A three-dimensional model fitting method and system for power grid facilities based on panoramic point cloud

Abstract

The present invention discloses a method and system for fitting a three-dimensional model of power grid facilities based on a panoramic point cloud, and relates to the technical field of three-dimensional model fitting. The method comprises: collecting panoramic point cloud data and high-definition image data of power grid facilities, and performing data preprocessing; importing a BIM model of power grid facilities, and establishing a homologous coordinate system with the panoramic point cloud data; aligning the panoramic point cloud data and the BIM model, and extracting key features of the power grid facilities from the aligned point cloud data; adjusting the vertices and edges of the BIM model based on the extracted key features, so that the BIM model matches the point cloud data, and ensuring the accuracy and reliability of the BIM model through consistency verification; generating a three-dimensional model consistent with the actual power grid facilities based on the corrected BIM model for real-time monitoring and defect detection. The present invention can reduce the operation and maintenance risks caused by data deviation, reduce human intervention, and improve the automation level of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional model fitting, and in particular to a three-dimensional model fitting method and system for power grid facilities based on panoramic point cloud. Background Art

[0002] With the rapid development of drone technology, the power grid industry has gradually adopted drones for inspection and maintenance of power facilities (such as towers, etc.). Drones can be used to efficiently collect three-dimensional point cloud data of power grid facilities. These data provide a basis for real-time monitoring, defect detection and maintenance decisions of facilities. However, the collection of point cloud data is limited by factors such as equipment accuracy and environmental changes, and usually has geometric deviations from existing three-dimensional building information models (BIM) models. The existing BIM model can intuitively reflect the structure and properties of power grid facilities, but there are still many technical challenges to align the point cloud data collected by drones with the BIM model and achieve accurate three-dimensional model fitting. These challenges include high-precision requirements for data alignment, noise filtering, and maintaining consistency of geometric features during data fitting.

[0003] At present, the power grid industry still lacks technical methods that can effectively solve the geometric deviation between point cloud data and BIM models. Existing point cloud data fitting algorithms often cannot meet the needs of efficiently and accurately correcting BIM models. Especially in the case of large-scale and complex geometric structures such as power grid facilities, how to preprocess, align, extract key features of point cloud data, and accurately adjust BIM models are still technical challenges in this field. Therefore, it is particularly important to develop a method that can automatically correct the deviation between point cloud and BIM model and generate a three-dimensional model that is highly consistent with the actual facilities. Summary of the invention

[0004] In response to the above problems, the present invention proposes a three-dimensional model fitting method and system for power grid facilities based on panoramic point cloud. Through the refined processing and automatic fitting of point cloud data, the corrected three-dimensional model is ensured to be highly consistent with the actual power facilities, which greatly improves the accuracy and reliability of the model.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0006] A three-dimensional model fitting method for power grid facilities based on panoramic point cloud, the method comprising:

[0007] The drone-mounted laser radar and high-definition camera collects panoramic cloud data and high-definition image data of power grid facilities;

[0008] Preprocessing the panoramic point cloud data to remove noise and redundant data, and using a filtering algorithm to balance the point cloud density;

[0009] Import the BIM model of the power grid facilities and establish a homologous coordinate system with the panoramic cloud data to ensure that the coordinate systems of different data sources can match each other;

[0010] Aligning the panoramic point cloud data with the BIM model, and extracting key features of the power grid facilities from the aligned point cloud data;

[0011] Adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data, and ensure the accuracy and reliability of the BIM model through consistency verification;

[0012] Based on the revised BIM model, a three-dimensional model consistent with the actual power grid facilities is generated for real-time monitoring and defect detection.

[0013] As a preferred solution of the present invention, the panoramic point cloud data is preprocessed, including filtering the geometric line segments in the panoramic point cloud data, and the formula is:

[0014] ;

[0015] Where E represents the energy function; N represents the number of all candidate line segments; represents the number of supporting points of the i-th line segment; is the density of points within a predetermined unit length; is the length of the i-th line segment; is the selection status of the i-th line segment, Indicates that the line segment is selected. Indicates that the line segment is not selected.

