Steel bar bundling distance measuring method and device based on layered recognition and medium

Abstract

The invention discloses a steel bar bundling distance measuring method and device based on layered recognition and a medium, and relates to the technical field of building construction. The method comprises the following steps: carrying out multi-angle three-dimensional scanning on a construction area by carrying multiple sensors through an unmanned aerial vehicle, and generating a point cloud data set containing a steel bar layer; performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain spatial distribution of reinforcing steel bar layers; performing model classification on the reinforcing steel bar layer by using a three-dimensional convolutional neural network model, and extracting design parameters to perform multi-dimensional model comparison to obtain a reinforcing steel bar model and reinforcing steel bar bundling data; and a multi-source data set of the reinforcing steel bars is introduced and aligned, a rule database of the reinforcing steel bars is obtained, and a multi-modal acceptance report of the reinforcing steel bar models and the reinforcing steel bar bundling data is generated according to the rule database. According to the method, the problems of inaccurate acquisition of steel bar bundling data and single comparison dimension in a complex construction scene are solved, and accurate measurement of steel bar spacing in a high-density area and intelligent management of construction quality are realized.

Description

Technical Field

[0001] This application relates to the field of construction technology, and in particular, to a method, device, and medium for measuring the distance of steel bar bundling based on hierarchical recognition. Background Art

[0002] In the current construction field, the detection of the quality of steel bar bundling is a key link to ensure the safety and durability of the project structure. Traditional steel bar distance measurement methods mainly rely on manual measurement or single-view drone scanning technology. When dealing with high-density areas such as multi-layer steel bar bundling and beam-column joints, it is often difficult to accurately identify the position and type of each layer of steel bars due to steel bar overlap and occlusion in the three-dimensional scan data, resulting in large measurement errors. At the same time, only relying on the flat method annotation for comparison ignores key information such as node details, lap anchorage, etc. in the project drawings and steel bar atlases, and cannot meet the acceptance requirements of complex structures, restricting the improvement of detection efficiency and accuracy.

[0003] Through the above analysis, the problems and defects of the existing technology are as follows: In the detection of the quality of steel bar bundling in the existing technology, the data acquisition is inaccurate, the efficiency is low, and it is easy to introduce human errors. The comparison dimension is single, and it is difficult to provide intuitive and effective rectification suggestions for the construction site. Summary of the Invention

[0004] The embodiments of this application provide a method, device, and medium for measuring the distance of steel bar bundling based on hierarchical recognition, which can solve the problems in the detection of the quality of steel bar bundling in the existing technology, such as inaccurate acquisition of steel bar bundling data, low efficiency, easy introduction of human errors, single comparison dimension, and difficulty in providing intuitive and effective rectification suggestions for the construction site.

[0005] In a first aspect, the embodiments of this application provide a method for measuring the distance of steel bar bundling based on hierarchical recognition. The method includes: performing multi-angle three-dimensional scanning on the construction area through a drone equipped with multiple sensors to generate a point cloud data set containing steel bar layers; performing hierarchical recognition on the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers; using a three-dimensional convolutional neural network model to classify the types of steel bar layers, extracting design parameters for multi-dimensional type comparison to obtain the steel bar type and steel bar bundling data; introducing multi-source data sets of steel bars and aligning them to obtain a rule database of steel bars, and generating a multi-modal acceptance report of the steel bar type and steel bar bundling data according to the rule database. The multi-modal acceptance report includes problem markings and deviation analysis of the steel bars.

[0006] In an implementation manner of the present application, hierarchical recognition is performed on the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers, specifically including: clustering the point cloud data set through the density clustering algorithm to obtain the contour clusters of single steel bars; calculating the average height value of the contour clusters to obtain the vertical distribution range of the steel bar layers and generating a three-dimensional spatial distribution map of the steel bars; combining the three-dimensional spatial distribution map with the multi-spectral texture information and generating a three-dimensional model of the steel bar bundling area through the three-dimensional reconstruction algorithm.

[0007] In an implementation manner of the present application, a three-dimensional convolutional neural network model is used to classify the types of steel bar layers, extract design parameters for multi-dimensional type comparison, and obtain the steel bar types and steel bar bundling data, specifically including: cropping the steel bar clusters into three-dimensional data blocks of a preset size, and performing rotation, translation, and noise enhancement on the three-dimensional data blocks; extracting the spatial features of the three-dimensional data blocks through a three-dimensional residual network and outputting the probability distribution of the steel bar types according to a preset steel bar specification mapping table; screening the classification results with probabilities exceeding a preset threshold to obtain the steel bar types and the corresponding confidence levels.

