A method for monitoring the safety of a structure in a multi-stage construction process of a building

Abstract

The present disclosure relates to a kind of building multi-stage construction process structure safety intelligent monitoring method, more particularly to the technical field of building construction safety monitoring.It includes: obtaining the building point cloud data of different construction stages;Building point cloud data is input into intelligent algorithm network model, and the plane registration point cloud data output by intelligent algorithm network model is obtained;Intelligent algorithm network model is used to segment and register building point cloud data;Structure safety feature points are selected from plane registration point cloud data;Based on structure safety feature points, calculate structure safety parameters, and structure safety parameters are used to represent the settlement deformation change of building in construction process.The present disclosure embodiment can intelligently process building point cloud data, and improve data processing efficiency.

Description

Technical Field

[0001] The present disclosure relates to the technical field of building construction safety monitoring, and in particular to a method for intelligently monitoring the structural safety of a building during a multi-stage construction process, as well as an intelligent monitoring device for the structural safety of a building during a multi-stage construction process, an electronic device, a storage medium, and a computer program product. Background Art

[0002] With the continuous development of the economy and the rapid urbanization in my country, more and more buildings are being built taller and deeper. During the construction process, particular attention must be paid to the overall safety of the building structure. Improper construction methods and procedures can adversely affect the overall safety of the structure, and in severe cases, cause structural collapse and damage. Therefore, it is of great significance to conduct structural safety monitoring during the construction process to promptly identify safety issues in buildings during construction.

[0003] In recent years, intelligent construction has been widely adopted across various fields. Visual processing technologies, particularly 3D point clouds, have begun to be widely applied in building construction. By acquiring comprehensive building information through 3D point clouds, comprehensive and accurate analysis of building safety conditions at different times can be achieved. However, current research on intelligent point cloud processing is still limited. The large amount of acquired building point cloud data still requires manual post-processing for building safety monitoring, resulting in low data processing efficiency. Summary of the Invention

[0004] In order to solve the above technical problems or at least partially solve the above technical problems, the present disclosure provides a method for intelligent monitoring of structural safety in a multi-stage construction process of a building, which can intelligently process building point cloud data and improve data processing efficiency.

[0005] In order to achieve the above objectives, the technical solutions provided by the embodiments of the present disclosure are as follows:

[0006] In a first aspect, the present disclosure provides a method for intelligently monitoring structural safety during a multi-stage construction process of a building, comprising:

[0007] Obtain building point cloud data at different construction stages;

[0008] Input the building point cloud data into the intelligent algorithm network model to obtain the plane registration point cloud data output by the intelligent algorithm network model; the intelligent algorithm network model is used to segment and register the building point cloud data;

[0009] Select structural safety feature points from the plane registration point cloud data;

[0010] Based on the structural safety characteristic points, the structural safety parameters are calculated. The structural safety parameters are used to characterize the changes in settlement and deformation of the building during the construction process.

[0011] As an optional implementation of the embodiment of the present disclosure, the intelligent algorithm network model comprises: a point cloud segmentation sub-model and a point cloud registration sub-model; the point cloud segmentation sub-model is configured to segment the building point cloud data into a plurality of planar point cloud data; and the point cloud registration sub-model is configured to register the planar point cloud data of different construction stages.

[0012] As an optional implementation of the embodiment of the present disclosure, after obtaining the building point cloud data of different construction stages, before inputting the building point cloud data into the intelligent algorithm network model to obtain the planar registration point cloud data output by the intelligent algorithm network model, the method further comprises: pre-processing the building point cloud data, the pre-processing comprising at least one of splicing, down-sampling and denoising.

[0013] As an optional implementation of the embodiment of the present disclosure, the training process of the point cloud segmentation sub-model comprises: obtaining sample data; the sample data comprising building point cloud data of any construction stage; training an initial model using the sample data, and adjusting the model parameters of the initial model until the model converges to obtain the point cloud segmentation sub-model; wherein the initial model is constructed based on a ShellNet algorithm, and comprises: a convolution layer, a convolution operator corresponding to the convolution layer, a deconvolution layer, and a multilayer perceptron; and the model parameters of the initial model comprise: the output point number, the feature channel number and the number of spherical shells of the convolution layer, and the number of neighborhood points corresponding to each spherical shell.

[0014] As an optional implementation of the embodiment of the present disclosure, the structural safety feature points comprise: bottom feature points, top feature points and intermediate feature points; selecting the structural safety feature points from the planar registration point cloud data comprises: determining the bottom feature points and the top feature points in the planar registration point cloud data; and selecting the intermediate feature points between the bottom feature points and the top feature points according to a preset interval.

[0015] As an optional implementation of the embodiment of the present disclosure, the structural safety parameters comprise: vertical displacement variation, horizontal displacement variation, inclination variation and deflection variation; and calculating the structural safety parameters based on the structural safety feature points comprises: calculating the vertical displacement variation based on the bottom feature points; calculating the horizontal displacement variation based on the top feature points; calculating the inclination variation based on the displacement components of the bottom feature points and the top feature points and the point interval, the point interval being the distance between the bottom feature points and the top feature points.

[0016] calculating the deflection variation based on the bottom feature points, the top feature points and the center feature points; the center feature points being the center points of the bottom feature points and the top feature points.

