an accident occurs and cannot prevent safety risks on the construction site. The safety supervision effect is extremely poor. Therefore, there is an urgent need for a method to take active precautions on the construction site and improve the effect of safety supervision on the construction site. Summary of the invention

[0003] The embodiments of the present application provide a method and system for intelligent construction site construction safety supervision based on artificial intelligence, which are used to take active precautions on the construction site and improve the effect of safety supervision on the construction site.

[0004] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0005] In a first aspect, a smart construction site construction safety supervision method based on artificial intelligence is provided, which is applied to a smart construction site construction safety supervision system, wherein the smart construction site construction safety supervision system includes a central control host and a drone, and the method includes:

[0006] At every preset time period, n drone aerial images of the construction site and point cloud data of the construction site are acquired by the drone, wherein n>2, and n is an integer, and the construction site includes at least two construction areas;

[0007] Generating a digital elevation model of the construction site according to the point cloud data;

[0008] Extracting elevation data from the digital elevation model and generating contour lines based on the elevation data;

[0009] Generate a digital orthophoto according to the n drone aerial images, and extract ground features and edge features in the drone aerial images;

[0010] Integrating the digital orthophoto, the ground features and the digital elevation model to obtain a vector map of the construction site;

[0011] Matching the edge features in the overlapping images of the n drone aerial images, and inputting the matched edge features into a pre-built machine learning model to generate an aerial triangulation model;

[0012] Creating a stereo image pair according to the aerial triangulation model, and generating a real-scene three-dimensional model according to the stereo image pair;

[0013] Integrating the real-scene three-dimensional model and the vector diagram of the construction site to obtain a digital map of the construction site, wherein the digital map of the construction site includes the contour lines;

[0014] Acquiring location data of personnel in the construction area through the safety helmets of the personnel in the construction area;

[0015] In the digital map of the construction site, risk factors are determined, and a safety hazard buffer zone analysis of the construction site is performed based on the risk factors and the personnel location data to obtain a safety hazard analysis warning zone of the construction site, wherein the safety hazard analysis warning zone of the construction site is a polygonal entity with a preset risk range;

[0016] The safe construction distance and the construction safety risk factor of the construction area are determined and displayed based on the safety hazard analysis warning area of ​​the construction site.

[0017] In a possible implementation manner of the first aspect, generating a digital elevation model of the construction site according to the point cloud data includes:

[0018] Constructing a two-dimensional grid in each of the construction areas, wherein the two-dimensional grid is composed of a plurality of grid units, and the area of ​​each of the grid units is consistent;

[0019] Calculating an average height value of each grid cell according to the point cloud data in each grid cell;

[0020] The point cloud data is converted into grid data by a preset interpolation algorithm, and each grid unit represents an average height value in the construction area;

[0021] The digital elevation model is generated according to all the grid cells.

[0022] In another possible implementation of the first aspect, integrating the digital orthophoto, the terrain features, and the digital elevation model to obtain a vector map of the construction site includes:

[0023] The coordinate system of the digital orthophoto, the coordinate system of the land feature and the coordinate system of the digital elevation model are converted into a unified coordinate system, and the common coordinates in the digital orthophoto and the land feature are used as control points;

[0024] According to the control points, affine transformation is performed on the non-common coordinates in the digital orthophoto and the ground feature to align the non-common coordinates in the digital orthophoto and the ground feature;

[0025] Extracting the geometric features of each feature from the aligned features;

[0026] If the geometric feature has a corresponding grid cell in the digital elevation model, the average height value of the grid cell is used as the height attribute of the geometric feature;

[0027] If the geometric feature does not have a corresponding grid unit in the digital elevation model, a height attribute of the geometric feature is calculated using a height estimation formula;

[0028] generating three-dimensional vector data according to the geometric features and the height attributes of the geometric features;

[0029] All the three-dimensional vector data are integrated to obtain a vector image of the construction site.

[0030] In another possible implementation manner of the first aspect, performing an affine transformation on the non-common coordinates in the digital orthophoto and the terrain feature according to the control point to align the non-common coordinates in the digital orthophoto and the terrain feature includes:

[0031] Constructing an affine transformation equation set according to at least three of the control points;

[0032] Solving the affine transformation equations using the least squares method to determine affine transformation parameters;

[0033] According to the affine transformation parameters, affine transformation is performed on the non-common coordinates in the digital orthophoto image and the terrain feature, so that the non-common coordinates in the digital orthophoto image and the terrain feature are aligned.

[0034] In another possible implementation manner of the first aspect, creating a stereo image pair according to the aerial triangulation model, and generating a real-scene three-dimensional model according to the stereo image pair includes:

[0035] According to the aerial triangulation model, randomly selecting two stereo images with overlapping areas, and calculating the disparity between the two stereo images;

[0036] Matching the overlapping areas of the two stereo images by a preset image matching algorithm to determine corresponding points of the same name in the two stereo images;

[0037] Performing stereoscopic correction on the two matched stereo images so that the points with the same name are located in the same row on the two stereo images to obtain a stereoscopic image pair;

[0038] Performing parallax calculation on the same-name points in the stereo image pair to obtain the three-dimensional coordinates of each pixel point in the two stereo images;

[0039] Three-dimensional point cloud data is generated according to the three-dimensional coordinates, and a real-scene three-dimensional model is generated according to the three-dimensional point cloud data.

[0040] In another possible implementation of the first aspect, the integration of the real-scene three-dimensional model and the vector map of the construction site to obtain a digital map of the construction site, wherein the digital map of the construction site includes the contour lines, comprises:

[0041] The vector map and the contour lines are projected onto the real-scene three-dimensional model to obtain a digital map of the construction site.

[0042] In another possible implementation of the first aspect, in the digital map of the construction site, determining a risk factor, and performing a safety hazard buffer zone analysis of the construction site according to the risk factor and the personnel location data to obtain a safety hazard analysis warning zone of the construction site, including:

[0043] Determining risk factors according to three-dimensional environmental features within a preset range of the construction area in the construction site digital map;

[0044] According to the layer of the risk factor, a safety hazard buffer zone radius is set, and a safety hazard buffer zone layer is generated;

[0045] Overlaying the potential safety hazard buffer layer with the personnel position to determine whether the personnel is located within the potential safety hazard buffer;

[0046] If the personnel is located in the potential safety hazard buffer zone, the potential safety hazard buffer zone is determined to be a potential safety hazard analysis warning zone.

[0047] In another possible implementation of the first aspect, the risk factors include terrain risk, falling object risk, and mechanical operation risk, the three-dimensional environmental features include slope features, building features, and large equipment features, and determining the risk factors according to the three-dimensional environmental features within a preset range of the construction area in the digital map of the construction site includes:

[0048] Determining whether there is a risk of water accumulation and / or a risk of sudden landform change based on the slope characteristics;

[0049] In the case where the risk of water accumulation and / or the risk of sudden terrain change exist, determining the risk factor to be the terrain risk;

[0050] Calculate the falling object risk range around the building according to the building characteristics, and determine that the risk factor is the high-altitude falling object risk when the percentage of the overlapping area between the falling object risk range and the preset range of the construction area is greater than a preset percentage;

[0051] Determine the type of the large equipment according to the large equipment characteristics, obtain the operating radius of the large equipment according to the large equipment type, and perform conflict detection between the operating radius and the building characteristics to determine whether there is a risk of collision between the large equipment and the building corresponding to the building characteristics during operation;

[0052] If there is a risk of collision between the large equipment and the building corresponding to the building characteristics during operation, the risk factor is determined to be mechanical operation risk.

[0053] In another possible implementation of the first aspect, determining the safe construction distance and the construction safety risk coefficient of the construction area according to the safety hazard analysis warning area of ​​the construction site includes:

[0054] Determine the impact range of the risk factor, and determine the safe construction distance based on the impact range and the safety hazard analysis warning area of ​​the construction site;

[0055] The number, type and preset risk level of the hazardous factors in each of the construction areas are determined, and the construction safety risk coefficient of the construction area is calculated based on the number, type and preset risk level of the hazardous factors.

[0056] In a second aspect, the present application provides a smart construction site safety supervision system, including:

[0057] Central control host;

[0058] The drone is connected to the central control host.