[0008] In an implementation manner of the present application, multi-source data sets of steel bars are introduced and aligned to obtain a rule database of steel bars, specifically including: obtaining the plane method annotation, project drawings, and steel bar atlas of the steel bars; parsing the steel bar spacing, lap length, and joint construction rules in the plane method annotation, and extracting the steel bar diameter, spacing, and quantity parameters, and converting them into calculation expressions; parsing the anchorage length and lap coefficient in the project drawings and the construction rules of the steel bar atlas, and defining multi-dimensional acceptance rules and logical judgment conditions; obtaining the rule database of the steel bars according to the calculation expressions, multi-dimensional acceptance rules, and logical judgment conditions.

[0009] In an implementation manner of the present application, a multi-modal acceptance report of the steel bar types and steel bar bundling data is generated according to the rule database, specifically including: performing spatial matching between the three-dimensional model and the rule database of the corresponding steel bar types, and calculating the spacing deviation and construction matching degree; marking the problem points in the three-dimensional model, distinguishing the deviation levels through color coding, and generating a deviation heat map and a construction comparison map.

[0010] In an implementation manner of the present application, after generating a point cloud data set including steel bar layers through multi-angle three-dimensional scanning of the construction area by a drone carrying multiple sensors, the method further includes: calculating the neighborhood distance distribution of each point in the point cloud data set through a statistical filter and removing the noise points exceeding a preset distance threshold; performing smoothing processing on the remaining point cloud data by using a Gaussian filter and dynamically adjusting based on a preset radius and weight function; converting the remaining point cloud data from the acquisition coordinate system to an engineering coordinate system with the steel bar distribution direction as the reference to complete spatial alignment.

[0011] In an implementation manner of the present application, after spatially matching the 3D model with the rule database of the corresponding steel bar model and calculating the spacing deviation and the construction matching degree, the method further includes: calculating the normal vector and curvature of the point cloud data set, and correcting the spatial deviation of the steel bar position based on the normal vector and curvature; using the iterative closest point algorithm to align the offsets between the acquisition coordinate system and the engineering coordinate system; after the acquisition coordinate system and the engineering coordinate system are aligned, evaluating the confidence interval of the matching result through a probability model.

[0012] In an implementation manner of the present application, after hierarchically identifying the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers, the method further includes: dynamically adjusting the clustering radius and the minimum point number threshold according to the distribution density of the steel bars; combining the curvature to dynamically adjust the hierarchical accuracy of the vertical distribution range.

[0013] In a second aspect, an embodiment of the present application further provides a steel bar bundling ranging device based on hierarchical identification. The device includes at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: perform multi-angle 3D scanning on the construction area through a drone carrying multiple sensors to generate a point cloud data set including steel bar layers; hierarchically identify the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers; use a 3D convolutional neural network model to classify the types of the steel bar layers, extract design parameters for multi-dimensional type comparison to obtain the steel bar model and the steel bar bundling data; introduce and align the multi-source data sets of the steel bars to obtain the rule database of the steel bars, and generate a multi-modal acceptance report of the steel bar model and the steel bar bundling data according to the rule database, and the multi-modal acceptance report includes problem markings and deviation analysis of the steel bars.

[0014] In a third aspect, an embodiment of the present application further provides a non-volatile computer storage medium for steel bar bundling ranging based on hierarchical identification, storing computer-executable instructions, and the computer-executable instructions are set to: perform multi-angle 3D scanning on the construction area through a drone carrying multiple sensors to generate a point cloud data set including steel bar layers; hierarchically identify the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers; use a 3D convolutional neural network model to classify the types of the steel bar layers, extract design parameters for multi-dimensional type comparison to obtain the steel bar model and the steel bar bundling data; introduce and align the multi-source data sets of the steel bars to obtain the rule database of the steel bars, and generate a multi-modal acceptance report of the steel bar model and the steel bar bundling data according to the rule database, and the multi-modal acceptance report includes problem markings and deviation analysis of the steel bars.

[0015] A method, device, and medium for measuring the distance of steel bar bundling based on hierarchical recognition provided by the embodiments of the present application. Through a drone carrying a high-frame-rate lidar and a multispectral camera, multi-angle three-dimensional scanning is realized. Combining deep learning algorithms, hierarchical recognition and model classification of steel bars are carried out, and multi-source design data such as flat method annotation, project drawings, and steel bar atlases are integrated to construct a comprehensive multi-dimensional acceptance rule library. It can accurately measure the spacing of steel bars in high-density areas, automatically complete multi-dimensional acceptance such as lap length and node structure matching degree, improve the efficiency and accuracy of construction quality inspection; generate a multi-modal report containing problem point markings, deviation values, rule violation entries, etc., and provide intuitive and effective rectification suggestions for the construction site; applicable to the intelligent management of the quality of steel bar projects in complex engineering projects such as super high-rise buildings and bridges. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation to the present application. In the drawings: Figure 1 It is a flowchart of a method for measuring the distance of steel bar bundling based on hierarchical recognition provided by the embodiments of the present application; Figure 2 It is an overall logical architecture diagram of a method for measuring the distance of steel bar bundling based on hierarchical recognition provided by the embodiments of the present application; Figure 3 It is an internal structure schematic diagram of a device for measuring the distance of steel bar bundling based on hierarchical recognition provided by the embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be clearly and completely described below in conjunction with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0018] The embodiments of the present application provide a method, device, and medium for measuring the distance of steel bar bundling based on hierarchical recognition, which solves the problems in the prior art that in the detection of the quality of steel bar bundling, the acquisition of steel bar bundling data is inaccurate, the efficiency is low and it is easy to introduce human errors, the comparison dimension is single, and it is difficult to provide intuitive and effective rectification suggestions for the construction site.