[0017] In a second aspect, the present disclosure provides a building multi-stage construction process structural safety intelligent monitoring device, which comprises:

[0018] Acquisition module, used to obtain building point cloud data at different construction stages;

[0019] The segmentation and registration module is used to input the building point cloud data into the intelligent algorithm network model to obtain the plane registration point cloud data output by the intelligent algorithm network model; the intelligent algorithm network model is used to segment and register the building point cloud data;

[0020] A selection module is used to select structural safety feature points from the plane registration point cloud data;

[0021] The calculation module is used to calculate the structural safety parameters based on the structural safety feature points. The structural safety parameters are used to characterize the changes in settlement and deformation of the building during the construction process.

[0022] As an optional implementation of the embodiment of the present disclosure, the intelligent algorithm network model includes: a point cloud segmentation sub-model and a point cloud registration sub-model; the point cloud segmentation sub-model is used to segment the building point cloud data into multiple plane point cloud data; the point cloud registration sub-model is used to align the plane point cloud data of different construction stages.

[0023] As an optional implementation of the embodiment of the present disclosure, the device further includes a preprocessing module for preprocessing the building point cloud data, where the preprocessing includes at least one of splicing, downsampling, and denoising.

[0024] As an optional implementation of the embodiment of the present disclosure, the device also includes a training module for training a point cloud segmentation sub-model, including: obtaining sample data; the sample data includes building point cloud data at any construction stage; using the sample data to train an initial model, and adjusting the model parameters of the initial model until the model converges to obtain a point cloud segmentation sub-model; wherein the initial model is constructed based on the ShellNet algorithm, including: a convolution layer, a convolution operator corresponding to the convolution layer, a deconvolution layer, and a multi-layer perceptron; the model parameters of the initial model include: the number of output points of the convolution layer, the number of feature channels and the number of spherical shells, and the number of neighborhood points corresponding to each spherical shell.

[0025] As an optional implementation of the embodiment of the present disclosure, the structural safety feature points include: bottom feature points, top feature points and middle feature points; the selection module is specifically used to: determine the bottom feature points and top feature points in the plane alignment point cloud data; and select the middle feature points between the bottom feature points and the top feature points according to a preset spacing.

[0026] As an optional implementation of the embodiment of the present disclosure, the structural safety parameter comprises: a vertical displacement change amount, a horizontal displacement change amount, an inclination change amount and a deflection change amount; the calculation module is specifically configured to: calculate the vertical displacement change amount based on the bottom feature point; calculate the horizontal displacement change amount based on the top feature point; calculate the inclination change amount based on the displacement component of the bottom feature point, the displacement component of the top feature point and the point spacing, and the point spacing is the distance between the bottom feature point and the top feature point; calculate the deflection change amount based on the bottom feature point, the top feature point and the center feature point, and the center feature point is the center point of the bottom feature point and the top feature point.

[0027] In a third aspect, the present disclosure provides an electronic device, comprising: a processor, a memory and a computer program stored on the memory and executable on the processor, and when the computer program is executed by the processor, the building multi-stage construction process structural safety intelligent monitoring method according to the first aspect or any optional implementation thereof is implemented.

[0028] In a fourth aspect, the present disclosure provides a computer readable storage medium, comprising: a computer program stored on the computer readable storage medium, and when the computer program is executed by a processor, the building multi-stage construction process structural safety intelligent monitoring method according to the first aspect or any optional implementation thereof is implemented.

[0029] In a fifth aspect, the present disclosure provides a computer program product, comprising: the computer program product comprises a computer program, and when the computer program runs on a computer, the computer program makes the computer implement the building multi-stage construction process structural safety intelligent monitoring method according to the first aspect or any optional implementation thereof.

[0030] The technical scheme provided by the embodiment of the present disclosure has the following advantages compared with the prior art:

[0031] The embodiment of the present disclosure provides a building multi-stage construction process structural safety intelligent monitoring method, wherein the method first acquires building point cloud data in different construction stages, then inputs the building point cloud data into an intelligent algorithm network model, and intelligently segments and registers the building point cloud data by the intelligent algorithm network model to output plane registration point cloud data, and then selects structural safety feature points from the plane registration point cloud data, so as to calculate structural safety parameters for representing the settlement deformation change of the building in the construction process according to the structural safety feature points. In this way, the embodiment of the present disclosure improves the data processing efficiency and accuracy by intelligently segmenting and registering the building point cloud data by the intelligent algorithm network model, so as to realize accurate monitoring of the structural safety of the building in the construction process and ensure the safety of the building construction. BRIEF DESCRIPTION OF DRAWINGS

[0032] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0033] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0034] Figure 1 A schematic diagram of a flow chart of a method for intelligently monitoring structural safety during a multi-stage construction process of a building according to an embodiment of the present disclosure;

[0035] Figure 2 A schematic diagram of the structure of the initial model provided in the embodiment of the present disclosure;

[0036] Figure 3 A schematic diagram of a device for intelligently monitoring the structural safety of a building during a multi-stage construction process according to an embodiment of the present disclosure;

[0037] Figure 4 This is a structural diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0038] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features therein can be combined with each other in the absence of conflict.

[0039] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0040] In order to solve some or all of the technical problems existing in the related art, the embodiments of the present disclosure provide a method for intelligently monitoring the structural safety of a building during a multi-stage construction process, wherein the method first obtains building point cloud data at different construction stages, then inputs these building point cloud data into an intelligent algorithm network model, and the intelligent algorithm network model intelligently segments and aligns the building point cloud data to output plane-aligned point cloud data, and then selects structural safety feature points from the plane-aligned point cloud data to calculate structural safety parameters used to characterize the changes in settlement and deformation of the building during the construction process based on these structural safety feature points. In this way, the embodiments of the present disclosure improve data processing efficiency and accuracy by intelligently segmenting and aligning the building point cloud data through the intelligent algorithm network model, thereby achieving precise monitoring of the building structure safety during the construction process and ensuring the safety of building construction.