[0059] Through the above technical solution, the aerial images and point cloud data of the construction site are regularly obtained by drones, and the changes of the construction site can be continuously monitored. The digital elevation model is generated through the point cloud data, and the digital orthophoto is generated in combination with the aerial images of the drone. A detailed three-dimensional model of the construction site can be constructed. The three-dimensional model of the construction site not only includes the features of the objects and the edge features, but also can accurately determine the three-dimensional spatial position of each construction area through the aerial triangulation model. Combined with the contour lines and vector maps, the terrain, buildings and construction progress of the construction site can be reflected in detail. The integration of the real-life three-dimensional model and the vector map generates a digital map of the construction site that is not only accurate, but also dynamically reflects the actual situation on site. The location of the personnel in the construction area is obtained by the location data of the construction personnel's helmets. Combined with the identification of dangerous factors in the digital map of the construction site, the safety hazard buffer zone analysis can be automatically performed to detect and warn construction personnel entering high-risk areas in real time, thereby significantly reducing the probability of accidents. Through the safety hazard analysis warning area in the digital map of the construction site, the safe construction distance can be automatically determined, and the construction safety risk factor of the construction area can be calculated and displayed intuitively in the system. The safe construction distance can effectively prevent personnel and equipment from entering the dangerous area. The construction safety risk factor can quantify the risk level of the construction site, which is conducive to safety early warning.

[0060] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 A flowchart of a method for intelligent construction site safety supervision based on artificial intelligence provided in an embodiment of the present application;

[0062] Figure 2 A schematic diagram of drawing a digital map of a construction site provided in an embodiment of the present application;

[0063] Figure 3 A schematic diagram of a digital map of a construction site provided in an embodiment of the present application;

[0064] Figure 4 A schematic diagram of a potential safety hazard buffer zone provided in an embodiment of the present application;

[0065] Figure 5 A schematic diagram of the structure of a smart construction site construction safety supervision system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0067] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), such directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0068] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that the combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0069] Figure 1 The following schematically shows a flow chart of a method for intelligent construction site safety supervision based on artificial intelligence according to an embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides a smart construction site construction safety supervision method based on artificial intelligence, which is applied to a smart construction site construction safety supervision system. The smart construction site construction safety supervision system includes a central control host and a drone. The method includes:

[0070] S110, acquiring n drone aerial images and point cloud data of the construction site at predetermined intervals through the drone, wherein n>2, and n is an integer, and the construction site includes at least two construction areas;

[0071] S120, generating a digital elevation model of the construction site according to the point cloud data;

[0072] S130, extracting elevation data from the digital elevation model, and generating contour lines according to the elevation data;

[0073] S140, generating a digital orthophoto according to the n drone aerial images, and extracting ground features and edge features in the drone aerial images;

[0074] S150, integrating the digital orthophoto, the ground features and the digital elevation model to obtain a vector map of the construction site;

[0075] S160, matching edge features in overlapping images of the n drone aerial images, and inputting the matched edge features into a pre-built machine learning model to generate an aerial triangulation model;

[0076] S170, creating a stereo image pair according to the aerial triangulation model, and generating a real-scene three-dimensional model according to the stereo image pair;

[0077] S180, integrating the real-scene 3D model and the vector map of the construction site to obtain a digital map of the construction site, wherein the digital map of the construction site includes contour lines;

[0078] S190, obtaining personnel location data in the construction area through the safety helmets of the personnel in the construction area;

[0079] S1100, in the digital map of the construction site, determine the risk factors, and perform a safety hazard buffer zone analysis on the construction site based on the risk factors and personnel location data to obtain a safety hazard analysis warning zone for the construction site, where the safety hazard analysis warning zone for the construction site is a polygon entity with a preset risk range;

[0080] S1110. Analyze the warning area for safety hazards at the construction site and determine the safe construction distance and the construction safety risk factor of the construction area and display them.

[0081] Figure 2 A flowchart of drawing a digital map of a construction site provided by an embodiment of the present application is shown. Figure 2 As shown in the figure, firstly, n drone aerial images and point cloud data of the construction site are obtained by drone. n represents the number of images obtained, and n>2 means that at least three or more aerial images need to be obtained. By using drones, the actual situation of the construction site can be captured from the air in all directions and from multiple angles. The image can not only record the overall appearance of the surface, but also capture different details of the objects at different flight altitudes and angles. At the same time, the point cloud data is obtained by scanning with laser radar or other high-precision sensors, which can accurately describe the three-dimensional spatial structure of the construction site. Point cloud data is composed of a large number of points, each of which has its spatial coordinates (X, Y, Z), so it can present the ups and downs of the terrain and the three-dimensional forms of various objects in detail. By combining aerial images and point cloud data, basic data can be provided for subsequent terrain modeling and feature extraction.

[0082] Digital elevation model (DEM) is a digital expression of surface height information, which can help accurately understand the terrain changes of the construction site. When generating DEM, the point cloud data must first be processed to remove noise and unnecessary points to ensure the accuracy of the data. Then, the point cloud data is converted into a raster form through an interpolation algorithm, and each raster unit represents the average height value in a certain area. In this way, an accurate elevation model can be generated to clearly show the undulations of the terrain. The generation of DEM is very important for the terrain analysis of the construction site. It can help identify high-risk areas, such as steep slopes, depressions, etc., which may require special safety measures in actual construction.

[0083] The next step is to extract elevation data from the digital elevation model and generate contour lines based on the elevation data. Contour lines are curves that connect all points of the same height on the ground. Through contour lines, you can intuitively see the ups and downs of the terrain. The process of generating contour lines is to first extract points with specific height values ​​from the DEM, and then connect the points into continuous lines. The generated contour lines can help construction personnel better understand the terrain features, such as the slope and direction of the slope, the scope of hills or depressions, etc. By analyzing the contour lines, the path planning of construction machinery can be determined to reduce the construction risks caused by complex terrain.

[0084] Generating digital orthophotos based on n drone aerial images and extracting ground features and edge features in the images are further steps. Digital orthophotos are geometrically corrected images that eliminate image deviations caused by terrain undulations and lens distortion during the shooting process, so that they can accurately reflect the true situation of the surface. When generating digital orthophotos, it is necessary to stitch together multiple drone images through technologies such as geometric correction and image fusion. Then, computer vision technology is used to extract ground features in the image, such as buildings, roads, vegetation, etc., as well as edge features such as walls and building outlines. Feature extraction can use deep learning algorithms such as convolutional neural networks (CNNs). Through learning from a large amount of training data, the model can automatically identify and extract important information in the image.

[0085] After the orthophoto is generated and feature extracted, the digital orthophoto, feature and digital elevation model are integrated to obtain a vector map of the construction site. A vector map is a simplified representation of geographic spatial data, using geometric elements such as points, lines and surfaces to represent feature. The integration process includes combining the extracted feature with the elevation data in the digital elevation model, and reflecting elements such as buildings, roads and terrain undulations in the vector map. The integrated vector map can provide more accurate geographic information, which is helpful for construction planning, safety monitoring during construction and subsequent construction records.

[0086] Next, it is necessary to match the edge features of overlapping images in n drone aerial images, and input the matched edge features into the pre-built machine learning model to generate an aerial triangulation model. In this process, by identifying the edge features of the same objects in multiple images, algorithms such as RANSAC are applied to screen out the best matching points, and then the matching points are input into the machine learning model for further processing. The aerial triangulation model can accurately calculate the three-dimensional coordinates of each point on the construction site and construct a three-dimensional spatial framework for the entire site.

[0087] Based on the aerial triangulation model, a stereo pair is created and a real-life 3D model is generated. Specifically, a stereo pair refers to the result of capturing the same target through two images taken from different perspectives. Through stereo vision technology, a 3D image of the target can be generated. The creation of a stereo pair requires combining the matching points in the aerial triangulation model to determine the shooting perspective and distance. The generated real-life 3D model can truly reproduce the 3D structure of the construction site, so that every building, equipment and even terrain details can be presented in the model.

[0088] Integrate the real-life 3D model and the vector map of the construction site to obtain a digital map of the construction site (such as Figure 2 The digital map of the construction site shown in the figure) includes contour lines. In this process, the previously generated 3D model and vector map are integrated, and the spatial information in the 3D model is superimposed on the vector map through the matching of geographic coordinates. At the same time, the contour lines will also be superimposed on the digital map to more accurately reflect the terrain characteristics of the construction site. The integrated digital map can not only display the plane information of the surface, but also reflect the three-dimensional undulations of the terrain, so that construction personnel can understand the site environment more intuitively. The digital map of the construction site is as follows: Figure 3 shown.