[0016] As a preferred solution of the present invention, the method of importing the BIM model of the power grid facilities and establishing a homologous coordinate system with the panoramic point cloud data includes:

[0017] Import the BIM model of the power grid facilities and define the vertex set M as ,in, represents the three-dimensional coordinates of the j-th vertex in the BIM model, is the total number of vertices;

[0018] Process the preprocessed panoramic point cloud data P to extract geometric features corresponding to the BIM model; ,in, represents the coordinates of the jth point in the panoramic point cloud data, is the number of points in the panoramic point cloud data;

[0019] A global Gaussian mixture model is constructed to describe the geometric distribution of the panoramic cloud data and the BIM model. The formula is:

[0020] ;

[0021] In the formula, is the mean vector of the kth global Gaussian mixture model component; is the mean value of the panoramic cloud data; is the mean value of the BIM model; is the covariance matrix of the kth global Gaussian mixture model component; is the covariance matrix of the panoramic cloud data; is the covariance matrix of the BIM model; is the covariance matrix between the panoramic point cloud data and the BIM model; Represents the transpose operation of a matrix or vector;

[0022] The global Gaussian mixture model is used to match geometric features, and the geometric consistency of the panoramic point cloud data and the BIM model is determined by comparing the mean and covariance in the local area.

[0023] Apply an affine or rigid transformation by computing a rotation matrix and translation vectors , complete the coordinate conversion, transform the panoramic point cloud data into the coordinate system of the BIM model, the formula is:

[0024] ;

[0025] In the formula, Represents the transformed point cloud coordinates;

[0026] Thin plate spline interpolation TPS is used to process complex deformation, which can be expressed as:

[0027] ;

[0028] In the formula, Represents a nonlinear transformation function.

[0029] As a preferred solution of the present invention, the panoramic point cloud data and the BIM model are aligned, including using the log-likelihood method to align the data.

[0030] Based on the constructed global Gaussian mixture model, the log-likelihood method is used to calculate the degree of alignment between each point in the panoramic point cloud data and the BIM model. The calculation formula is:

[0031] ;

[0032] In the formula, Indicates the midpoint of the panoramic point cloud data Log-likelihood with BIM model; For point The probability of belonging to any part of the BIM model; is the weight of the kth global Gaussian mixture model component; K is the total number of Gaussian components; is the kth Gaussian distribution lower point The probability density of and the covariance matrix Sure;

[0033] The log-likelihood of all panoramic point cloud data is summed and the log-likelihood value is maximized to determine the optimal match between the panoramic point cloud data and the BIM model and achieve data alignment.

[0034] As a preferred solution of the present invention, the panoramic point cloud data and the BIM model are aligned, and a point cloud registration algorithm is used, specifically a multi-resolution ICP algorithm combined with a Super4PCS algorithm for global registration, and the formula is:

[0035] ;

[0036] In the formula, represents the error function in the registration process; A is the total number of registered points;

[0037] By adjusting the rotation matrix and translation vectors To minimize the error function value, the panoramic point cloud data and the BIM model can be aligned.

[0038] As a preferred solution of the present invention, extracting key features of power grid facilities from the aligned point cloud data specifically includes:

[0039] Edge feature extraction: Extract edge features from point cloud data that has been transformed and aligned to the BIM model coordinate system. Identify edge points and generate contour lines by calculating the curvature value of each point. The formula is:

[0040] ;

[0041] Where: is the transformed point cloud coordinate The curvature value of For point The bth point in the neighborhood; is the center point in the neighborhood; B is the total number of neighborhood points;

[0042] Corner point feature extraction: By analyzing the local geometric shape of each point, determine whether the point is a corner point. The shape index formula of the corner point is:

[0043] ;

[0044] Where: , , are the curvature eigenvalues ​​of the local area of ​​the point cloud respectively; is the shape index, which indicates the strength of local geometric features;