[0041] A method for intelligent monitoring of structural safety during a multi-stage construction process of a building provided in an embodiment of the present disclosure can be implemented by an intelligent monitoring device for structural safety during a multi-stage construction process of a building or an electronic device, and the electronic device includes but is not limited to a vehicle terminal, a server, a personal computer, a laptop computer, a tablet computer, a smart phone, etc. The operating system of the electronic device may include Android, a mobile operating system (iOS) developed by Apple, an operating system (Windows) developed by Microsoft Corporation of the United States, etc., and the embodiment of the present disclosure does not limit this. The electronic device can be run alone to implement the present disclosure, or it can be connected to a network and implement the present disclosure through interactive operations with other computer devices in the network. Among them, the network in which the electronic device is located includes but is not limited to the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), etc.

[0042] It should be noted that the protection scope of the method for intelligent monitoring of structural safety during a multi-stage construction process of a building described in the embodiment of the present disclosure is not limited to the execution order of the steps listed in this embodiment. All solutions implemented by adding, reducing or replacing steps in the existing technology based on the principles of the present disclosure are included in the protection scope of the present disclosure.

[0043] like Figure 1 As shown, Figure 1 This is a flow chart of a method for intelligently monitoring the structural safety of a building during a multi-stage construction process according to an embodiment of the present disclosure. This method can be executed by an intelligent monitoring device for the structural safety of a building during a multi-stage construction process, wherein the device can be implemented using software and / or hardware and can generally be integrated into an electronic device. Figure 1 As shown, the method mainly includes the following steps S101 to S104:

[0044] S101. Acquire building point cloud data at different construction stages.

[0045] In some embodiments, 3D laser scanners, binocular cameras, and other equipment are used to acquire building point cloud data at different construction stages. A multi-point measurement approach can be employed, with the number of measurement points appropriately planned based on the actual size of the construction site. Measurement points can be spaced at regular intervals to collect building point cloud data at different construction stages.

[0046] For example, a FARO Focus Premium 3D laser scanner is used for color point cloud collection. The scanner has a measurement range of 70 m, a measurement error of 1 mm, a measurement rate of 2 million points / second, and an angular measurement accuracy of 19 arc seconds (vertical angle / horizontal angle).

[0047] In some embodiments, after acquiring the building point cloud data at different construction stages, the building point cloud data is preprocessed, and the preprocessing includes but is not limited to: splicing, downsampling, and denoising.

[0048] In the process of splicing building point cloud data, the building point cloud data of multiple measuring points are unified into the same coordinate system to complete the point cloud splicing and obtain the complete point cloud data of the entire building.

[0049] Multi-point measurement can obtain accurate scene distribution, but due to the large and dense volume of building point cloud data, it is necessary to downsample the building point cloud data. In the process of downsampling the building point cloud data, random downsampling and other sampling methods can be used to delete the building point cloud data without losing important features of the point cloud, which can improve the efficiency of subsequent point cloud processing and reduce the computational pressure of subsequent algorithms.

[0050] In some embodiments, denoising building point cloud data includes: using a statistical filter to remove outliers from the downsampled point cloud. The statistical filter performs a statistical analysis on the neighborhood of each point and calculates the average distance from it to all points in the neighborhood. Assuming that the average distance between any point in the point cloud and its k nearest neighbors satisfies a Gaussian distribution, a distance threshold σ is determined based on a standard range defined by the global distance mean and variance. When the average distance between a point and its k nearest neighbors is greater than the threshold σ, the point is determined to be an outlier and is removed from the data. Denoising building point cloud data is performed to remove a large number of noise points and isolated points in the building point cloud data, thereby improving the accuracy of point cloud processing.

[0051] The above embodiment obtains building point cloud data at different construction stages. Optionally, the building point cloud is pre-processed to retain valid point cloud data, which is conducive to improving the efficiency and accuracy of subsequent point cloud processing.

[0052] S102: Input the building point cloud data into the intelligent algorithm network model to obtain the plane registration point cloud data output by the intelligent algorithm network model.

[0053] Among them, the intelligent algorithm network model is used to segment and register building point cloud data.

[0054] In some embodiments, the intelligent algorithm network model includes: a point cloud segmentation sub-model and a point cloud registration sub-model. The point cloud segmentation sub-model is used to segment the building point cloud data into multiple planar point cloud data; the point cloud registration sub-model is used to register the planar point cloud data at different construction stages.

[0055] The point cloud segmentation sub-model is configured to: randomly extract sampling points from the building point cloud data; determine a point set centered on the sampling point based on the sampling point and the preset number of neighborhood points, and distribute the point set on a spherical shell centered on the sampling point; expand the radius of the sphere centered on the sampling point, and when the number of points within the sphere equals the preset number of neighborhood points, use the sphere of the current radius as the first spherical shell; continue to expand the radius of the sphere from the current radius until the number of points contained between the sphere and the first spherical shell equals the preset number of neighborhood points, and use the sphere at this time as the second spherical shell; process the points within the first and second spherical shells to obtain planar point cloud data. The processing of the points within the first and second spherical shells includes feature pooling, feature dimension upscaling, and sequential convolution. Feature pooling is to aggregate the feature information of the first and second spherical shells through a maximum pooling layer; feature dimension upscaling is to perform dimension upscaling on the feature information through a multi-layer perceptron; and sequential convolution is to perform one-dimensional convolution on the feature information in an inside-out order.