[0019] The technical solutions proposed by the embodiments of the present application will be described in detail below with reference to the drawings.

[0020] Figure 1This is a flowchart of a steel bar bundling ranging method based on hierarchical recognition provided by an embodiment of the present application. As Figure 1 shown, a steel bar bundling ranging method based on hierarchical recognition provided by an embodiment of the present application specifically includes the following steps: Step 10: Use a drone equipped with multiple sensors to perform multi-angle three-dimensional scanning on the construction area to generate a point cloud data set containing steel bar layers.

[0021] In this step, in order to overcome the problems of occlusion and overlap existing in traditional single-view scanning, a drone is used to carry a high-frame-rate lidar and a multi-spectral camera for multi-angle three-dimensional scanning. The drone performs omnidirectional scanning on key areas such as beam-column joints according to the preset circumferential path planning to ensure that each angle can be captured. The accuracy of the lidar can reach ±1 mm, ensuring the high quality of the point cloud data. In addition, the multi-spectral camera can provide additional texture information to help further improve the accuracy of steel bar recognition. In practical applications, the drone can collect data of tens of thousands of points and generate a point cloud data set containing a large number of steel bars.

[0022] Step 20: Perform hierarchical recognition on the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers.

[0023] In this step, in order to effectively solve the problems of overlap and occlusion between multi-layer steel bars, a point cloud layering algorithm accelerated by GPU is adopted, which is mainly divided into two steps: First, use the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm to perform preliminary clustering on the point cloud data, and then use the trained ResNet-3D model to classify the types of each layer of steel bars.

[0024] As an optional embodiment, after using a drone equipped with multiple sensors to perform multi-angle three-dimensional scanning on the construction area to generate a point cloud data set containing steel bar layers, the method may further include: calculating the neighborhood distance distribution of each point in the point cloud data set through a statistical filter, removing noise points exceeding the preset distance threshold; performing smoothing processing on the remaining point cloud data using a Gaussian filter, and dynamically adjusting based on the preset radius and weight function; converting the remaining point cloud data from the acquisition coordinate system to the engineering coordinate system with the steel bar distribution direction as the reference to complete spatial alignment, refer to Figure 2 .

[0025] In this step, noise points usually refer to those isolated points or outliers that have nothing to do with the real structure. They may interfere with subsequent clustering analysis. These noise points can be removed by a Statistical Outlier Removal Filter. For each point, calculate the average distance to its nearest neighbor points, calculate the standard deviation of the average distances of all points. If the average distance of a certain point exceeds the set threshold (usually the mean plus several times the standard deviation), it is considered a noise point and removed.

[0026]

[0027] Among them, di is the average distance from the i-th point to its nearest neighbor, dist(pi, pj) represents the Euclidean distance between two points, and k is the number of neighbors; the threshold can be set as: threshold = mean(d) + n × std(d), where mean(d) and std(d) represent the mean and standard deviation of the average distances of all points respectively, and n is an empirical value, usually taking 2 or 3.

[0028] To further improve the data quality, a Gaussian Filter can be used to smooth the point cloud. The Gaussian Filter adjusts the position of each point based on the distances of its surrounding points by assigning a weight to each point, thereby achieving a smoothing effect. Define a sphere with a radius of r as the local area. Within this area, apply the Gaussian weight function according to the distance of each point from the center point: Among them, d is the distance from the point to the center point, and σ controls the speed of weight decay. Update the new position of each point to the weighted average position: It is necessary to convert the point cloud data from the coordinate system used during acquisition to a coordinate system more suitable for analysis. In steel bar detection, rotate and translate the coordinate system to better adapt to the direction and distribution of the steel bars. Use the rigid body transformation matrix T to describe the change from the source coordinate system to the target coordinate system: Among them, R is the rotation matrix and t is the translation vector. For each point P, apply the transformation to obtain the new coordinate P': P' = TP.