[0045] Geometric parameter extraction: Based on the aligned point cloud data, the geometric dimensions of the power grid facilities are calculated, including height, width, length and inclination angle of the components;

[0046] The geometric parameter extraction formula is:

[0047] ;

[0048] Where: D represents the Euclidean distance between two points; and are the three-dimensional coordinates of two points;

[0049] Automatically classify and annotate the extracted edge features, corner features, and geometric parameter features, identify feature points through a machine learning algorithm based on geometric morphology, classify feature points as key structural features, and automatically match them with the key structures of the BIM model;

[0050] Use color or intensity information in high-definition image data to enhance the annotation accuracy of feature points and ensure that components of different materials or types can be distinguished.

[0051] As a preferred solution of the present invention, the method of adjusting the vertices and edges of the BIM model based on the extracted key features so that the BIM model matches the point cloud data includes:

[0052] An automatic correction algorithm based on rules, including edge smoothness, vertex movement range, and curvature matching, automatically adjusts the vertices and edges of the BIM model according to the extracted key features, reduces geometric deviations, and ensures that the BIM model is aligned with the point cloud data;

[0053] Manual fine-tuning combined with user input to customize key structural parts;

[0054] The system provides real-time feedback to display the alignment between the BIM model and the point cloud data, ensuring that the adjustment results meet the expected requirements.

[0055] As a preferred solution of the present invention, the method for ensuring the accuracy and reliability of the BIM model through consistency verification includes:

[0056] Calculate the geometric error W between the BIM model and the point cloud data using the formula:

[0057] ;

[0058] In the formula, is the depth map error, is the normal map error, is the height error, , and are the weights of the error terms respectively;

[0059] Verify the accuracy of the BIM model by calculating geometric errors.

[0060] A three-dimensional model fitting system for power grid facilities based on panoramic point cloud, based on the above-mentioned three-dimensional model fitting method for power grid facilities based on panoramic point cloud, comprises:

[0061] The data acquisition module is used to collect panoramic cloud data and high-definition image data of power grid facilities through the laser radar and high-definition camera carried by the drone;

[0062] A data preprocessing module, used to preprocess the panoramic point cloud data, remove noise and redundant data, and balance the point cloud density using a filtering algorithm;

[0063] BIM model import module, used to import BIM models of power grid facilities;

[0064] A coordinate conversion module is used to establish a homologous coordinate system between the BIM model and the panoramic point cloud data, and to align the panoramic point cloud data and the BIM model;

[0065] A feature extraction module is used to extract key features of power grid facilities from the aligned point cloud data, including edge features, corner features and geometric parameter features;

[0066] The model correction module is used to adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data;

[0067] Model verification module, used to ensure the accuracy and reliability of BIM models through consistency verification;

[0068] The 3D model generation module is used to generate a 3D model consistent with the actual power grid facilities based on the revised BIM model for real-time monitoring and defect detection.

[0069] The beneficial effects of the present invention are as follows: by accurately fitting the panoramic point cloud data collected by the drone and the BIM model, the geometric error between the point cloud data and the BIM model can be effectively reduced, thereby ensuring the accuracy and consistency of the three-dimensional model; through the consistency verification method, multiple geometric error indicators such as the depth map, normal map and height error are used to significantly improve the accuracy of the model correction, thereby ensuring that the generated BIM model is highly consistent with the actual facilities; in addition, the automated data alignment and feature extraction method of the present invention can quickly and accurately extract the edge features, corner features and geometric parameter features of the power grid facilities, thereby reducing human intervention and improving the automation level of the system; it can not only improve the monitoring and maintenance efficiency of the power grid facilities, but also reduce the operation and maintenance risks caused by data deviations, thereby achieving more efficient fault detection and decision support. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0071] Figure 1 is a flow chart of the method of the present invention;