[0056] The point cloud registration sub-model is configured as follows: taking the planar point cloud data of any construction stage as the source point cloud data, and the planar point cloud data of the previous construction stage as the target point cloud data; determining matching point pairs from the source point cloud data and the target point cloud data; and calculating the pose transformation matrix based on the matching point pairs to transform the source point cloud data into the target point cloud data.

[0057] The following will explain the point cloud segmentation sub-model and the point cloud configuration sub-model in detail:

[0058] Point Cloud Segmentation Sub-Model

[0059] During the construction process, the overall structure of the building has differences in spatial distribution, and the structural stress and deformation conditions of different spatial planes are also different. Therefore, in order to be able to fully and accurately evaluate the safety characteristics of the entire building during the construction process, the embodiment of the present disclosure uses a point cloud segmentation sub-model to segment the building point cloud data into multiple different plane point cloud data based on the three-dimensional spatial distribution of the building, which is convenient for subsequent structural safety monitoring.

[0060] Optionally, the point cloud segmentation sub-model is built based on the ShellNet algorithm.

[0061] The ShellNet algorithm consists of a convolutional layer (encoder), a deconvolutional layer (decoder), and a convolution operator (ShellConv). ShellConv defines convolution on a domain that can be divided by concentric spherical shells. It then defines the order of convolutions from the inner shell to the outer shell to enable neighborhood point queries and address point cloud disorder. By establishing concentric spherical shells, ShellConv achieves a larger receptive field without increasing the number of layers. The receptive field is defined as the area of ​​the input that a convolutional neural network feature can see.

[0062] The point cloud segmentation sub-model includes a convolutional layer, a convolution operator corresponding to the convolutional layer, and a deconvolution layer; it may also include a multilayer perceptron (MLP).

[0063] In some embodiments, the training process of the point cloud segmentation sub-model includes: first obtaining sample data; the sample data includes building point cloud data at any construction stage; then using the sample data to train the initial model, adjusting the model parameters of the initial model until the model converges to obtain the point cloud segmentation sub-model.

[0064] The initial model is built based on the ShellNet algorithm, including: convolutional layer, convolution operator corresponding to the convolutional layer, deconvolution layer, and multi-layer perceptron; the model parameters of the initial model include: the number of output points of the convolutional layer, the number of feature channels and the number of spherical shells, as well as the number of neighborhood points corresponding to each spherical shell.

[0065] Optionally, the sample data is divided into a training set, a test set, and a validation set according to a preset ratio. Exemplarily, the preset ratio is 7:2:1. The training set is used to estimate the model, the validation set is used to determine the network structure or model parameters that control model complexity, and the test set is used to verify the performance of the converged model.

[0066] When constructing the initial model, for example, Figure 2 As shown, Figure 2 A schematic diagram of the structure of the initial model provided in the embodiment of the present disclosure. The initial model may include three convolutional layers (labeled as encoders in the figure) and three deconvolutional layers (labeled as decoders in the figure). Each convolutional layer contains a convolution operator ShellConv. The i-th layer encoder, i = 0, 1, 2, represents the first layer encoder, the second layer encoder, and the third layer encoder, respectively. The model parameters of the initial model include: the number of output points N of the convolutional layer i , number of feature channels C i , number of spherical shells S i And the number of neighborhood points K corresponding to each spherical shell i .

[0067] For the initial model, the number of encoder output points N iSet them to 512, 128, and 32 respectively; set the number of feature channels C i Set them to 128, 256 and 512 respectively; set the number of spherical shells S i Set them to 4, 2, and 1 respectively; set the number of neighborhood points K corresponding to each spherical shell i Set to 32, 16, and 8. The number of points contained in each spherical shell is defined as the size of the spherical shell, and the value is set to 8. C in the decoder is set to 64 in the last convolution for segmentation.

[0068] The initial model also includes a multi-layer perceptron, Figure 2 The initial model shown in Figure 1 includes three MLPs, with the feature channels corresponding to each MLP initially set to 128, 64, and 6, respectively. The number of feature channels in the last layer corresponds to the number of multidimensional spaces of the building structure. Furthermore, to prevent overfitting during training, this embodiment of the disclosure can include a regularization (Dropout) layer between the MLPs, with a ratio of 0.5.

[0069] After the initial model is built, the sample data is used to train the initial model, and the model parameters of the initial model are adjusted until the model converges to obtain the point cloud segmentation sub-model.

[0070] Optionally, the initial model is trained using sample data according to preset hyperparameters. Preset hyperparameters include, but are not limited to, batch size, number of iterations, and learning rate. For example, the batch size is set to 4, the number of iterations is set to 50, and the learning rate is set to 0.001. Furthermore, the model training method can be based on an adaptive motion estimation algorithm (Adam optimization algorithm), and the loss function for model training can be a cross-entropy loss function. Model evaluation metrics include, but are not limited to, accuracy, precision, recall, and F1-score.

[0071] Taking a single point p and its neighborhood point set q as an example, point p and neighborhood point set q satisfy the relationship shown in the following formula (1):

[0072]

[0073] Where F is the input feature of a feature channel; n is the number of layers; q∈Ω p ,Ω p It is the neighborhood point set of the nearest neighbor query. It should be noted that ShellConv uses a set of multi-scale concentric spheres to define the area. The part between the two spheres forms a spherical shell. The point cloud data contained in each spherical shell is a set of neighborhood point sets Ω. p ;w(q) n is the convolution weight of the neighborhood point set q, which is a fixed-size trainable parameter vector.