[0029] As an alternative embodiment, hierarchically identify the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers. Specifically, it can include: Step 201: Cluster the point cloud data set through the density clustering algorithm to obtain the contour clusters of single steel bars; Step 202: Calculate the average height value of the contour clusters to obtain the vertical distribution range of the steel bar layers and generate a three-dimensional spatial distribution map of the steel bars; Step 203: Combine the three-dimensional spatial distribution map with the multi-spectral texture information and generate a three-dimensional model of the steel bar bundling area through the three-dimensional reconstruction algorithm.

[0030] In this step, for the steel bar point cloud data, different steel bar distribution characteristics in different scenarios can be adapted by adjusting ε (representing the neighborhood radius of a point. If the number of points included within this range exceeds MinPts, it is considered a core point) and MinPts (specifying the minimum number of points that a core point should contain at least within its ε neighborhood). Since steel bars are usually arranged vertically, a smaller ε value is set on the horizontal plane to capture the contour of a single steel bar, while a larger ε value is set in the vertical direction to distinguish different steel bar layers. The specific steps are as follows: Determine appropriate ε and MinPts: Adjust the parameters according to actual engineering experience and tests to ensure that the steel bars can be accurately divided while avoiding over-segmentation; Traverse all points and mark them as core points, boundary points, or noise points. For each core point, all the points that meet the conditions around it are grouped into the same cluster. Repeat this process until all points are assigned to the corresponding clusters or marked as noise. After completing the preliminary clustering, according to the average height (i.e., the z-axis coordinate) of each cluster of point sets, they are divided into different layers. For example, assuming there are three layers of steel bars at the beam-column joint of a construction site, the upper and lower layers can be distinguished by calculating the average height of each cluster of point sets.

[0031] Considering that there may be irregular distributions in actual construction, it is also necessary to combine manual inspection or other auxiliary means to optimize and verify the results of automatic division to ensure the accuracy of the final results.

[0032] Step 30: Use a three-dimensional convolutional neural network model to classify the types of steel bar layers, extract design parameters for multi-dimensional type comparison, and obtain steel bar types and steel bar bundling data.

[0033] As an optional embodiment, using a three-dimensional convolutional neural network model to classify the types of steel bar layers, extract design parameters for multi-dimensional type comparison, and obtain steel bar types and steel bar bundling data, specifically may include: Step 301: Crop the steel bar clusters into three-dimensional data blocks of a preset size, and perform rotation, translation, and noise enhancement on the three-dimensional data blocks; Step 302: Extract the spatial features of the three-dimensional data blocks through a three-dimensional residual network, and output the probability distribution of the steel bar types according to a preset steel bar specification mapping table; Step 303: Screen the classification results with probabilities exceeding a preset threshold to obtain the steel bar types and the corresponding confidence levels.

[0034] In this step, once the position information of each steel bar layer is determined by the DBSCAN algorithm, the next task is to use the deep learning model ResNet-3D to accurately classify the types of steel bars within each layer. This process involves extracting features from the preprocessed point cloud data and using the trained model to identify the specific specifications of the steel bars (such as Φ12, Φ16, etc.). The following are the specific steps: Before inputting into the ResNet-3D model, the point cloud data needs to be further prepared and enhanced to ensure that the model can accurately identify the steel bar types; for each segmented steel bar cluster, a sub-region containing the steel bar is cropped according to its bounding box and its size is normalized to a fixed size (for example, a cube of 128×128×128) for unified processing by the model; to improve the generalization ability of the model, data augmentation operations can be performed on the dataset, including rotation, translation, scaling, and adding random noise, which helps the model better adapt to changes in the actual scenario. The core of the ResNet-3D model lies in its powerful feature extraction ability, which automatically learns complex patterns in the input point cloud data through multiple layers of convolutional neural networks. Input layer: The normalized point cloud data is used as the input, with a shape of N×D, where N is the number of points and D is the feature dimension of each point (usually including XYZ coordinates and possible additional attributes such as color or normal). Convolutional layer: 3D convolutional kernels are used to perform convolutional operations on the input data to extract spatial features. The size of the convolutional kernel can be adjusted according to specific circumstances, and common settings are 3×3×3 or 5×5×5. Activation function: The ReLU activation function is applied after each layer of convolution to increase the non-linear expression ability. Pooling layer: The spatial size of the feature map is reduced through max pooling operations while retaining the most important feature information. ResNet-3D is designed based on the Residual Network structure, which can effectively alleviate the vanishing gradient problem in deep networks, thus allowing the construction of deeper network structures. The basic module consists of multiple residual blocks, each of which contains two or more convolutional layers, plus a skip connection, enabling the output to be directly passed to the next layer. The global average pooling layer is used to compress the feature map into a fixed-length vector, and then a fully connected layer is connected for the final classification task.