[0089] After that, the location data of the personnel in the construction area is obtained through the hard hats of the workers in the construction area. Specifically, the hard hats are usually equipped with positioning devices, such as GPS modules or RFID chips, which can monitor the location of workers in real time. When workers enter or leave a specific area, the location data will be updated and transmitted to the central control device. By analyzing the location data, the distribution of workers in each construction area can be understood in real time.

[0090] In the digital map of the construction site, the risk factors are determined, and the buffer zone analysis of safety hazards at the construction site is performed based on the risk factors and personnel location data to obtain the warning zone for safety hazard analysis at the construction site. Risk factors can include steep slopes, high-altitude operation areas, heavy machinery operation areas, etc. The buffer zone analysis is based on the risk factors and their impact range, combined with the real-time location data of the workers, to calculate the areas where safety hazards may exist, and generate warning zones represented by polygonal entities. The warning zone can help construction managers to warn in advance and take measures to prevent accidents.

[0091] The last step is to determine the safe construction distance and the construction safety risk factor of the construction area according to the warning area of ​​the construction site safety hazard analysis and display. In this process, the warning area obtained by analysis will be used to calculate the minimum safe distance between the construction personnel and the dangerous area. At the same time, the construction risk is comprehensively evaluated based on various dangerous factors in the area to obtain the construction safety risk factor. This factor can help construction management personnel evaluate the safety of the current construction conditions and display it in real time through the central control equipment so that safety decisions can be made in a timely manner.

[0092] This embodiment uses a drone to regularly obtain aerial images and point cloud data of the construction site, which can continuously monitor changes in the construction site. A digital elevation model is generated through point cloud data, and a digital orthophoto is generated in combination with drone aerial images, so that a detailed three-dimensional model of the construction site can be constructed. The three-dimensional model of the construction site not only includes land features and edge features, but also can accurately determine the three-dimensional spatial position of each construction area through an aerial triangulation model. Combined with contour lines and vector maps, the topography, buildings and construction progress of the construction site can be reflected in detail. The integration of the real-life three-dimensional model and the vector map generates a digital map of the construction site that is not only accurate, but also dynamically reflects the actual situation on site. The location of personnel in the construction area is obtained through the location data of the construction personnel's helmets. Combined with the identification of dangerous factors in the digital map of the construction site, a safety hazard buffer zone analysis can be automatically performed to detect and warn construction personnel entering high-risk areas in real time, thereby significantly reducing the probability of accidents. Through the safety hazard analysis warning area in the digital map of the construction site, the safe construction distance can be automatically determined, and the construction safety risk factor of the construction area can be calculated and intuitively displayed in the system. The safe construction distance can effectively prevent personnel and equipment from entering dangerous areas, and the construction safety risk factor can quantify the risk level of the construction site, which is conducive to safety warning.

[0093] In one implementation of this embodiment, generating a digital elevation model of a construction site according to point cloud data includes the following steps:

[0094] S210, constructing a two-dimensional grid in each construction area, where the two-dimensional grid is composed of a plurality of grid units, and the area of ​​each grid unit is consistent;

[0095] S220, calculating the average height value of each grid cell according to the point cloud data in each grid cell;

[0096] S230, converting the point cloud data into grid data by a preset interpolation algorithm, wherein each grid unit represents an average height value in the construction area;

[0097] S240: Generate a digital elevation model based on all grid cells.

[0098] In the process of generating a digital elevation model of a construction site based on point cloud data, it is first necessary to construct a two-dimensional grid in each construction area. The two-dimensional grid consists of multiple grid cells, and each grid cell represents a fixed surface area in the construction area. In order to ensure the uniformity of data processing and simplify the subsequent calculation process, the area of ​​each grid cell is usually set to be consistent. This means that all grid cells in the entire construction area have the same length and width. For example, each grid cell is set to 1 meter × 1 meter or 0.5 meter × 0.5 meter, so that the entire construction area is divided into a regular matrix. In specific implementations, the density of the grid can be adjusted by setting the size and shape of the grid, so as to accurately represent the surface details or roughly summarize the terrain features according to actual needs.

[0099] Next, it is necessary to calculate the average height value of each grid cell based on the point cloud data in each grid cell. Point cloud data usually contains a large number of three-dimensional coordinate points in the construction area, and each coordinate point contains three components of X, Y, and Z, which respectively represent the plane position and height of the point. When calculating the average height of a grid cell, it is necessary to consider all Z coordinate values ​​in the cell. The specific implementation method can be carried out through the following process: First, find all the point cloud data points that fall within the range of the grid cell, then add up the Z coordinate values ​​of the points and divide them by the number of points in the grid cell to obtain the average height value of the grid cell. If there are no point cloud data points in some grid cells, they can be estimated by interpolation or other methods to ensure that each grid cell has a reasonable height value.

[0100] After calculating the average height value of each grid cell, the next step is to convert the point cloud data into raster data through a preset interpolation algorithm. The selection and application of the interpolation algorithm is crucial to generating high-quality raster data. Common interpolation algorithms include inverse distance weighted (IDW) and Kriging interpolation (Kriging). The algorithm infers the height value of the unknown point by analyzing and calculating the known data points, thereby generating a continuous elevation model. Specifically, the inverse distance weighted interpolation method assumes that the data points with a closer distance have a greater impact on the interpolation result, while the data points with a farther distance have a smaller impact. Therefore, when performing interpolation calculations, a greater weight is given to the points with a closer distance to calculate the height value of the grid cell. The Kriging interpolation method is based on statistical principles and constructs a variation function model to perform more accurate interpolation calculations on spatial data. Regardless of the interpolation method used, the ultimate goal is to generate a complete set of raster data from point cloud data. Each raster cell contains the average height information in the construction area, which provides basic data for generating a digital elevation model.

[0101] Finally, a digital elevation model is generated based on all the grid cells. A digital elevation model (DEM) is a two-dimensional or three-dimensional digital surface that can represent the undulations of the terrain. The raster data generated by the above steps can visualize the terrain height information of the construction area as a continuous surface model. When generating a DEM, the data of each grid cell needs to be smoothed to eliminate noise or outliers and ensure the smoothness and continuity of the model. Specifically, the raster data can be processed by applying a Gaussian smoothing algorithm to reduce the impact of errors and make the digital elevation model more accurate.

[0102] In summary, the process of constructing a two-dimensional grid, calculating the average height of grid cells, using interpolation algorithms to generate grid data, and finally generating a digital elevation model is a complete and detailed technical process for converting point cloud data into construction site elevation information. Each step is closely linked. The DEM generated by this method is not only suitable for traditional construction scenarios, but also can provide important digital support for the construction and management of smart construction sites.

[0103] This implementation method effectively ensures that the final digital elevation model can accurately and comprehensively reflect the terrain characteristics of the construction area, thereby effectively ensuring the safety monitoring of the construction process, by constructing a two-dimensional grid, calculating the average height of the grid cells, and using an interpolation algorithm to generate grid data to finally generate a digital elevation model.

[0104] In one implementation of this embodiment, the digital orthophoto, the ground feature and the digital elevation model are integrated to obtain a vector map of the construction site, including the following steps:

[0105] S310, converting the coordinate system of the digital orthophoto, the coordinate system of the land feature and the coordinate system of the digital elevation model into a unified coordinate system, and using the common coordinates of the digital orthophoto and the land feature as control points;

[0106] S320, performing affine transformation on the non-common coordinates in the digital orthophoto and the ground feature according to the control points, so as to align the non-common coordinates in the digital orthophoto and the ground feature;

[0107] S330, extracting the geometric features of each feature from the aligned feature features;

[0108] S340, if the geometric feature has a corresponding grid cell in the digital elevation model, taking the average height value of the grid cell as the height attribute of the geometric feature;

[0109] S350, if the geometric feature does not have a corresponding grid unit in the digital elevation model, a height attribute of the geometric feature is calculated using a height estimation formula;

[0110] S360, generating three-dimensional vector data according to the geometric features and the height attributes of the geometric features;

[0111] S370. Integrate all three-dimensional vector data to obtain a vector diagram of the construction site.