[0074] The structure of the trained point cloud segmentation sub-model is similar to that of the initial model, such as Figure 2 As shown, the model parameters of the two are different.

[0075] After training the point cloud segmentation sub-model to convergence, when applying the point cloud segmentation sub-model, the process of segmenting the building point cloud data includes: first, randomly extracting a number of points from the building point cloud data of any construction stage as sampling points; setting the number of neighborhood points to obtain a point set centered on the sampling point, and distributing the point set on the corresponding concentric spherical shells; expanding the radius of the sphere centered on the sampling point; when the number of points within the sphere meets the set number of neighborhood points, the sphere of that radius is used as the first spherical shell; continuing to expand the sphere radius until the number of points contained between the sphere radius and the first spherical shell meets the set number of neighborhood points, and the sphere of this radius is used as the second spherical shell. Then, the feature information in each spherical shell is aggregated through a maximum pooling layer, and the features are then dimensionalized through a multi-layer perceptron. Further one-dimensional convolution is performed in an inside-out order, and the output is multiple planar point cloud data obtained by segmenting the building point cloud data of that construction stage.

[0076] refer to Figure 2 As shown in the structural diagram, the building point cloud data at different construction stages is taken as input. The building point cloud data is first downsampled layer by layer and passed through three convolution operators ShellConv. The number of feature channels of the corresponding output increases layer by layer to obtain an N2×C2 matrix, where N2 represents the number of sampling points extracted from the building point cloud data and C2 represents the number of feature channels of each sampling point. The N2 feature sampling points are decoded and the convolution operator is used. The number of output points of the decoder increases layer by layer and the number of feature channels decreases layer by layer. The encoder and decoder with the same matrix size are spliced ​​to obtain an N×C feature matrix. Furthermore, the N×C feature matrix is ​​input into the Multilayer Perceptron (MLP) to output N×k seg Matrix, representing k seg Planar point cloud data. Among them, k seg Indicates the number of labels obtained by segmenting the building point cloud data. In this invention, the number of labels represents the multidimensional space number of the building structure.

[0077] Point Cloud Registration Sub-Model

[0078] The construction process often involves multiple phases. To monitor building safety throughout the entire construction process, it is necessary to understand the changing patterns of building point cloud data at different stages and to perform point cloud registration for these phases. The goal of point cloud registration is to find a coordinate transformation or rigid body transformation (such as translation or rotation) that accurately aligns corresponding points across multiple point clouds. Registration ensures consistency and alignment between point clouds from different locations or perspectives, providing more comprehensive and accurate 3D information.

[0079] In some embodiments, after the building point cloud data is segmented by the point cloud segmentation sub-model, the registration stage includes coarse registration and fine registration. The plane point cloud data of any construction stage can be used as the source point cloud data, and the plane point cloud data of the previous construction stage can be used as the target point cloud data to determine the center coordinates of the point cloud data, and extract the feature points around the building for coarse registration. It can be understood that the intelligent algorithm network model also includes a coarse registration sub-module. The coarse registration sub-module is used to perform a relatively rough registration when the transformation between the source point cloud data and the target point cloud data is completely unknown, and its main purpose is to provide a better initial value of the transformation for fine registration. Subsequently, the plane point cloud data after coarse registration is used as the input of the point cloud registration sub-model for further fine registration.

[0080] The point cloud registration sub-model is built based on the Iterative Closest Point (ICP) algorithm. It takes as input the segmented planar point cloud data from different construction stages, using the planar point cloud data from any construction stage as the source point cloud data and the planar point cloud data from the previous construction stage as the target point cloud data. It then searches for matching point pairs, assuming the data is noisy, and removes erroneous point pairs. It then solves for the pose transformation matrix, which is used to transform the source point cloud data into the target point cloud data.

[0081] Among them, the ICP algorithm is an iterative method to correct the rigid body transformation between two point clouds to minimize the distance between the point clouds.

[0082] Specifically, for any point P in the source point cloud data P i , first perform the initial transformation T0 and get point P i '; Find the point P from the target point cloud data Q i ′ is the nearest point Q i , point P i and Q i As a set of matching point pairs; calculate the sum of the distance differences of all matching point pairs, and solve the transformation matrix when the sum of the distance differences is minimized. The transformation matrix T(R, t) can be calculated according to the following formula (2):

[0083]

[0084] Where n represents the number of points in the source point cloud data P, R is the rotation matrix, and t is the translation matrix.

[0085] Then determine whether the convergence conditions are met, which include at least one of the following: the maximum distance difference of the matching point pair is less than a preset threshold; the difference between the maximum distance of the matching point pair and the sum of the distance differences of the two previous and subsequent iterations is less than a preset threshold; whether the number of iterations is greater than a preset number.

[0086] If the convergence condition is met, the calculated transformation matrix T(R, t) is output as the optimal transformation matrix; if the convergence condition is not met, the initial transformation matrix is ​​updated to Continue iterating until convergence and output the final pose transformation matrix.

[0087] In some embodiments, after convergence, the accuracy of point cloud registration can be ensured by measuring the size of indicators such as the root mean square error (RMSE). The calculation formula for RMSE is as follows:

[0088]

[0089] In formula (3), n is the number of points in the plane point cloud data, D i is the Euclidean distance between the matching point pairs after registration, is the true value of the Euclidean distance between matching point pairs, which is usually 0.

[0090] The above embodiment uses an intelligent algorithm network model to segment and align building point cloud data, thereby improving the efficiency and intelligence level of point cloud processing.