[0035] For example, learning rate: the initial value is set to 0.001 and adjusted dynamically according to the training progress; batch size: select an appropriate batch size according to the hardware conditions, such as 32 or 64; loss function: use the cross-entropy loss function to measure the difference between the prediction result and the true label; optimizer: the Adam optimizer is widely used because of its adaptive learning rate adjustment mechanism; the last layer of the model is usually a fully connected layer, which is responsible for mapping the extracted features to specific steel bar types, and the number of nodes in the output layer is equal to the number of steel bar types to be classified. Softmax function: Apply the Softmax function after the fully connected layer to convert the scores of each type into a probability distribution, indicating the possibility of belonging to a certain category; threshold judgment: set a confidence threshold (such as 0.9), and only when the prediction probability of a certain category exceeds this threshold, it is considered that the sample belongs to this category; otherwise, mark it as uncertain or require manual review; result output: for each input steel bar cluster, the model outputs its most likely steel bar model (such as Φ12, Φ16, etc.) and the corresponding confidence score.

[0036] Step 40: Introduce and align the multi-source dataset of steel bars to obtain the rule database of steel bars, and generate a multi-modal acceptance report of steel bar models and steel bar bundling data according to the rule database. The multi-modal acceptance report includes problem markings and deviation analysis of steel bars.

[0037] As an optional embodiment, introducing and aligning the multi-source dataset of steel bars to obtain the rule database of steel bars may specifically include: Step 401: Obtain the plane method annotation, project drawings, and steel bar atlas of steel bars; Step 402: Analyze the steel bar spacing, lap length, and joint construction rules in the plane method annotation, extract the steel bar diameter, spacing, and quantity parameters, and convert them into calculation expressions; Step 403: Analyze the anchorage length and lap coefficient in the project drawings, as well as the construction rules of the steel bar atlas, and define multi-dimensional acceptance rules and logical judgment conditions; Step 404: Obtain the rule database of steel bars according to the calculation expressions, multi-dimensional acceptance rules, and logical judgment conditions.

[0038] In this step, the multi-source data parsing engine integrates design parameters from multiple sources such as plane method annotations, project drawings (CAD / DWG format), and steel bar atlases (PDF format), and constructs a multi-dimensional acceptance rule library.

[0039] Parse the flat method annotation: First, use OCR (Optical Character Recognition) technology to identify the relevant content of the flat method annotation from the design documents, which are usually contained in specific tables or annotation areas. According to the predefined templates or pattern matching algorithms, extract the specific information of the steel bars, such as diameter (Φ), spacing (@), quantity, etc. For example, "Φ12@150" represents steel bars with a diameter of 12mm and a spacing of 150mm. Convert the extracted data into a form that can be processed by a computer and perform basic logical verification to ensure the correctness of the data. For example, check whether the spacing meets the specification requirements. Among them, L is the length of the component, N is the quantity of steel bars. For a given spacing S, the quantity of steel bars N can be deduced.

[0040] Extract project drawings: Import the project drawings in CAD / DWG format into the system and use vectorization tools to convert them into vector graphics that are easy to process. Adopt image processing techniques such as edge detection and contour analysis to automatically identify the key features in the drawings, such as beam-column joints and steel bar arrangements. Extract specific numerical information, such as anchorage length and lap length, through pattern recognition algorithms. For example, the "anchorage length = 900mm" marked in the detail drawing of the joint can be directly read and recorded. The anchorage length La: It is determined according to the diameter of the steel bar and the concrete strength grade. The lap length Ld: It is calculated based on the type of steel bar and the connection method. The formula is: Ld = α ⋅ d. Among them, α is the lap coefficient and d is the diameter of the steel bar.

[0041] Read the construction rules of the steel bar atlas: Use a PDF parsing tool to open the steel bar atlas file and locate to the relevant chapters or paragraphs. Based on natural language processing (NLP) technology, extract the specific construction rules from the text, such as "when the diameter of the steel bar is greater than 20mm, the hook angle should be 135 degrees". Map the extracted rules to the internal data structure of the system for subsequent comparison calls. The hook angle θ may need to be determined according to the diameter d of the steel bar and the application scenario. For example, in areas with higher seismic requirements, the hook angle θ may be set to 135 degrees or larger.

[0042] Convert the information from different data sources to the same coordinate system or reference frame for subsequent processing and analysis. Unify the point cloud data, design parameters, and specification requirements within a common spatial reference system. First, a unified coordinate system needs to be determined, which can be achieved by selecting a certain reference point (such as a key node of the building structure) as the origin. Then, use the rigid body transformation matrix T to convert all data sources to this coordinate system. Among them, R is the rotation matrix and t is the translation vector.

[0043] After completing the coordinate transformation, perform a consistency check to ensure that there are no conflicts between the various data sources. For example, check whether the positions of the steel bars in the point cloud data match the markings in the project drawings.