[0112] Convert the coordinate systems of digital orthophotos, features, and digital elevation models into a unified coordinate system to ensure spatial consistency between different data sources. Each data type (digital orthophotos, features, digital elevation models) typically uses a different coordinate system, such as UTM, WGS84, or geodetic coordinate system. If the data is not unified into the same coordinate system, there will be deviations in the spatial relationship between the data, making it difficult to accurately align and integrate the data in subsequent steps. Through the coordinate conversion algorithm, the coordinate systems of all data are converted into a unified coordinate system, such as WGS84, to ensure that they are represented in the same spatial reference system.

[0113] After the coordinate system conversion is completed, the common coordinates in the digital orthophoto and the feature are selected as control points. Control points refer to points with consistent positions in different data sources. They are used for subsequent affine transformation to ensure the geometric alignment of the data. In actual operation, obvious geographic feature points, such as building corners and road intersections, can be selected as control points. The control points can be clearly identified in the image and feature data. By comparing the positions of the control points, the geometric deviation between the data can be detected and corrected.

[0114] According to the selected control points, the non-common coordinates in the digital orthophoto and the features are affine transformed. Affine transformation is a geometric transformation used to correct the deviation between data so that they are aligned in space. The specific operations of affine transformation include translation, rotation, scaling and shearing. By using the coordinates of the control points, the affine transformation matrix is ​​constructed to make corresponding geometric adjustments to the non-common coordinate points (i.e., those points that do not exist in the image and the features) to make them consistent with the control points. The purpose of this step is to eliminate the geometric distortion or deviation caused by different data collection methods, thereby ensuring the spatial alignment of all data in a unified coordinate system.

[0115] After completing the data alignment, the geometric features of each feature are extracted from the aligned features. Geometric features include points, lines, and surfaces, which describe the spatial form of the feature. The way to extract geometric features usually relies on spatial analysis algorithms, such as using edge detection technology to identify the outline of a building, or using cluster analysis to identify linear features such as roads and rivers. The extraction of geometric features not only involves the identification of spatial coordinates, but also includes the extraction of attribute information, such as the type of building, the width of the road, etc. If the extracted geometric features have corresponding grid cells in the digital elevation model (DEM), the average height value of the grid cell is used as the height attribute of the geometric feature. The digital elevation model is composed of regular grids, and each grid cell represents the average elevation of a specific area. In this process, it is necessary to first match the geometric features of the feature with the grid position in the DEM. The nearest neighbor method or bilinear interpolation method is usually used to determine the grid cell corresponding to the geometric feature point. Then, the average elevation of all points in the cell is calculated and used as the height attribute of the feature. The main effect of this operation is to add height information to the feature, so that the feature can be expanded from a two-dimensional plane to a three-dimensional form.

[0116] If the extracted geometric feature does not have a corresponding grid cell in the digital elevation model, due to the lack of direct elevation data, this embodiment uses a height estimation formula to calculate the height attribute of the geometric feature. Specifically, the height attribute (height value) of the geometric feature can be calculated using the height estimation formula based on the elevation data of the adjacent grid cells to approximate the actual terrain height as much as possible, thereby giving reasonable three-dimensional attributes to the ground feature that lacks direct elevation data. The height estimation formula of this embodiment is:

[0117] ;

[0118] Among them, z is the estimated height value, z i is the elevation value of the adjacent grid, w i is the nearest elevation value.

[0119] Generate 3D vector data based on the extracted geometric features and their height attributes. In this step, all the geometric features of the objects are combined with their corresponding height information to generate complete 3D vector data through 3D modeling technology. Vector data can be expressed in the form of polygonal meshes, point clouds, etc. to represent the shape and position of each object in 3D space. For example, the outline and height of a building can be represented by a polygonal cylinder, and a road can be represented by a line segment with a height attribute. The generated 3D vector data can accurately reflect the real spatial position and geometric shape of each object on the construction site.

[0120] Finally, by integrating all the 3D vector data, a complete construction site vector map is obtained. Specifically, different categories of 3D features, such as buildings, roads, terrain and other data, are integrated to form a unified 3D scene. The integration process is used to ensure the consistency of the height of the building and the ground, and to deal with the spatial overlap between objects. The final generated construction site vector map is a high-precision 3D model that contains not only the geometric information of all objects, but also the height information and spatial relationship of the objects.

[0121] In summary, through the implementation of the above steps, the final generated construction site vector map has high-precision spatial positioning and three-dimensional display capabilities.

[0122] The vector map of this implementation method can not only accurately reflect the distribution and form of objects on the construction site, but also display the height information of each object, making the spatial layout of the entire construction site clearer and more intuitive. High-precision three-dimensional vector maps can provide important references for construction planning, reduce errors in construction, improve construction efficiency, and ultimately help achieve safer and more efficient project management.

[0123] In one implementation of this embodiment, affine transformation is performed on the non-common coordinates in the digital orthophoto and the ground feature according to the control points to align the non-common coordinates in the digital orthophoto and the ground feature, including the following steps:

[0124] S410, constructing an affine transformation equation group according to at least three control points;

[0125] S420, using the least square method to solve the affine transformation equations to determine the affine transformation parameters;

[0126] S430, performing affine transformation on the non-common coordinates in the digital orthophoto and the ground feature according to the affine transformation parameters, so as to align the non-common coordinates in the digital orthophoto and the ground feature.

[0127] An affine transformation is a linear transformation used to map one coordinate system in a two-dimensional space to another coordinate system. To perform an affine transformation, you first need to construct a set of equations based on at least three control points. The control points should be located on the digital orthophoto and the features, and should be clearly identifiable in both coordinate systems. Each control point represents a specific location, and they are represented by their own coordinates in the digital orthophoto and the features. Assume that the original coordinates of the three selected control points are , the corresponding target coordinates are The mathematical expression of affine transformation is:

[0128] ;

[0129] Where (x, y) is the original coordinate, (x', y') is the transformed coordinate, and a, b, c, d, e, f are the six parameters of the affine transformation. The parameters control the scaling, rotation, translation, and shearing of the image. By substituting the original coordinates of each control point into the above formula, the following linear equations can be obtained:

[0130] ;

[0131] Representing the equation in matrix form, we can construct the following matrix equation:

[0132] ;

[0133] In the matrix equation, the matrix on the left represents the original coordinates of the control point, and the vector on the right is the corresponding target coordinate. Through this system of equations, a pair of equations can be established for each control point, thus forming a linear system for solving the six parameters of the affine transformation. It is necessary to select at least three control points, because only in this way can a solvable system of equations be obtained, thus ensuring that the affine transformation has a unique solution.

[0134] After constructing the above affine transformation equations, the next step is to use the least squares method to solve the equations and determine the parameters of the affine transformation. The least squares method is a statistical method used to minimize the sum of squared errors in data fitting to find the optimal solution. In the context of affine transformation, the least squares method is used to minimize the sum of squared distances between the transformed control point coordinates and the target coordinates. To this end, the above matrix equation can be expressed as:

[0135] ;

[0136] Among them, A is a known matrix containing the original coordinates of the control points, P is the six parameter vectors of the affine transformation, and Y is the target coordinate vector of the control points. The goal of the least squares method is to solve P so that the equation error is minimized. The sum of squares of the error can be expressed as:

[0137] ;

[0138] By taking the derivative of the error function and setting it to zero, we get the following normal equation:

[0139] ;

[0140] Solving this equation, we get the parameter vector P:

[0141] ;

[0142] In the above formula is the transpose of matrix A, is its inverse matrix. In summary, the six parameters of affine transformation can be calculated: a, b, c, d, e, f. Parameters play a key role in affine transformation and directly affect the alignment accuracy between digital orthophotos and ground features. The advantage of using the least squares method is that it can effectively process the noise and errors in the control point data and find the optimal affine transformation parameters to minimize the errors between control points, thereby ensuring the accuracy and stability of image calibration.