[0091] S103: Select structural safety feature points from the plane registration point cloud data.

[0092] During construction, the top and bottom of a building structure are vulnerable to settlement and deformation, requiring special attention. Therefore, in the disclosed embodiments, structural safety feature points may include top and bottom feature points. First, the bottom and top feature points are selected from the plane registration point cloud data, and then other points are selected at a preset spacing as structural safety feature points. The preset spacing can be set according to the size of the building plan.

[0093] In some embodiments, the center point coordinates of the plane registration point cloud data are extracted, and the top and bottom feature points are determined based on the plane dimensions of the plane registration point cloud data. Intermediate feature points between the top and bottom feature points are obtained at a preset spacing. The top, bottom, and intermediate feature points are used as structural safety feature points.

[0094] S104. Calculate structural safety parameters based on the structural safety feature points.

[0095] Structural safety parameters reflect the overall change trend of the building structure during construction and characterize the building's settlement and deformation changes during construction. Structural safety parameters include, but are not limited to, displacement change, inclination change, and deflection change. Displacement change includes vertical displacement change and lateral displacement change (horizontal displacement change).

[0096] In some embodiments, in the process of calculating the displacement change, the vertical displacement change is calculated based on the bottom feature points.

[0097] Specifically, the vertical displacement change is calculated according to the following formula (4):

[0098]

[0099] In formula (4), Represents the structural safety feature point N at the current construction stage i Projection in the Z-axis direction; Represents the structural safety characteristic point N of the previous construction stage i Projection in the Z-axis direction.

[0100] The horizontal displacement change is calculated based on the top feature point. Specifically, the horizontal displacement change is calculated according to the following formula (5):

[0101]

[0102] In formula (5), Represents the structural safety feature point M at the current construction stage i Projection in the X-axis direction; Represents the structural safety characteristic point M of the previous construction stage i Projection in the X-axis direction.

[0103] It should be noted that during the construction process, the bottom of the building is prone to settlement changes and the top is prone to horizontal displacement. The structural safety feature point N i It can be the bottom feature point, structural safety feature point M i Can be a top feature point.

[0104] In some embodiments, in the process of calculating the inclination change, the inclination change is calculated based on the displacement component of the bottom feature point, the displacement component of the top feature point, and the point spacing, where the point spacing is the distance between the bottom feature point and the top feature point.

[0105] Specifically, the tilt angle change Δα is calculated according to the following formula (6): i :

[0106] Δα i =α i -α i ′(6)

[0107] In formula (6),

[0108]

[0109] in, Represents the structural safety feature point M at the current construction stage i The displacement component can be the structural safety feature point M in the current construction stage. i Projection in the Z-axis direction; Represents the structural safety feature point N at the current construction stage i The displacement component can be the structural safety characteristic point N in the current construction stage. i Projection in the Z-axis direction; Represents the structural safety characteristic point M of the previous construction stage i The displacement component can be the structural safety feature point M of the previous construction stage i Projection in the Z-axis direction; Represents the structural safety characteristic point N of the previous construction stage i The displacement component can be the projection of the structural safety characteristic point Ni in the previous construction stage in the Z-axis direction. Represents the structural safety feature point M at the current construction stage i 、N i The distance between.

[0110] In some embodiments, in the process of calculating the deflection change, the deflection change is calculated based on the bottom feature point, the top feature point and the center feature point; wherein the center feature point is the center point of the bottom feature point and the top feature point.

[0111] Specifically, the deflection change Δf is calculated according to the following formula (7): i :

[0112] Δf i =f i -f i ′(7)

[0113] In formula (7),

[0114]

[0115] In formula (7-1) and (7-2),

[0116]

[0117] Pi Point is the structural safety characteristic point M in the current construction stage i 、N i The center point of the point; P represents the current construction stage i The displacement component of a point can be either vertical or horizontal. Indicates M i 、P i the distance between them; Indicates P i 、N i The distance between them.

[0118] After calculating the structural safety parameters, the system compares them with the parameter ranges specified in relevant construction regulations and traditional on-site monitoring data to analyze the overall structural changes during construction and the building's most dangerous locations. Furthermore, for buildings not yet under construction, the calculated structural safety parameters for existing buildings can be used to provide guidance and control on key issues in the subsequent construction of other buildings, such as the specific stages of construction and the protection of the most dangerous locations, effectively ensuring the overall safety of the structure during future construction.

[0119] In summary, the disclosed embodiments provide a method for intelligently monitoring the structural safety of a building during a multi-stage construction process. The method first acquires building point cloud data at different construction stages, then inputs this building point cloud data into an intelligent algorithm network model. The intelligent algorithm network model then intelligently segments and registers the building point cloud data to output plane-registered point cloud data. Structural safety feature points are then selected from the plane-registered point cloud data to calculate structural safety parameters that characterize the changes in settlement and deformation of the building during construction based on these structural safety feature points. Thus, the disclosed embodiments improve data processing efficiency and accuracy by intelligently segmenting and registering building point cloud data using the intelligent algorithm network model, thereby achieving precise monitoring of the building's structural safety during construction and ensuring building construction safety.