[0044] Build a rule library: Extract all design parameters from the plane method annotation, project drawings, and steel bar atlas, and organize them into a form that is easy to query. Store information such as steel bar specifications, spacing, and anchorage length in a database table for quick retrieval. Based on the above design parameters, define a complete set of acceptance rules. These rules can be divided into multiple dimensions, such as steel bar spacing, lap length, and node structure matching degree. Each rule should clearly define its scope of application, calculation method, and allowable deviation range. For example: Steel bar spacing: Design value vs measured value, allowable deviation ±5mm. Lap length: Calculated according to the steel bar diameter d, generally required to be not less than 30d. Node structure matching degree: The shape matching degree should reach more than 95%.

[0045] In addition to specific engineering design parameters, relevant national standards (such as 《GB 50010》) and industry codes (such as 《16G101-1》) also need to be integrated. This step usually involves a large amount of text parsing work. Use natural language processing technology to extract key clauses from the documents and convert them into logical judgment conditions that can be executed by a computer.

[0046] A rule engine can also be developed to automatically apply the above rules for comparison based on the input 3D model data. The core of the rule engine is a decision tree, which selects corresponding processing paths according to different input conditions. When checking the stirrup spacing, if it is found that the measured value deviates from the design value by more than the set threshold, a corresponding warning mechanism is triggered.

[0047] As an optional embodiment, generate a multimodal acceptance report of steel bar models and steel bar bundling data according to the rule database. Specifically, it can include: Step 405: Perform spatial matching between the 3D model and the rule database of the corresponding steel bar model, and calculate the spacing deviation and structure matching degree; Step 406: Mark the problem points in the 3D model, distinguish the deviation levels through color coding, and generate a deviation heat map and a structure comparison map.

[0048] In this step, for any non-compliant situations, mark them in the 3D model for subsequent viewing and processing. Different colors can be used to represent different types of problems, with red for severe deviations (such as excessive steel bar spacing) and yellow for minor deviations (such as insufficient anchorage length); add text labels on the 3D model to describe the specific problems and the corresponding code provisions. For example, "Problem P-102: The stirrup spacing at the beam-column joint exceeds the standard by 12mm, violating Article 5.3.2 of '16G101-1'". Based on the comparison results, generate a detailed inspection report. These reports not only contain accurate location information and numerical deviations but also come with detailed graphical explanations and rectification suggestions. The following are the main components of the report: Diagrammatic report: Display the 3D model annotation diagram (showing the actual positions of the steel bars and the marked problem points), deviation heat map (using the shade of color to represent the degree of deviation for quick identification of severe problems), and node structure comparison diagram (showing the differences between the actual node and the designed node to help analyze the reasons). Use visualization means to help engineers quickly understand the problems; the data table can list the coordinates of all problem points, deviation values, and their corresponding violated provisions. For example, the actual steel bar spacing at a certain point is 158mm, the designed value is 150mm, and the deviation value is +8mm.

[0049] For example, intelligent comparison and marking and multi-modal report generation, as shown in Table 1: Table 1 Multi-modal Report

[0050] Propose targeted improvement measures according to the specific situation and cite relevant national standards or industry codes as the basis. For example, it is recommended to adjust the stirrup spacing to the designed value and refer to Article 5.3.2 of '16G101-1' for construction.

[0051] As an alternative embodiment, after spatially matching the three-dimensional model with the rule database of the corresponding steel bar models and calculating the spacing deviation and structure matching degree, the method may further include: calculating the normal and curvature of the point cloud data set, and correcting the spatial deviation of the steel bar positions based on the normal and curvature; using the iterative closest point algorithm to align the offsets between the acquisition coordinate system and the engineering coordinate system; after aligning the acquisition coordinate system and the engineering coordinate system, evaluating the confidence interval of the matching result through a probability model.

[0052] In this step, to facilitate subsequent classification tasks, features can be extracted from the preprocessed point cloud data: for each point, find the set of points in its neighborhood, and then use principal component analysis (PCA) to calculate the normal direction of the point; use the spatial distribution characteristics of the points in the neighborhood to estimate the curvature, which reflects the degree of surface change. The normal direction can be obtained from the first principal component of PCA; the curvature can be calculated according to the ratio of the minimum eigenvalue to the total eigenvalue: , where λ0, λ1, and λ2 are the three eigenvalues obtained by PCA and are sorted in descending order.

[0053] As an alternative embodiment, after hierarchically identifying the point cloud dataset based on the density clustering algorithm to obtain the spatial distribution of the steel bar layer, the method may further include: dynamically adjusting the clustering radius and the minimum point number threshold according to the distribution density of the steel bars; and dynamically adjusting the hierarchical accuracy of the vertical distribution range in combination with the curvature.