[0143] After obtaining the parameters of the affine transformation, apply it to all non-common coordinates in the digital orthophoto and the features to align the two. The essence of the affine transformation is to map the points of the original coordinate system to the target coordinate system. Using the parameters (a, b, c, d, e, f) solved above, we can align each non-common point. Perform coordinate transformation. The transformation can be expressed as:

[0144] ;

[0145] For each pixel coordinate in the image Apply the transformation formula to calculate its new position (x i ',y i '). In this way, all coordinate points between the digital orthophoto and the ground features are remapped to ensure that the two are aligned in a unified coordinate system. To further ensure the accuracy of the alignment, the transformed coordinates can be compared with the expected target coordinates and the residuals can be calculated:

[0146] ;

[0147] By calculating and analyzing the residual, the accuracy and effect of the affine transformation can be evaluated. Generally speaking, if the residual is less than a preset threshold, the transformation is considered successful. Otherwise, it may be necessary to reselect control points or adjust the transformation parameters to achieve a better alignment effect.

[0148] This implementation method realizes the affine transformation and alignment between digital orthophotos and ground features. In the whole process, the construction of the affine transformation equation group ensures the rationality of the mathematical model of the transformation, and the use of the least squares method to solve the problem ensures the optimality of the parameters and the robustness to noise. Finally, the precise alignment of spatial data is achieved by aligning the coordinates by applying the transformation parameters, which can significantly improve the accuracy and reliability of image and feature data processing in geographic information systems.

[0149] In one implementation of this embodiment, creating a stereo pair of images based on an aerial triangulation model, and generating a real-scene three-dimensional model based on the stereo pair of images includes the following steps:

[0150] S510, randomly selecting two stereo images with overlapping areas according to the aerial triangulation model, and calculating the parallax between the two stereo images;

[0151] S520, matching the overlapping areas of the two stereo images by a preset image matching algorithm to determine corresponding points of the same name in the two stereo images;

[0152] S530, performing stereoscopic correction on the two matched stereo images so that the points with the same name are located in the same row on the two stereo images to obtain a stereoscopic image pair;

[0153] S540, performing parallax calculation on the same-name points in the stereo image pair to obtain the three-dimensional coordinates of each pixel point in the two stereo images;

[0154] S550, generating three-dimensional point cloud data according to the three-dimensional coordinates, and generating a real-scene three-dimensional model according to the three-dimensional point cloud data.

[0155] First, the aerial triangulation model uses the geometric information in the aerial images to accurately calculate the spatial coordinates of the objects in each image by analyzing the relative position relationship between the images. The construction of the model needs to consider the internal and external orientation parameters of the image. The internal orientation parameters include the focal length, principal point position and lens distortion coefficient of the camera, while the external orientation parameters involve the position and posture of the camera during shooting. Through the parameters, the pixel coordinates of the image can be converted into the coordinates of the object in three-dimensional space. After the aerial triangulation model is established, two images with sufficient overlapping areas can be selected for subsequent processing.

[0156] Next, when constructing a stereo pair, two stereo images with sufficient overlapping areas are randomly selected from the aerial triangulation model. The images must be captured with an appropriate overlap range, usually 60%-80%, to ensure the accuracy of stereo matching and disparity calculation. The image content in the overlapping area must be as consistent as possible, which can reduce the parallax between the images and ensure that the position difference of the same object in the two images is as small as possible in subsequent image matching. In the process, the images can be corrected to the same scale and resolution through geometric transformation, which facilitates disparity calculation and matching of same-name points. The calculation of disparity is based on the difference in pixel position of the same object point in the two images. Through the triangulation formula, the difference is converted into the depth information of the object, that is, the distance of the object from the camera. The larger the parallax, the closer the object is to the camera, and vice versa.

[0157] In the image matching stage, the overlapping areas of the two stereo images are matched point by point through preset image matching algorithms, such as SIFT (Scale Invariant Feature Transform) and SURF (Speeded Up Robust Features). The core of the matching process is to identify and extract feature points in the image. Feature points are usually corners, edges or texture-rich areas in the image. The matching algorithm will first identify a large number of feature points in one image, and then find the corresponding feature points in another image, and finally form pairs of points with the same name. In this process, it is necessary to overcome interference factors such as noise and illumination changes to ensure the accuracy of the matching. The number and accuracy of the pairs of points with the same name directly affect the effect of subsequent 3D reconstruction. Therefore, multiple iterative matching and error correction methods are usually used to improve the matching accuracy to ensure that the final pairs of points with the same name are evenly distributed and accurately matched.

[0158] In the stereo correction step, the matched stereo images will undergo geometric transformation and be adjusted to the ideal stereo geometric relationship. Specifically, the purpose of stereo correction is to align the same-name points of the two images in the vertical direction so that they are located in the same row in the stereo pair. The advantage of doing this is that it greatly simplifies the process of disparity calculation, making the subsequent depth information extraction more intuitive and efficient. The correction process usually uses affine transformation or projective transformation methods to resample and interpolate the image so that the geometric structure of the image meets the requirements of stereo vision. The corrected image not only visually presents an ideal stereo effect, but also provides a convenient basis for subsequent disparity calculations.

[0159] In the disparity calculation step, the disparity value of each same-name point is calculated using the corrected stereo image pair through a stereo matching algorithm. The calculation of the disparity value is based on the geometric relationship between the images, usually using the principle of binocular stereo vision, that is, based on the baseline distance of the camera (the difference in shooting position of the two images) and the disparity value, the three-dimensional coordinates of each pixel point are calculated using the triangulation formula. The specific formula is:

[0160] ;

[0161] Where Z is the depth of the object (the distance from the camera), f is the focal length of the camera, B is the baseline distance of the camera, and d is the parallax value. By calculating pixel by pixel, we can get the depth map of the entire image, which represents the three-dimensional spatial position of each point in the image.

[0162] In the process of generating 3D point cloud data, the depth information obtained by the above calculation is used to combine the 2D coordinates of each pixel (the position in the image) with its corresponding depth value and convert it into a point in 3D space. The points constitute a 3D point cloud. Point cloud is a common form of 3D data representation, in which each point has its corresponding 3D coordinates. In order to further improve the quality of 3D point cloud, post-processing of point cloud is usually performed, including removing noise points, filling holes, and improving the smoothness and accuracy of point cloud through filtering technology. The processed point cloud is denser and has higher accuracy, which can accurately reflect the shape and spatial distribution of objects in the actual scene.

[0163] Based on the processed 3D point cloud data, the last step is to generate a real-life 3D model. This process usually uses a surface reconstruction algorithm to convert discrete point cloud data into a continuous 3D surface model. Common surface reconstruction algorithms include Delaunay triangulation and Poisson reconstruction. The algorithm connects neighboring points in the point cloud to generate a triangular mesh or other geometric surface, making the surface of the 3D model continuous and smooth. After the surface model is generated, the texture information (i.e., color and details) of the original image can be superimposed on the 3D model through texture mapping technology, making the model more realistic in visual effect. The real-life 3D model generated by this process can intuitively display the spatial structure of the terrain and buildings, and because of the use of real image textures, the visual effect of the model is close to the real scene.

[0164] This implementation starts with an aerial triangulation model, and after stereo image selection, image matching, stereo correction, parallax calculation and point cloud generation, a high-precision real-scene 3D model is finally generated. The real-scene 3D model has high-precision geometric characteristics, can accurately reflect the spatial structure of the real world, and can provide detailed 3D descriptions for large-scale terrain or buildings in a relatively short period of time, greatly improving the efficiency and quality of spatial data acquisition.

[0165] In one implementation of this embodiment, the real-scene 3D model and the vector map of the construction site are integrated to obtain a digital map of the construction site, wherein the digital map of the construction site includes contour lines, including the following steps:

[0166] S610, projecting the vector map and the contour lines into the real-scene three-dimensional model to obtain a digital map of the construction site.

[0167] First, vector maps and contour lines are projected into the real-life 3D model. Vector maps are usually two-dimensional graphic data that represent key elements in the construction site, such as the boundaries of buildings, the routes of roads, and the paths of pipelines. Contour lines are lines that represent changes in terrain elevation and are usually generated from elevation data. The purpose of projection is to integrate the two-dimensional data with the three-dimensional model so that vector maps and contour lines can be accurately displayed in three-dimensional space. Specifically, the projection process requires mapping each point of the vector map and contour lines to the surface of the three-dimensional model through coordinate transformation. First, it is necessary to ensure that the three-dimensional model uses the same coordinate system as the vector map and contour lines. If the coordinate system is inconsistent, it may cause data misalignment or distortion. Therefore, coordinate system transformation must be performed, usually using coordinate transformation tools in the geographic information system (GIS). Under the premise of consistent coordinate systems, the projection algorithm is then used to transform the coordinates of each point in the two-dimensional vector map to the corresponding point in the three-dimensional model. This process involves interpolation calculations on the surface of the three-dimensional model to ensure that the projected vector map and contour lines accurately match the terrain of the three-dimensional model. The projected vector diagrams and contour lines will be displayed on the surface of the 3D model and become part of the 3D space, so that the information that originally existed only on the 2D plane can accurately reflect the terrain and structural layout of the construction site in a 3D environment.