[0120] like Figure 3 As shown, Figure 3 This is a schematic diagram of a device for intelligently monitoring the structural safety of a building during a multi-stage construction process provided by an embodiment of the present disclosure. The device includes:

[0121] Acquisition module 301, for acquiring building point cloud data at different construction stages;

[0122] The segmentation and registration module 302 is used to input the building point cloud data into the intelligent algorithm network model to obtain the plane registration point cloud data output by the intelligent algorithm network model; the intelligent algorithm network model is used to segment and register the building point cloud data;

[0123] A selection module 303 is used to select structural safety feature points from the plane registration point cloud data;

[0124] The calculation module 304 is used to calculate the structural safety parameters based on the structural safety feature points. The structural safety parameters are used to characterize the settlement and deformation changes of the building during the construction process.

[0125] As an optional implementation of the embodiment of the present disclosure, the intelligent algorithm network model includes: a point cloud segmentation sub-model and a point cloud registration sub-model; the point cloud segmentation sub-model is used to segment the building point cloud data into multiple plane point cloud data; the point cloud registration sub-model is used to align the plane point cloud data of different construction stages.

[0126] As an optional implementation of the embodiment of the present disclosure, the device further includes a preprocessing module for preprocessing the building point cloud data, where the preprocessing includes at least one of splicing, downsampling, and denoising.

[0127] As an optional implementation of the embodiment of the present disclosure, the device also includes a training module for training a point cloud segmentation sub-model, including: obtaining sample data; the sample data includes building point cloud data at any construction stage; using the sample data to train an initial model, and adjusting the model parameters of the initial model until the model converges to obtain a point cloud segmentation sub-model; wherein the initial model is constructed based on the ShellNet algorithm, including: a convolution layer, a convolution operator corresponding to the convolution layer, a deconvolution layer, and a multi-layer perceptron; the model parameters of the initial model include: the number of output points of the convolution layer, the number of feature channels and the number of spherical shells, and the number of neighborhood points corresponding to each spherical shell.

[0128] As an optional implementation of the embodiment of the present disclosure, the structural safety feature points include: bottom feature points, top feature points and middle feature points; the selection module 303 is specifically used to: determine the bottom feature points and top feature points in the plane alignment point cloud data; and select the middle feature points between the bottom feature points and the top feature points according to a preset spacing.

[0129] As an optional implementation of the embodiment of the present disclosure, the structural safety parameters include: vertical displacement change, horizontal displacement change, inclination change and deflection change; the calculation module 304 is specifically used to: calculate the vertical displacement change based on the bottom feature point; calculate the horizontal displacement change based on the top feature point; calculate the inclination change based on the displacement component of the bottom feature point and the displacement component of the top feature point and the point spacing, where the point spacing is the distance between the bottom feature point and the top feature point; calculate the deflection change based on the bottom feature point, the top feature point and the center feature point; the center feature point is the center point of the bottom feature point and the top feature point.

[0130] like Figure 4As shown, an embodiment of the present disclosure provides an electronic device comprising: a processor 401, a memory 402, and a computer program stored in the memory 402 and executable on the processor 401. When executed by the processor 401, the computer program implements the various processes of the method for intelligently monitoring structural safety during a multi-stage construction process of a building in the above-described method embodiment. The same technical effects can be achieved, and to avoid repetition, they will not be described here.

[0131] An embodiment of the present disclosure provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, each process of the method for intelligent monitoring of structural safety during a multi-stage construction process of a building in the above-mentioned method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0132] The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0133] An embodiment of the present disclosure provides a computer program product, which stores a computer program. When the computer program is executed by a processor, each process of the method for intelligent monitoring of structural safety during a multi-stage construction process of a building in the above-mentioned method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0134] Those skilled in the art will appreciate that embodiments of the present disclosure may be provided as methods, systems, or computer program products. Thus, the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0135] In the several embodiments provided by the present disclosure, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0136] In the present disclosure, a processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0137] In this disclosure, 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. Memory is an example of a computer-readable medium.

[0138] In this disclosure, computer-readable media includes permanent and non-permanent, removable and non-removable storage media. Storage media can be implemented by any method or technology to store information, and 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 technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic 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 transitory media such as modulated data signals and carrier waves.

[0139] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device that includes the element.

[0140] The above are merely specific embodiments of the present disclosure, intended to enable those skilled in the art to understand and implement the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure is not limited to these embodiments, but is to be construed in the broadest manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent monitoring of structural safety during a multi-stage construction process of a building, characterized in that: include: Obtain building point cloud data at different construction stages; Inputting the building point cloud data into an intelligent algorithm network model to obtain plane registration point cloud data output by the intelligent algorithm network model; The intelligent algorithm network model is used to segment and register the building point cloud data; The intelligent algorithm network model includes: a point cloud segmentation sub-model and a point cloud registration sub-model; the point cloud segmentation sub-model is used to segment the building point cloud data into multiple plane point cloud data; the point cloud registration sub-model is used to register the plane point cloud data at different construction stages; Wherein, the point cloud segmentation sub-model is configured as follows: randomly extracting sampling points from the building point cloud data; determining a point set centered on the sampling point according to the sampling point and the preset number of neighborhood points, and the point set is distributed on a spherical shell centered on the sampling point; expanding the radius of the sphere centered on the sampling point, and when the number of points in the sphere is equal to the preset number of neighborhood points, using the sphere of the current radius as the first spherical shell; continuing to expand the radius of the sphere from the current radius until the number of points contained between the sphere and the first spherical shell is equal to the preset number of neighborhood points, using the sphere at this time as the second spherical shell; processing the points in the first spherical shell and the second spherical shell to obtain planar point cloud data; wherein, processing the points in the first spherical shell and the second spherical shell includes: feature pooling, feature dimensionality increase and sequential convolution; feature pooling is to aggregate the feature information of the first spherical shell and the second spherical shell through the maximum pooling layer; feature dimensionality increase is to increase the dimension of the feature information through the multi-layer perceptron; sequential convolution is to perform one-dimensional convolution on the feature information in an inside-out order; The point cloud registration sub-model is configured to: use the plane point cloud data of any construction stage as source point cloud data, and the plane point cloud data of the previous construction stage as target point cloud data; determine matching point pairs from the source point cloud data and the target point cloud data; calculate the pose transformation matrix based on the matching point pairs to transform the source point cloud data into the target point cloud data; Selecting structural safety feature points from the plane registration point cloud data, the structural safety feature points including: bottom feature points, top feature points and middle feature points; Based on the structural safety feature points, structural safety parameters are calculated. The structural safety parameters are used to characterize the settlement and deformation changes of the building during construction. The structural safety parameters include: vertical displacement change, horizontal displacement change, inclination angle change, and deflection change.