[0072] Figure 2 It is a system modular structure diagram of the present invention. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0074] like Figure 1 As shown in FIG. 1 , an embodiment of the present invention provides a method for fitting a three-dimensional model of a power grid facility based on a panoramic point cloud, comprising the following steps:

[0075] S1: Collect panoramic cloud data and high-definition image data of power grid facilities through the lidar and high-definition camera carried by the drone;

[0076] When collecting data, it is necessary to determine the appropriate flight path and altitude to cover power grid facilities (such as transmission lines, towers, substations, etc.) and surrounding areas; data from multiple perspectives and time points should be ensured during the collection process to facilitate subsequent processing. Panoramic point cloud data is generated by laser radar (LiDAR) technology and contains a collection of points in three-dimensional space. Each point records information such as distance, angle, and reflection intensity, reflecting the geometric shape, spatial position, and structure of the object, but it usually does not contain color or texture information; high-definition image data is an image or video taken by a high-definition camera, which mainly contains two-dimensional visual information of the scene, such as color, texture, light and shadow effects, etc., which is used to assist in the calibration of point cloud data, provide surface detail information, and help extract and verify the features of the model.

[0077] S2: Preprocess the panoramic point cloud data to remove noise and redundant data, and use filtering algorithms to balance the point cloud density to retain valid point cloud data of power grid facilities and ensure data quality;

[0078] In one embodiment, the panoramic point cloud data is preprocessed, including filtering the geometric line segments in the panoramic point cloud data, and the formula is:

[0079] ;

[0080] Where E represents the energy function, which is used to measure the matching degree between the line segments selected in the panoramic point cloud data and the geometric information of their supporting points. A larger E value indicates that the geometric information is more consistent with expectations; N represents the number of all candidate line segments; represents the number of supporting points of the i-th line segment, that is, the number of points that can be screened out by the distance threshold in the panoramic point cloud data; It is the density of points within a predetermined unit length, which is used to measure the distribution density of points on a line segment and defines the minimum point density that needs to be achieved on the line segment; is the length of the i-th line segment. Longer line segments require more support points to satisfy geometric feature matching. is the selection status of the i-th line segment, Indicates that the line segment is selected. Indicates that the line segment is not selected.

[0081] The formula calculates whether the density of points on each line segment meets the expected standard (given by OK) to determine whether the line segment should be selected. ≥1, it means that the number of supporting points of the line segment is sufficient and the line segment can be considered to conform to the geometric information; if it is insufficient, that is, <1, then take 0 to avoid selecting line segments with too few supporting points.

[0082] S3: Import the BIM model (Building Information Modeling) of the power grid facilities and establish a homologous coordinate system with the panoramic cloud data to ensure that the coordinate systems of different data sources can match each other;

[0083] The BIM model includes the material properties and geometric structure information of the power grid facilities, and identifies and records information in the BIM model such as the coordinates of the tower base, tower height, and the positions of each vertex for subsequent correction processing; it applies affine transformation, rigid transformation, or nonlinear transformation comprehensive algorithms to enhance the accuracy of coordinate conversion and improve data consistency.

[0084] In one specific embodiment, step S3 includes:

[0085] S31: Import the BIM model of the power grid facilities and define the vertex set M as ,in, represents the three-dimensional coordinates of the j-th vertex in the BIM model, is the total number of vertices;

[0086] S32: Process the preprocessed panoramic point cloud data P to extract geometric features corresponding to the BIM model; ,in, represents the coordinates of the jth point in the panoramic point cloud data, is the number of points in the panoramic point cloud data;

[0087] S33: Construct a global Gaussian mixture model to describe the geometric distribution of the panoramic cloud data and BIM model. The formula is:

[0088] ;