[0168] During the projection process, the specific implementation method can be realized through software, such as using professional tools such as ArcGIS, QGIS or AutoCAD. Taking ArcGIS as an example, first import the real-life 3D model, vector map and contour data, then use ArcGIS's "Project to Surface" function, select the vector map and contour data as the projection object, and select the 3D model as the target surface for projection. The projection algorithm will automatically calculate the corresponding position of each vector point and contour line on the surface of the 3D model, and generate a 3D displayed vector map and contour line. At this point, the digital map of the construction site has a preliminary prototype, and all two-dimensional geographic information has been embedded in the 3D model.

[0169] This implementation can integrate two-dimensional geographic information with the terrain model in three-dimensional space to generate a digital map of the construction site containing contour lines. The digital map of the construction site can not only visually present the overall picture of the construction site, but also has accurate elevation and location information, which provides important support for project planning, monitoring of construction progress, and management of the construction site. Through the method, construction management personnel can more clearly understand the terrain changes and building layout on site, so as to better plan construction steps, optimize resource allocation, and promptly respond to various problems that may arise during the construction process. The level of digital management of the construction site has been significantly improved.

[0170] In one implementation of this embodiment, in a digital map of a construction site, a risk factor is determined, and a safety hazard buffer zone analysis of the construction site is performed based on the risk factor and personnel location data to obtain a safety hazard analysis warning zone of the construction site, including the following steps:

[0171] S710, determining the risk factors according to the three-dimensional environmental characteristics within a preset range of the construction area in the digital map of the construction site;

[0172] S720, according to the layer of the risk factor, setting the radius of the safety hazard buffer zone, and generating a safety hazard buffer zone layer;

[0173] S730, superimposing the safety hazard buffer layer and the personnel position to determine whether the personnel is located in the safety hazard buffer;

[0174] S740: If the personnel are located in the safety hazard buffer zone, determine the safety hazard buffer zone as a safety hazard analysis warning zone.

[0175] First, in the digital map of the construction site, it is necessary to determine the possible risk factors based on the three-dimensional environmental characteristics within the preset range of the construction area. The three-dimensional environmental characteristics include but are not limited to the ups and downs of the terrain, the location and height of the buildings, the direction of the roads, the layout of the underground pipelines, and the distribution of mechanical equipment. The characteristics directly affect the possible risk factors that may arise during the construction process. For example, there may be a risk of collapse in areas with large terrain undulations, there may be a risk of falling objects around high-rise buildings, and there may be safety hazards in excavation operations near underground pipelines.

[0176] In actual operation, it is first necessary to collect three-dimensional environmental data of the construction site, which can be obtained through technologies such as drone aerial photography, laser scanning or geographic information system (GIS). The data is then imported into the digital map of the construction site and processed through spatial analysis tools. For example, through GIS tools, the height difference of the terrain can be automatically analyzed to identify possible landslides or collapse areas; through the three-dimensional models of buildings and equipment, areas where there may be a risk of falling objects can be determined. Each identified risk factor needs to be marked on the digital map and given corresponding attribute information, such as the type of risk, the level of risk, the possible scope of impact, etc.

[0177] After the risk factors are determined, the radius of the safety hazard buffer zone needs to be set according to the nature of each risk factor, and the corresponding safety hazard buffer zone layer needs to be generated. The radius of the buffer zone determines the impact range of the risk factor, and is usually determined based on relevant safety standards, equipment operating specifications, and specific construction conditions. For example, for the risk of falling objects from high altitudes, the radius of the buffer zone may be set based on the height of the building; for the operational risk of mechanical equipment, the radius of the buffer zone may be determined based on the operating radius and safety distance of the equipment.

[0178] When implementing this, you can use the buffer generation tool in GIS software. First, set the corresponding buffer radius based on the attribute information of each hazard factor. For example, the falling object risk area of ​​a building may be set to a radius of 10 meters. Then, use the buffer generation tool to generate a circular or irregular buffer layer with the coordinates of the hazard factor as the center, such as Figure 4 As shown in the figure, Figure 1 is a point-shaped safety hazard buffer zone, Figure 2 is a line-shaped safety hazard buffer zone, and Figure 3 is a surface-shaped safety hazard buffer zone. The buffer zone layer is represented by different colors or line types in the digital map to facilitate identification by construction management personnel.

[0179] In the process of generating buffer zones, multiple factors need to be considered, including the overlap between buffer zones, the relationship between buffer zones and existing buildings or facilities, etc. If there is overlap between buffer zones, the weighted average method can be used to calculate the hazard level of the overlapping area and give special marks in the layer. The generated buffer layer not only shows the impact range of each hazard factor, but also provides basic data support for subsequent safety hazard analysis.

[0180] After generating the safety hazard buffer layer, the next step is to overlay the layer with real-time personnel location data to determine whether construction workers are in the danger zone. Personnel location data is usually obtained through GPS positioning devices, mobile devices or RFID tags and updated in real time on the digital map.

[0181] The process of overlay analysis involves spatial position matching and data query. Specifically, in GIS software, spatial query tools can be used to determine whether each person's location point falls within a buffer zone. This process is usually automated, and through simple conditional judgment, if the coordinates of the person's location point coincide with the geometric range of a buffer zone, it means that the person is in a dangerous area.

[0182] During implementation, the activity trajectory and predicted location of personnel can also be considered to provide early warning of potential dangers. For example, by analyzing the speed and direction of movement of personnel, their possible future location can be predicted and it can be determined whether they will enter a certain danger buffer zone.

[0183] When a person is detected in a potential safety hazard buffer zone, the buffer zone is identified as a potential safety hazard analysis warning zone. This means that there is an actual safety risk in the area and immediate attention needs to be paid and appropriate safety protection measures should be taken. Warning zones are usually marked in eye-catching colors on digital maps and may be accompanied by alarm prompts or other warning signals.

[0184] In addition to spatial overlay analysis, the process of generating warning zones can also classify and prioritize warning zones based on risk levels and buffer types. For example, some high-risk warning zones may require immediate evacuation, while other low-risk areas may only require enhanced monitoring. For multi-level warning zone management, different identification colors or warning symbols can be used so that managers can quickly identify and respond.

[0185] In actual applications, the generation of warning zones is not only a response to real-time data, but can also be integrated into the safety management system of the entire construction site, automatically generating alarm records, notifying relevant persons in charge, and generating corresponding safety reports. The integrated management method can significantly improve the safety level of the construction site and reduce the occurrence of safety accidents.

[0186] The digital map of the construction site in this implementation method successfully generates a safety hazard analysis warning zone based on risk factors and personnel location data. The warning zone can not only intuitively display the areas with potential safety risks in the construction site, but also reflect the activities of personnel in real time, providing timely safety warnings for construction management personnel. The safety management method based on three-dimensional environmental characteristics, buffer zone analysis and spatial superposition can significantly improve the safety management efficiency of the construction site, effectively prevent and respond to various safety hazards, and provide important technical guarantees for the safety of personnel during the construction process.

[0187] In one implementation of this embodiment, the risk factors include terrain risk, high-altitude falling object risk and mechanical operation risk, and the three-dimensional environmental characteristics include slope characteristics, building characteristics and large equipment characteristics. According to the three-dimensional environmental characteristics within a preset range of the construction area in the digital map of the construction site, determining the risk factors includes the following steps:

[0188] S810. Determine whether there is a risk of water accumulation and / or a risk of sudden landform change based on slope characteristics;

[0189] S820: In the case of water accumulation risk and / or terrain abrupt change risk, the risk factor is determined to be terrain risk;

[0190] S830, calculating the falling object risk range around the building according to the building characteristics, and determining the risk factor as high-altitude falling object risk when the percentage of the overlapping area between the falling object risk range and the preset range of the construction area to the preset range of the construction area is greater than a preset percentage;

[0191] S840, determining the type of the large equipment according to the characteristics of the large equipment, obtaining the operating radius of the large equipment according to the type of the large equipment, and performing a conflict detection between the operating radius and the building characteristics to determine whether there is a risk of collision between the large equipment and the building corresponding to the building characteristics during operation;

[0192] S850. If there is a risk of collision between a large piece of equipment and a building corresponding to the building characteristics during operation, the risk factor is determined to be mechanical operation risk.