2. The method according to claim 1, characterized in that After acquiring the building point cloud data at different construction stages and before inputting the building point cloud data into the intelligent algorithm network model and obtaining the plane registration point cloud data output by the intelligent algorithm network model, the method further includes: Preprocessing is performed on the building point cloud data, where the preprocessing includes at least one of stitching, downsampling, and denoising.

3. The method according to claim 1, characterized in that The training process of the point cloud segmentation sub-model includes: Acquire sample data; the sample data includes building point cloud data at any construction stage; train an initial model using the sample data, and adjust model parameters of the initial model until the model converges to obtain the point cloud segmentation sub-model; The initial model is constructed based on the ShellNet algorithm, including: a convolutional layer, a convolution operator corresponding to the convolutional layer, a deconvolution layer, and a multilayer perceptron; the model parameters of the initial model include: the number of output points of the convolutional layer, the number of feature channels and the number of spherical shells, and the number of neighborhood points corresponding to each spherical shell.

4. The method according to claim 1, wherein The selecting of structural safety feature points from the plane registration point cloud data includes: Determining bottom feature points and top feature points in the plane registration point cloud data; According to a preset distance, an intermediate feature point between the bottom feature point and the top feature point is selected.

5. The method according to claim 4, characterized in that The calculating of the structural safety parameters based on the structural safety feature points includes: Calculating the vertical displacement change based on the bottom feature points; Calculating the horizontal displacement change based on the top feature point; Calculating the tilt angle variation based on the displacement component of the bottom feature point, the displacement component of the top feature point, and a point spacing, where the point spacing is the distance between the bottom feature point and the top feature point; The deflection variation is calculated based on the bottom feature point, the top feature point, and a center feature point; the center feature point is the center point between the bottom feature point and the top feature point.

6. An intelligent monitoring device for structural safety during a multi-stage construction process of a building, characterized in that: include: Acquisition module, used to obtain building point cloud data at different construction stages; A segmentation and registration module is used to input the building point cloud data into an intelligent algorithm network model to obtain plane registration point cloud data output by the intelligent algorithm network model; The intelligent algorithm network model is used to segment and register the building point cloud data; The intelligent algorithm network model includes: a point cloud segmentation sub-model and a point cloud registration sub-model; the point cloud segmentation sub-model is used to segment the building point cloud data into multiple plane point cloud data; the point cloud registration sub-model is used to register the plane point cloud data at different construction stages; Wherein, the point cloud segmentation sub-model is configured as follows: randomly extracting sampling points from the building point cloud data; determining a point set centered on the sampling point according to the sampling point and the preset number of neighborhood points, and the point set is distributed on a spherical shell centered on the sampling point; expanding the radius of the sphere centered on the sampling point, and when the number of points in the sphere is equal to the preset number of neighborhood points, using the sphere of the current radius as the first spherical shell; continuing to expand the radius of the sphere from the current radius until the number of points contained between the sphere and the first spherical shell is equal to the preset number of neighborhood points, using the sphere at this time as the second spherical shell; processing the points in the first spherical shell and the second spherical shell to obtain planar point cloud data; wherein, processing the points in the first spherical shell and the second spherical shell includes: feature pooling, feature dimensionality increase and sequential convolution; feature pooling is to aggregate the feature information of the first spherical shell and the second spherical shell through the maximum pooling layer; feature dimensionality increase is to increase the dimension of the feature information through the multi-layer perceptron; sequential convolution is to perform one-dimensional convolution on the feature information in an inside-out order; The point cloud registration sub-model is configured to: use the plane point cloud data of any construction stage as source point cloud data, and the plane point cloud data of the previous construction stage as target point cloud data; determine matching point pairs from the source point cloud data and the target point cloud data; calculate the pose transformation matrix based on the matching point pairs to transform the source point cloud data into the target point cloud data; A selection module, configured to select structural safety feature points from the plane registration point cloud data, wherein the structural safety feature points include: bottom feature points, top feature points, and middle feature points; A calculation module is used to calculate structural safety parameters based on the structural safety feature points. The structural safety parameters are used to characterize the changes in settlement and deformation of the building during construction. The structural safety parameters include: vertical displacement change, horizontal displacement change, inclination angle change, and deflection change.

7. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the method for intelligent monitoring of structural safety during a multi-stage construction process of a building as claimed in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that include: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for intelligently monitoring structural safety during a multi-stage construction process of a building according to any one of claims 1 to 5 is implemented.

9. A computer program product, characterized in that include: The computer program product includes a computer program, and when the computer program is run on a computer, the computer is enabled to implement the method for intelligent monitoring of structural safety during a multi-stage construction process of a building according to any one of claims 1 to 5.

Images