[0054] In this step, by dynamically adjusting the hierarchical accuracy of the vertical distribution range in combination with the previously calculated curvature information, an appropriate hierarchical accuracy can be obtained in different curvature regions, thereby further improving the accuracy of distance measurement.

[0055] The above is the method embodiment proposed in this application. Based on the same inventive concept, the embodiments of this application also provide a steel bar bundling distance measurement device based on hierarchical identification, and its structure is as Figure 3 shown.

[0056] Figure 3 This is a schematic diagram of the internal structure of a steel bar bundling distance measurement device based on hierarchical identification provided by the embodiments of this application. As Figure 3 shown, the device includes: at least one processor 301; and a memory 302 communicatively connected to the at least one processor; wherein, the memory 302 stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor 301 to enable the at least one processor 301 to: perform multi-angle three-dimensional scanning on the construction area through a drone equipped with multiple sensors to generate a point cloud dataset containing a steel bar layer; hierarchically identify the point cloud dataset based on the density clustering algorithm to obtain the spatial distribution of the steel bar layer; use a three-dimensional convolutional neural network model to classify the steel bar layer, extract design parameters for multi-dimensional model comparison to obtain the steel bar model and steel bar bundling data; introduce and align the multi-source datasets of the steel bars to obtain a rule database of the steel bars, and generate a multi-modal acceptance report of the steel bar model and steel bar bundling data according to the rule database, and the multi-modal acceptance report includes problem markings and deviation analysis of the steel bars.

[0057] Some embodiments of this application provide corresponding to Figure 1A non-volatile computer storage medium for measuring the distance of steel bar bundling based on hierarchical recognition stores computer-executable instructions, which are set as follows: use a drone equipped with multiple sensors to perform multi-angle three-dimensional scanning on the construction area to generate a point cloud data set containing steel bar layers; perform hierarchical recognition on the point cloud data set based on the density clustering algorithm to obtain the spatial distribution of the steel bar layers; use a three-dimensional convolutional neural network model to classify the types of steel bar layers, extract design parameters for multi-dimensional type comparison to obtain the steel bar type and steel bar bundling data; introduce and align the multi-source data sets of steel bars to obtain a rule database of steel bars, and generate a multi-modal acceptance report for the steel bar type and steel bar bundling data according to the rule database. The multi-modal acceptance report includes problem markings and deviation analysis of the steel bars.

[0058] Each embodiment in this application is described in a progressive manner. For the same or similar parts between the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the Internet of Things devices and media, since they are basically similar to the method embodiments, the description is relatively simple, and for the relevant parts, reference can be made to the partial description of the method embodiments.

[0059] The systems and media provided in the embodiments of this application correspond one-to-one with the methods. Therefore, the systems and media also have beneficial technical effects similar to those of their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be elaborated here.

[0060] Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can 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 code.

[0061] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of this application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 a process or multiple processes and / or blocks Figure 1a device for the functions specified in one or more boxes.

[0062] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one Figure 1 one process or more processes and / or boxes Figure 1 a box or more boxes.

[0063] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 one process or more processes and / or boxes Figure 1 a box or more boxes.

[0064] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0065] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0066] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The 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 technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassette tapes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0067] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the phrase "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising said element.

[0068] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A method for measuring the distance between steel bar bundles based on layered recognition, characterized in that: The method comprises: Use drones equipped with multiple sensors to perform multi-angle 3D scanning of the construction area and generate a point cloud dataset including the steel bar layer; Performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain the spatial distribution of the steel bar layer; Using a three-dimensional convolutional neural network model to classify the steel bar layer, extracting design parameters for multi-dimensional model comparison, and obtaining steel bar models and steel bar bundling data; Multi-source data sets of steel bars are introduced and aligned to obtain a rule database of steel bars, and a multi-modal acceptance report of steel bar models and steel bar bundling data is generated according to the rule database. The multi-modal acceptance report includes problem marking and deviation analysis of the steel bars.

2. A method for measuring the distance between steel bars bundled based on layered identification according to claim 1, characterized in that: The step of performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain the spatial distribution of the steel bar layer specifically includes: Clustering the point cloud data set by a density clustering algorithm to obtain a contour cluster of a single steel bar; Calculating the average height value of the contour cluster to obtain the vertical distribution range of the steel bar layer and generate a three-dimensional spatial distribution map of the steel bars; The three-dimensional spatial distribution map is combined with the multi-spectral texture information, and a three-dimensional model of the steel bar bundling area is generated through a three-dimensional reconstruction algorithm.