[0193] First, the digital map of the construction site provides detailed three-dimensional environmental features within the construction area, including slope features. Slope features can be obtained by analyzing the terrain data of the three-dimensional model. The analysis of slope features can use the terrain slope map method to determine the overall slope change of the area by calculating the local slope of each point. For example, the slope calculation formula can be used:

[0194] ;

[0195] in, Indicates height change, and Respectively represent the change in horizontal distance. Through this method, a slope distribution map of a region can be generated. If the slope of some areas is small and in a low-lying position, the area may be at risk of water accumulation under the influence of rainfall or other water sources. Similarly, if the slope of some areas changes suddenly, such as from gentle to steep, the area may be at risk of sudden terrain changes. Such sudden changes are usually potential areas for geological disasters such as landslides and collapses.

[0196] The assessment of waterlogging risk can be further combined with rainfall forecasts and on-site drainage system layout. For example, using geographic information system (GIS) tools, surface water flow paths and waterlogging areas under different rainfall amounts can be simulated. Through simulation, areas prone to waterlogging can be identified, thereby determining whether there is a risk of waterlogging in the area.

[0197] Once it is determined that there is a risk of water accumulation or abrupt topographic changes, it is identified as a topographic risk area. During construction, the area requires special attention and measures such as drainage channel construction or slope adjustment can be taken to reduce the risk.

[0198] The analysis of building characteristics is also based on three-dimensional digital maps, and the characteristics include the building's height, structure type, and location. The calculation of the risk of falling objects from high altitudes first requires determining the height of the building and the safe distance around it. The risk range of falling objects from a building is usually related to the height and structural form of the building. For example, for a high-rise building, its risk range of falling objects can be calculated as 1.5 to 2 times its height, which covers the possible falling area of ​​falling objects from high altitudes under the action of gravity.

[0199] The falling object risk range can be calculated using a geometric formula. For example, assuming the height of a building is H, its falling object risk range R can be expressed as:

[0200] ;

[0201] Among them, k is an empirical coefficient, which is usually between 1.5 and 2. This means that the falling object risk area around the building is a circular area centered on the building with a radius of R.

[0202] Next, the overlap area between the risk range and the preset range of the construction area needs to be calculated. If the percentage of the overlap area to the preset range of the construction area is greater than a preset safety threshold (for example, 20%), then this area is considered to have a risk of falling objects from high altitude. This method helps to identify and prevent risks that may cause injuries to construction workers or damage to equipment.

[0203] During the implementation process, managers can use drones or laser scanning and other technical means to regularly update the three-dimensional feature information of the building to ensure the accuracy of the risk range calculation. This method can achieve efficient risk assessment and monitoring and reduce the safety hazards caused by the risk of falling objects from high altitudes.

[0204] Large equipment feature analysis is to identify and prevent risks caused by equipment operation during construction. The features of large equipment include the type, size, operating radius, and movement path of the equipment. Based on the features, the possible impact range of the equipment during operation can be calculated, such as the boom length and rotation radius of the crane.

[0205] The operating radius of the equipment can be calculated in the following way: For a typical tower crane equipment, the operating radius can be expressed as the maximum length of the boom plus a safety buffer distance.

[0206] Once the operating radius of the equipment is determined, it needs to be represented as an area in the 3D environment model and the overlap of this area with other building features is detected. Through spatial analysis with building features, it is determined whether large equipment will have a collision risk with the building during operation. If there is an overlapping area and there is not enough space within the area to avoid collision, it is considered to be a risk of mechanical operation.

[0207] In order to improve the accuracy of conflict detection, a real-time data feedback system can be used to monitor the real-time position and operating radius of large equipment through sensors installed on the equipment. Combined with the dynamic update of the three-dimensional model, real-time monitoring and early warning of equipment operation risks can be achieved.

[0208] This implementation method is based on the three-dimensional environmental characteristics of the construction site, combined with slope characteristics, building characteristics and large equipment characteristics, and can accurately calculate and determine the existence and scope of various risks. For the risks of water accumulation and sudden terrain changes, through detailed analysis and simulation of terrain data, potential geological disasters and safety issues can be identified and prevented in advance. For the risk of falling objects from high altitudes, through analysis of building heights and surrounding environment, potential areas affected by falling objects are determined to ensure construction safety. For mechanical operation risks, through spatial conflict detection of large equipment operating radius and building characteristics, potential safety hazards caused by equipment operation are prevented. The hazard factor identification method based on three-dimensional digital maps can effectively improve the safety management level of the construction site, reduce the occurrence of safety accidents, and ensure the smooth progress of construction.

[0209] In one implementation of this embodiment, determining the safe construction distance and the construction safety risk factor of the construction area according to the safety hazard analysis warning area of ​​the construction site includes the following steps:

[0210] S910. Determine the impact range of the risk factor and determine the safe construction distance based on the impact range and the safety hazard analysis warning area of ​​the construction site;

[0211] S920. Determine the number, type and preset risk level of the hazardous factors in each construction area, and calculate the construction safety risk coefficient of the construction area based on the number, type and preset risk level of the hazardous factors.

[0212] In the safety management of construction sites, it is first necessary to clarify the impact range of various risk factors. Risk factors may include terrain risks, falling objects risks, and mechanical operation risks. The determination of the impact range usually relies on a comprehensive analysis of the three-dimensional environmental characteristics of the construction site, such as obtaining detailed terrain and building data of the construction site through GIS (Geographic Information System) technology or drone remote sensing. During the analysis process, geometric and physical models can be used to calculate the impact area of ​​each risk factor. For example, for the risk of falling objects from high altitudes, the potential impact area can be calculated based on the height of the building and its falling object trajectory, which is usually simulated by combining the free fall formula with factors such as wind speed.

[0213] After the impact range is determined, the next step is to set the safe construction distance based on the range and the safety hazard analysis warning area of ​​the construction site. The calculation of the construction distance needs to take into account the cumulative effect of each risk factor and the superposition effect between different risk factors. In order to ensure construction safety, the safe construction distance should at least cover the maximum impact range of all determined risk factors, and add a certain safety buffer zone on this basis. For example, for the high-altitude falling object area, if the impact range is 20 meters, the safe construction distance can be set to 25 meters to increase safety. The determination of the impact range and safe construction distance usually needs to be adjusted in combination with the actual situation on site, relevant industry standards and the specific requirements of the construction project to ensure the safety of construction personnel and equipment.

[0214] After clarifying the impact range of various risk factors and setting the safe construction distance, the next step is to conduct a quantitative analysis of the risk factors in each construction area. The core of this step is to determine the number, type and preset risk level of risk factors in each construction area. The number of risk factors can be determined through the results of preliminary data collection and analysis. For example, by analyzing the terrain, building characteristics and equipment layout in the construction area, the number of terrain risk points, high-altitude falling object risk areas and mechanical operation risk areas in each area can be identified.

[0215] The types of risk factors are classified according to their specific properties, such as the aforementioned terrain risk, falling object risk, and mechanical operation risk. Each type of risk factor may have different risk impacts on the construction area, so it is necessary to identify and evaluate its risk level separately. The preset risk level can be based on historical data, industry standards, or expert experience, and is usually divided into three levels: low, medium, and high. For example, if the terrain in a certain area changes frequently and there is a lot of water accumulation, the terrain risk level of the area may be assessed as high.

[0216] After determining the number, type and preset risk level of the risk factors, the construction safety risk factor of the construction area can be further calculated. The calculation formula of the risk factor can be expressed as:

[0217] ;

[0218] Among them, n represents the total number of risk factors, the number of risk factors i represents the number of risk factors of the i-th category, the risk level i represents the preset risk level of this type of risk factors, and the impact range coefficient i represents the weighted coefficient of the impact range of this type of risk factors to the overall risk.