[0089] In the formula, is the mean vector of the kth global Gaussian mixture model component, including the geometric center of the panoramic point cloud data and the BIM model in a local area; is the mean of the global point cloud data, indicating the center point of the local area in the point cloud; is the mean value of the BIM model, indicating the center point of the geometric structure in the BIM model; is the covariance matrix of the kth global Gaussian mixture model component, reflecting the geometric distribution of the panoramic point cloud data and the BIM model in the local area; is the covariance matrix of the panoramic point cloud data, which represents the geometric distribution of the point cloud in the area; is the covariance matrix of the BIM model, which indicates the geometric distribution of the BIM model in the area; is the covariance matrix between the panoramic point cloud data and the BIM model, indicating the geometric correlation between the two; Represents the transpose operation of a matrix or vector;

[0090] The global Gaussian mixture model is used to match geometric features. The geometric consistency of the panoramic point cloud data and the BIM model in the local area is compared to determine the geometric consistency of the two. The global Gaussian mixture model provides information support for the alignment of the coordinate system by describing the geometric consistency or inconsistency between the point cloud and the BIM model.

[0091] S34: Apply an affine or rigid transformation by calculating a rotation matrix and translation vectors , complete the coordinate conversion, transform the panoramic point cloud data into the coordinate system of the BIM model, the formula is:

[0092] ;

[0093] In the formula, Represents the transformed point cloud coordinates;

[0094] Rotation Matrix Indicates the adjustment of the panoramic point cloud data in direction, translation vector Indicates the displacement adjustment of the panoramic point cloud data, which is used to align the center of the point cloud with the center of the BIM model;

[0095] S35: In some complex geometric cases, nonlinear transformations (such as thin plate spline interpolation, TPS) can be used to handle more complex deformations, expressed as:

[0096] ;

[0097] In the formula, Represents a nonlinear transformation function, which is used to describe complex geometric deformations.

[0098] S4: Align the panoramic point cloud data with the 3D BIM model, and extract key features of the power grid facilities from the aligned point cloud data, including geometric features, visual features, etc.;

[0099] In one embodiment, the panoramic point cloud data and the BIM model are aligned, including using the log-likelihood method to align the data, specifically:

[0100] Based on the constructed global Gaussian mixture model, the log-likelihood method is used to calculate the degree of alignment between each point in the panoramic point cloud data and the BIM model. The calculation formula is:

[0101] ;

[0102] In the formula, Indicates the midpoint of the panoramic point cloud data The log-likelihood of the point cloud coordinates (pre-processed without rotation and translation transformation) and the BIM model is used to determine the matching degree between the BIM model and the point cloud; For point The probability of belonging to any part of the BIM model; is the weight of the kth global Gaussian mixture model component, indicating the influence of this component on the overall model; K is the total number of Gaussian components; For the kth Gaussian distribution, point The probability density of and the covariance matrix Sure;

[0103] The log-likelihood of all panoramic point cloud data is summed and the log-likelihood value is maximized to determine the optimal match between the panoramic point cloud data and the BIM model and achieve data alignment.

[0104] In another embodiment, the panoramic point cloud data and the BIM model are aligned, and a point cloud registration algorithm is used, specifically a multi-resolution ICP (Iterative Closest Point) algorithm combined with a Super4PCS (Super Four-Points Congruent Sets) algorithm for global registration, and the formula is:

[0105] ;

[0106] In the formula, represents the error function in the registration process; A is the total number of registered points;

[0107] By adjusting the rotation matrix and translation vectors To minimize the error function value, the panoramic point cloud data and the BIM model can be aligned.

[0108] Furthermore, the key features of the power grid facilities are extracted from the aligned point cloud data, including:

[0109] Edge feature extraction: Extract edge features from point cloud data that has been transformed and aligned to the BIM model coordinate system. Identify edge points and generate contour lines by calculating the curvature value of each point. The formula is:

[0110] ;

[0111] Where: is the transformed point cloud coordinate The curvature value of For point The bth point in the neighborhood; is the center point in the neighborhood; B is the total number of neighborhood points;

[0112] Corner point feature extraction: By analyzing the local geometric shape of each point, determine whether the point is a corner point. The shape index formula of the corner point is:

[0113] ;