3. A method for measuring the distance between steel bar bundles based on layered identification according to claim 2, characterized in that: The method of using a three-dimensional convolutional neural network model to classify the steel bar layer, extracting design parameters for multi-dimensional model comparison, and obtaining steel bar models and steel bar bundling data specifically includes: Cutting the steel bar cluster into a three-dimensional data block of a preset size, and rotating, translating and noise enhancing the three-dimensional data block; Extracting the spatial features of the three-dimensional data block through a three-dimensional residual network, and outputting the probability distribution of the steel bar model according to a preset steel bar specification mapping table; The classification results whose probability exceeds the preset threshold are screened to obtain the steel bar model and the corresponding confidence level.

4. A method for measuring distance between steel bar bundles based on layered identification according to claim 3, characterized in that: The multi-source data sets of steel bars are introduced and aligned to obtain a rule database of steel bars, which specifically includes: Obtain rebar plan markings, project drawings and rebar atlases; Analyze the steel bar spacing, lap length and node construction rules in the flat method annotation, extract the steel bar diameter, spacing and quantity parameters, and convert them into calculation expressions; Analyze the anchorage length and lap coefficient in the project drawings, as well as the construction rules of the steel bar atlas, and define multi-dimensional acceptance rules and logical judgment conditions; According to the calculation expression, multi-dimensional acceptance rules and logical judgment conditions, a rule database of steel bars is obtained.

5. The method for measuring the distance between steel bar bundles based on layered identification according to claim 2, characterized in that: The generating of the multimodal acceptance report of the steel bar model and steel bar bundling data according to the rule database specifically includes: Spatially matching the three-dimensional model with a rule database of corresponding steel bar models, and calculating spacing deviation and structural matching degree; Problem points are marked in the three-dimensional model, deviation levels are distinguished by color coding, and deviation heat maps and structural comparison maps are generated.

6. A method for measuring the distance between steel bar bundles based on layered identification according to claim 2, characterized in that: After the construction area is scanned in three dimensions from multiple angles by using a drone equipped with multiple sensors to generate a point cloud data set including a steel bar layer, the method further includes: Calculating the neighborhood distance distribution of each point in the point cloud data set by using a statistical filter, and removing noise points that exceed a preset distance threshold; A Gaussian filter is used to smooth the remaining point cloud data, and dynamic adjustments are made based on the preset radius and weight function; The remaining point cloud data are converted from the acquisition coordinate system to the engineering coordinate system based on the distribution direction of the steel bars to complete the spatial alignment.

7. A method for measuring distance of steel bar bundling based on layered identification according to claim 6, characterized in that: After spatially matching the three-dimensional model with the rule database of the corresponding steel bar model and calculating the spacing deviation and the structural matching degree, the method further includes: Calculating the normal and curvature of the point cloud data set, and correcting the spatial deviation of the steel bar position based on the normal and curvature; Aligning the offsets of the acquisition coordinate system and the engineering coordinate system using an iterative closest point algorithm; After the acquisition coordinate system and the engineering coordinate system are aligned, the confidence interval of the matching result is evaluated by a probability model.

8. A method for measuring distance of steel bar bundling based on layered identification according to claim 7, characterized in that: After performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain the spatial distribution of the steel bar layer, the method further includes: Dynamically adjust the clustering radius and minimum point count threshold according to the distribution density of steel bars; In combination with the curvature, the stratification accuracy of the vertical distribution range is dynamically adjusted.

9. A steel bar bundling distance measuring device based on layer recognition, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Use drones equipped with multiple sensors to perform multi-angle 3D scanning of the construction area and generate a point cloud dataset including the steel bar layer; Performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain the spatial distribution of the steel bar layer; Using a three-dimensional convolutional neural network model to classify the steel bar layer, extracting design parameters for multi-dimensional model comparison, and obtaining steel bar models and steel bar bundling data; Multi-source data sets of steel bars are introduced and aligned to obtain a rule database of steel bars, and a multi-modal acceptance report of steel bar models and steel bar bundling data is generated according to the rule database. The multi-modal acceptance report includes problem marking and deviation analysis of the steel bars.

10. A non-volatile computer storage medium for measuring the distance of steel bar bundling based on layered identification, storing computer executable instructions, characterized in that: The computer executable instructions are configured to: Use drones equipped with multiple sensors to perform multi-angle 3D scanning of the construction area and generate a point cloud dataset including the steel bar layer; Performing hierarchical identification on the point cloud data set based on a density clustering algorithm to obtain the spatial distribution of the steel bar layer; Using a three-dimensional convolutional neural network model to classify the steel bar layer, extracting design parameters for multi-dimensional model comparison, and obtaining steel bar models and steel bar bundling data; Multi-source data sets of steel bars are introduced and aligned to obtain a rule database of steel bars, and a multi-modal acceptance report of steel bar models and steel bar bundling data is generated according to the rule database. The multi-modal acceptance report includes problem marking and deviation analysis of the steel bars.

Images