[0219] This implementation method first determines the impact range of the dangerous factors and sets the safe construction distance to effectively ensure the basic safety boundaries of the construction site and prevent construction personnel and equipment from entering high-risk areas. Then, by quantitatively analyzing the number, type and risk level of dangerous factors in the construction area, a comprehensive construction safety risk coefficient is calculated. This risk coefficient provides construction managers with an intuitive safety risk indicator, enabling them to make more scientific decisions based on risk assessment results, such as adjusting construction plans, adding safety protection measures or replanning the layout of the construction area. Ultimately, safety management methods based on scientific analysis can significantly improve the safety of construction sites, reduce the occurrence of safety accidents, and ensure the smooth progress of construction.

[0220] The present application also provides a smart construction site safety supervision system. Figure 5 As shown, including:

[0221] Central control host 10;

[0222] The drone 20 is connected to the central control host.

[0223] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0224] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. Computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0225] Computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, the instruction device being implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1A function specified in one or more boxes.

[0226] Computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide the instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0227] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0228] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0229] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0230] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to the process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0231] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A smart construction site safety supervision method based on artificial intelligence, characterized in that: Applied to a smart construction site construction safety supervision system, the smart construction site construction safety supervision system includes a central control host and a drone, and the method includes: At every preset time period, n drone aerial images of the construction site and point cloud data of the construction site are acquired by the drone, wherein n>2, and n is an integer, and the construction site includes at least two construction areas; Generating a digital elevation model of the construction site according to the point cloud data; Extracting elevation data from the digital elevation model and generating contour lines based on the elevation data; Generate a digital orthophoto according to the n drone aerial images, and extract ground features and edge features in the drone aerial images; Integrating the digital orthophoto, the ground features and the digital elevation model to obtain a vector map of the construction site; Matching the edge features in the overlapping images of the n drone aerial images, and inputting the matched edge features into a pre-built machine learning model to generate an aerial triangulation model; Creating a stereo image pair according to the aerial triangulation model, and generating a real-scene three-dimensional model according to the stereo image pair; Integrating the real-scene three-dimensional model and the vector diagram of the construction site to obtain a digital map of the construction site, wherein the digital map of the construction site includes the contour lines; Acquiring location data of personnel in the construction area through the safety helmets of the personnel in the construction area; In the digital map of the construction site, risk factors are determined, and a safety hazard buffer zone analysis of the construction site is performed based on the risk factors and the personnel location data to obtain a safety hazard analysis warning zone of the construction site, wherein the safety hazard analysis warning zone of the construction site is a polygonal entity with a preset risk range; The safe construction distance and the construction safety risk factor of the construction area are determined and displayed based on the safety hazard analysis warning area of ​​the construction site.

2. The method according to claim 1, characterized in that Generating a digital elevation model of the construction site according to the point cloud data comprises: Constructing a two-dimensional grid in each of the construction areas, wherein the two-dimensional grid is composed of a plurality of grid units, and the area of ​​each of the grid units is consistent; Calculating an average height value of each grid cell according to the point cloud data in each grid cell; The point cloud data is converted into grid data by a preset interpolation algorithm, and each grid unit represents an average height value in the construction area; The digital elevation model is generated according to all the grid cells.

3. The method according to claim 2, characterized in that The integrating the digital orthophoto, the ground feature and the digital elevation model to obtain the vector map of the construction site includes: The coordinate system of the digital orthophoto, the coordinate system of the land feature and the coordinate system of the digital elevation model are converted into a unified coordinate system, and the common coordinates in the digital orthophoto and the land feature are used as control points; According to the control points, affine transformation is performed on the non-common coordinates in the digital orthophoto and the ground feature to align the non-common coordinates in the digital orthophoto and the ground feature; Extracting the geometric features of each feature from the aligned features; If the geometric feature has a corresponding grid cell in the digital elevation model, the average height value of the grid cell is used as the height attribute of the geometric feature; If the geometric feature does not have a corresponding grid unit in the digital elevation model, a height attribute of the geometric feature is calculated using a height estimation formula; generating three-dimensional vector data according to the geometric features and the height attributes of the geometric features; All the three-dimensional vector data are integrated to obtain a vector image of the construction site.

4. The method according to claim 3, characterized in that The step of performing affine transformation on the non-common coordinates in the digital orthophoto and the ground feature according to the control points so as to align the non-common coordinates in the digital orthophoto and the ground feature comprises: Constructing an affine transformation equation set according to at least three of the control points; Solving the affine transformation equations using the least squares method to determine affine transformation parameters; According to the affine transformation parameters, affine transformation is performed on the non-common coordinates in the digital orthophoto and the terrain features, so that the non-common coordinates in the digital orthophoto and the terrain features are aligned.

5. The method according to claim 1, characterized in that The step of creating a stereo image pair according to the aerial triangulation model, and generating a real-scene three-dimensional model according to the stereo image pair includes: According to the aerial triangulation model, randomly selecting two stereo images with overlapping areas, and calculating the disparity between the two stereo images; Matching the overlapping areas of the two stereo images by a preset image matching algorithm to determine corresponding points of the same name in the two stereo images; Performing stereoscopic correction on the two matched stereo images so that the points with the same name are located in the same row on the two stereo images to obtain a stereoscopic image pair; Performing parallax calculation on the same-name points in the stereo image pair to obtain the three-dimensional coordinates of each pixel point in the two stereo images; Three-dimensional point cloud data is generated according to the three-dimensional coordinates, and a real-scene three-dimensional model is generated according to the three-dimensional point cloud data.

6. The method according to claim 1, characterized in that The integration of the real-scene three-dimensional model and the vector diagram of the construction site to obtain a digital map of the construction site, wherein the digital map of the construction site includes the contour lines, includes: The vector map and the contour lines are projected onto the real-scene three-dimensional model to obtain a digital map of the construction site.

7. The method according to claim 1, characterized in that In the digital map of the construction site, the risk factors are determined, and the safety hazard buffer zone of the construction site is analyzed according to the risk factors and the personnel location data to obtain the safety hazard analysis warning zone of the construction site, including: Determining risk factors according to three-dimensional environmental features within a preset range of the construction area in the construction site digital map; According to the layer of the risk factor, a safety hazard buffer zone radius is set, and a safety hazard buffer zone layer is generated; Overlaying the potential safety hazard buffer layer with the personnel position to determine whether the personnel is located within the potential safety hazard buffer; If the personnel is located in the potential safety hazard buffer zone, the potential safety hazard buffer zone is determined to be a potential safety hazard analysis warning zone.

8. The method according to claim 7, characterized in that The risk factors include terrain risk, high-altitude falling object risk and mechanical operation risk, the three-dimensional environmental characteristics include slope characteristics, building characteristics and large equipment characteristics, and the risk factors are determined according to the three-dimensional environmental characteristics within the preset range of the construction area in the digital map of the construction site, including: Determining whether there is a risk of water accumulation and / or a risk of sudden landform change based on the slope characteristics; In the case where the risk of water accumulation and / or the risk of sudden terrain change exist, determining the risk factor to be the terrain risk; Calculate the falling object risk range around the building according to the building characteristics, and determine that the risk factor is the high-altitude falling object risk when the percentage of the overlapping area between the falling object risk range and the preset range of the construction area to the preset range of the construction area is greater than a preset percentage; Determine the type of the large equipment according to the large equipment characteristics, obtain the operating radius of the large equipment according to the large equipment type, and perform conflict detection between the operating radius and the building characteristics to determine whether there is a risk of collision between the large equipment and the building corresponding to the building characteristics during operation; If there is a risk of collision between the large equipment and the building corresponding to the building characteristics during operation, the risk factor is determined to be mechanical operation risk.

9. The method according to claim 8, characterized in that The determining of the safe construction distance and the construction safety risk factor of the construction area according to the safety hazard analysis warning area of ​​the construction site includes: Determine the impact range of the risk factor, and determine the safe construction distance based on the impact range and the safety hazard analysis warning area of ​​the construction site; The number, type and preset risk level of the hazardous factors in each of the construction areas are determined, and the construction safety risk coefficient of the construction area is calculated based on the number, type and preset risk level of the hazardous factors.

10. A smart construction site construction safety supervision system, applied to the smart construction site construction safety supervision method based on artificial intelligence according to any one of claims 1 to 9, characterized in that: include: Central control host; The drone is connected to the central control host.

Images