[0114] Where: , , are the curvature eigenvalues ​​of the local area of ​​the point cloud respectively; is the shape index, which indicates the strength of local geometric features;

[0115] Geometric parameter extraction: Based on the aligned point cloud data, the geometric dimensions of the power grid facilities are calculated, including height, width, length and inclination angle of the components;

[0116] The geometric parameter extraction formula is:

[0117] ;

[0118] Where: D represents the Euclidean distance between two points; and are the three-dimensional coordinates of two points;

[0119] Automatically classify and annotate the extracted edge features, corner features, dimensions and geometric parameter features, identify the feature points through a machine learning algorithm based on geometric morphology, classify the feature points as key structural features (such as pillars, beams, joints, etc.), and automatically match them with the key structures of the BIM model;

[0120] Use the color or intensity information in high-definition image data to further enhance the annotation accuracy of feature points and ensure accurate distinction between components of different materials or different types.

[0121] S5: Adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data, and ensure the accuracy and reliability of the BIM model through consistency verification;

[0122] Specifically, step S5 includes:

[0123] An automatic correction algorithm based on rules, including edge smoothness, vertex movement range, and curvature matching, automatically adjusts the vertices and edges of the BIM model according to the extracted key features, reduces geometric deviations, and ensures that the BIM model is aligned with the point cloud data;

[0124] Manual fine-tuning combined with user input to customize key structural parts;

[0125] The system provides real-time feedback to display the alignment between the BIM model and the point cloud data, ensuring that the adjustment results meet the expected requirements.

[0126] Calculate the geometric error W between the BIM model and the point cloud data using the formula:

[0127] ;

[0128] In the formula, is the depth map error, is the normal map error, is the height error, , and are the weights of the error terms respectively;

[0129] Verify the accuracy of the BIM model by calculating geometric errors.

[0130] S6: Generate a three-dimensional model consistent with the actual power grid facilities based on the revised BIM model for real-time monitoring and defect detection.

[0131] like Figure 2 As shown, another embodiment of the present invention provides a three-dimensional model fitting system for power grid facilities based on panoramic point cloud, based on the above-mentioned three-dimensional model fitting method for power grid facilities based on panoramic point cloud, including:

[0132] The data acquisition module is used to collect panoramic cloud data and high-definition image data of power grid facilities through the laser radar and high-definition camera carried by the drone;

[0133] The data preprocessing module is used to preprocess the panoramic point cloud data, remove noise and redundant data, and use filtering algorithms to balance the point cloud density;

[0134] BIM model import module, used to import BIM models of power grid facilities;

[0135] Coordinate conversion module, used to establish a homologous coordinate system between the BIM model and the panoramic point cloud data, and to align the panoramic point cloud data and the BIM model;

[0136] A feature extraction module is used to extract key features of power grid facilities from the aligned point cloud data, including edge features, corner features and geometric parameter features;

[0137] The model correction module is used to adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data;

[0138] Model verification module, used to ensure the accuracy and reliability of BIM models through consistency verification;

[0139] The 3D model generation module is used to generate a 3D model consistent with the actual power grid facilities based on the revised BIM model for real-time monitoring and defect detection.

[0140] In summary, the present invention can effectively reduce the geometric error between point cloud data and BIM model and ensure the accuracy and consistency of three-dimensional model by accurately fitting the panoramic point cloud data collected by drone and BIM model; through the consistency verification method, multiple geometric error indicators such as depth map, normal map and height error are used to significantly improve the accuracy of model correction, ensuring that the generated BIM model is highly consistent with the actual facilities; in addition, the automated data alignment and feature extraction method of the present invention can quickly and accurately extract the edge features, corner features and geometric parameter features of power grid facilities, reduce human intervention and improve the automation level of the system; it can not only improve the monitoring and maintenance efficiency of power grid facilities, but also reduce the operation and maintenance risks caused by data deviation, and achieve more efficient fault detection and decision support.

[0141] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A three-dimensional model fitting method for power grid facilities based on panoramic point cloud, characterized in that: The method comprises: The drone-mounted laser radar and high-definition camera collects panoramic cloud data and high-definition image data of power grid facilities; Preprocessing the panoramic point cloud data to remove noise and redundant data, and using a filtering algorithm to balance the point cloud density; Import the BIM model of the power grid facilities and establish a homologous coordinate system with the panoramic cloud data to ensure that the coordinate systems of different data sources can match each other; Aligning the panoramic point cloud data with the BIM model, and extracting key features of the power grid facilities from the aligned point cloud data; The key features of the power grid facilities are extracted from the aligned point cloud data, specifically including: Edge feature extraction: Extract edge features from point cloud data that has been transformed and aligned to the BIM model coordinate system. Identify edge points and generate contour lines by calculating the curvature value of each point. The formula is: ； Where: is the transformed point cloud coordinate The curvature value of For point The bth point in the neighborhood; is the center point in the neighborhood; B is the total number of neighborhood points; Corner point feature extraction: By analyzing the local geometric shape of each point, determine whether the point is a corner point. The shape index formula of the corner point is: ； Where: , , are the curvature eigenvalues ​​of the local area of ​​the point cloud respectively; is the shape index, which indicates the strength of local geometric features; Geometric parameter feature extraction: Based on the aligned point cloud data, the geometric dimensions of the power grid facilities are calculated, including height, width, length and inclination angle of the components; The geometric parameter extraction formula is: ； Where: D represents the Euclidean distance between two points; and are the three-dimensional coordinates of two points; Automatically classify and annotate the extracted edge features, corner features, and geometric parameter features, identify feature points through a machine learning algorithm based on geometric morphology, classify feature points as key features, and automatically match them with the key structures of the BIM model; Use the color or intensity information in high-definition image data to enhance the annotation accuracy of feature points and ensure that components of different materials or types can be distinguished; Adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data, and ensure the accuracy and reliability of the BIM model through consistency verification; Based on the revised BIM model, a three-dimensional model consistent with the actual power grid facilities is generated for real-time monitoring and defect detection.

2. A three-dimensional model fitting method for power grid facilities based on panoramic point cloud according to claim 1, characterized in that: The panoramic point cloud data is preprocessed, including filtering the geometric line segments in the panoramic point cloud data. The formula is: ； Where E represents the energy function; N represents the number of all candidate line segments; represents the number of supporting points of the i-th line segment; is the density of points within a predetermined unit length; is the length of the i-th line segment; is the selection status of the i-th line segment, Indicates that the line segment is selected. Indicates that the line segment is not selected.

3. The method for fitting a three-dimensional model of power grid facilities based on panoramic point cloud according to claim 1, characterized in that: The method of importing the BIM model of the power grid facilities and establishing a homologous coordinate system with the panoramic point cloud data includes: Import the BIM model of the power grid facilities and define the vertex set M as ,in, represents the three-dimensional coordinates of the j-th vertex in the BIM model, is the total number of vertices; Process the preprocessed panoramic point cloud data P to extract geometric features corresponding to the BIM model; ,in, represents the coordinates of the jth point in the panoramic point cloud data, is the number of points in the panoramic point cloud data; A global Gaussian mixture model is constructed to describe the geometric distribution of the panoramic cloud data and the BIM model. The formula is: ； In the formula, is the mean vector of the kth global Gaussian mixture model component; is the mean value of the panoramic cloud data; is the mean value of the BIM model; is the covariance matrix of the kth global Gaussian mixture model component; is the covariance matrix of the panoramic cloud data; is the covariance matrix of the BIM model; is the covariance matrix between the panoramic point cloud data and the BIM model; Represents the transpose operation of a matrix or vector; The global Gaussian mixture model is used to match geometric features, and the geometric consistency of the panoramic point cloud data and the BIM model is determined by comparing the mean and covariance in the local area. Apply an affine or rigid transformation by computing a rotation matrix and translation vectors , complete the coordinate conversion, transform the panoramic point cloud data into the coordinate system of the BIM model, the formula is: ； In the formula, Represents the transformed point cloud coordinates; Thin plate spline interpolation TPS is used to process complex deformation, which can be expressed as: ； In the formula, Represents a nonlinear transformation function.

4. A method for fitting a three-dimensional model of power grid facilities based on panoramic point cloud according to claim 3, characterized in that: The panoramic point cloud data and the BIM model are aligned, including using the log-likelihood method to align the data, specifically: Based on the constructed global Gaussian mixture model, the log-likelihood method is used to calculate the degree of alignment between each point in the panoramic point cloud data and the BIM model. The calculation formula is: ； In the formula, Indicates the midpoint of the panoramic point cloud data Log-likelihood with the BIM model; For point The probability of belonging to any part of the BIM model; is the weight of the kth global Gaussian mixture model component; K is the total number of Gaussian components; is the kth Gaussian distribution lower point The probability density of and the covariance matrix Sure; The log-likelihood of all panoramic point cloud data is summed and the log-likelihood value is maximized to determine the optimal match between the panoramic point cloud data and the BIM model and achieve data alignment.

5. The method for fitting a three-dimensional model of power grid facilities based on panoramic point cloud according to claim 3, characterized in that: The panoramic point cloud data and the BIM model are aligned, and a point cloud registration algorithm is used, specifically a multi-resolution ICP algorithm combined with a Super4PCS algorithm for global registration. The formula is: ； In the formula, represents the error function in the registration process; A is the total number of registered points; By adjusting the rotation matrix and translation vectors To minimize the error function value, the panoramic point cloud data and the BIM model can be aligned.

6. The method for fitting a three-dimensional model of power grid facilities based on panoramic point cloud according to claim 1, characterized in that: The method of adjusting the vertices and edges of the BIM model based on the extracted key features so that the BIM model matches the point cloud data includes: An automatic correction algorithm based on rules, including edge smoothness, vertex movement range, and curvature matching, automatically adjusts the vertices and edges of the BIM model based on the extracted key features, reduces geometric deviations, and ensures that the BIM model is aligned with the point cloud data; Manual fine-tuning combined with user input to customize key structural parts; The system provides real-time feedback to display the alignment between the BIM model and the point cloud data, ensuring that the adjustment results meet the expected requirements.

7. The method for fitting a three-dimensional model of power grid facilities based on panoramic point cloud according to claim 1, characterized in that: The method of ensuring the accuracy and reliability of the BIM model through consistency verification includes: Calculate the geometric error W between the BIM model and the point cloud data using the formula: ； In the formula, is the depth map error, is the normal map error, is the height error, , and are the weights of the error terms, respectively; Verify the accuracy of the BIM model by calculating geometric errors.

8. A three-dimensional model fitting system based on a three-dimensional model fitting method of power grid facilities based on panoramic point cloud according to any one of claims 1 to 7, characterized in that: The system comprises: The data acquisition module is used to collect panoramic cloud data and high-definition image data of power grid facilities through the laser radar and high-definition camera carried by the drone; A data preprocessing module, used to preprocess the panoramic point cloud data, remove noise and redundant data, and balance the point cloud density using a filtering algorithm; BIM model import module, used to import BIM models of power grid facilities; A coordinate conversion module is used to establish a homologous coordinate system between the BIM model and the panoramic point cloud data, and to align the panoramic point cloud data and the BIM model; A feature extraction module is used to extract key features of power grid facilities from the aligned point cloud data, including edge features, corner features and geometric parameter features; The model correction module is used to adjust the vertices and edges of the BIM model based on the extracted key features to match the BIM model with the point cloud data; Model verification module, used to ensure the accuracy and reliability of BIM models through consistency verification; The 3D model generation module is used to generate a 3D model consistent with the actual power grid facilities based on the revised BIM model for real-time monitoring and defect detection.

Images