Power distribution network point cloud model construction system based on three-dimensional visualization

Abstract

Disclosed in the present invention is a power distribution network point cloud model construction system based on three-dimensional visualization, the system comprising: a data collection module, which collects model data of a power distribution network by means of receiving a signal reflected by a LiDAR scanner; a data denoising module, which performs denoising processing on the collected model data, so as to eliminate a data error caused by noise; a point cloud reconstruction module, which simplifies denoised point cloud data, screening important features in the power distribution network, and constructing a point cloud model using the screened data; a visualization module, which uses a visualization technique to present the constructed model, so as to realize an interactive visualization operation; and an operation state evaluation module, which evaluates the safety performance of the power distribution network on the basis of fault data and a historical operation record of the power distribution network. In this way, the operation condition of the power distribution network can be visually known by means of a three-dimensional visualization interface, and thus the operation and maintenance efficiency is improved, and accurate data support is provided for the management and operation and maintenance of the power distribution network.

Description

A distribution network point cloud model construction system based on 3D visualization Technical Field

[0001] The present invention belongs to the technical field of distribution network model construction, and in particular relates to a distribution network point cloud model construction system based on three-dimensional visualization. Background Art

[0002] With the development of three-dimensional visualization technology, the demand for the construction of distribution network point cloud models is becoming increasingly urgent. The distribution network visualization model is the core information of distribution network operations. The construction of the distribution network model can provide a visualization interface and simulation basis for the daily production scheduling of the distribution network. With the construction of smart grids, the traditional method of constructing distribution network models can no longer meet the detection requirements of distribution network entry production.

[0003] Traditional automatic construction systems mostly only consider two-dimensional plane information and cannot fully express the three-dimensional spatial information of the distribution network. Although existing systems can achieve three-dimensional visualization, the visualization effect is often not ideal. The expression of the model often only stays at the visualization level, lacking the analysis and evaluation function of the distribution network operation status, and cannot provide effective decision-making support for technical personnel.

[0004] Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a distribution network point cloud model construction system based on three-dimensional visualization to solve the three-dimensional visualization problem of the existing technology, but the visualization effect is often not ideal, the expression of the model often only stays at the visualization level, lacks the analysis and evaluation function of the distribution network operation status, and cannot provide effective decision support for technical personnel.

[0006] The technical solution of the present invention is:

[0007] The distribution network point cloud model construction system based on 3D visualization includes:

[0008] Data acquisition module: collects model data of the distribution network by receiving the signal reflected by the lidar scanner;

[0009] Data denoising module: denoises the collected model data to eliminate data errors caused by noise;

[0010] Point cloud reconstruction module: simplifies the denoised point cloud data, filters important features in the distribution network, and constructs a point cloud model based on the filtered data;

[0011] Visualization module: Use visualization technology to present the constructed model and realize interactive visualization operation;

[0012] Operation status assessment module: Evaluate the safety performance of the distribution network based on its fault data and historical operation records.

[0013] In the above-mentioned distribution network point cloud model construction system based on three-dimensional visualization, the data acquisition module scans the distribution network data through a lidar scanner deployed in the target area of ​​the distribution network, starts the lidar scanner, receives the signal reflected back by the lidar scanner, collects the texture and material information of the object surface, and sends the collected data information to the data denoising module.

[0014] In the above-mentioned distribution network point cloud model construction system based on 3D visualization, the data denoising module analyzes the spatial distribution of discrete points. The discrete point cloud is far away from the subject point cloud, while the subject point cloud has a high density and is relatively concentrated. The obtained point cloud data set with noise points is represented as D = {D i ,i=1,2,...,n},D i Represents any point cloud in the data set. When the average distance from all points in the point cloud data set to the remaining sub-neighborhoods obeys Gaussian distribution, that is, P i It obeys Gaussian distribution and the shape of the distribution is determined by the mean and standard deviation. The specific calculation formula of the mean and standard deviation is:

[0015] Among them, Q z represents the mean, P i It represents the average distance to k points in the field. The maximum distance threshold for data denoising can be expressed as: max =d′+n*B z

[0016] Among them, d' represents the average distance from the point cloud to the k points in the neighborhood, n represents the standard deviation coefficient, and when the distance exceeds d max When the distance is less than d max It is determined to be a non-noise point. Specifically, data denoising includes the following steps:

[0017] S1. Input point cloud data D with noise points. For any point cloud, find its k nearest points through nearest neighbor search and calculate the average distance from it to each point in its neighborhood.

[0018] S2, calculate the distance between the point cloud and k adjacent points and each point cloud and calculate the average distance, and get the set of average distances {eq1,eq2,eq 3,... eq n};

[0019] S3. Calculate the mean and standard deviation, and obtain the maximum threshold d based on the Gaussian distribution according to the mean and standard deviation.max =d′+n*B z ;

[0020] S4, compare the mean of the calculated point cloud and its k nearest neighbors with the maximum threshold d max In comparison, the noise point removal is finally completed.

[0021] In the above-mentioned distribution network point cloud model construction system based on 3D visualization, the point cloud reconstruction module reduces the redundancy of point cloud data, extracts important features in the point cloud, simplifies the point cloud, filters the curvature value, and constructs the point cloud model using the filtered point cloud data. The specific method of point cloud simplification is as follows:

[0022] S1. When n point clouds P on a sphere int (x i ,y i ,z i )(i=1,2,3,...,n), its center coordinates can be expressed as ct(x ct ,y ct ,z ct ), the radius can be expressed as: R=(xx ct ) 2 +(yy ct ) 2 +(zz ct ) 2

[0023] Where C represents the spherical fitting error. From the above calculation, the spherical linear equation can be expressed as: 2 +y 2 +z 2 -2(x*x ct +y*y ct +z*z ct )+C=0

[0024] Among them, x 2 、y 2 、z 2 Represents the square of the distance between the coordinates of the point and the origin. By substituting each point in the point cloud data into the equation, if the calculated result is equal to 0, it means that the point is located on the fitted sphere. A local coordinate system is established and P is int As the coordinate far point, the X-axis, Y-axis and Z-axis directions are the same as the absolute coordinate system directions. int The points near the point are translated to obtain their spatial coordinates in the newly created local coordinate system;

[0025] S2. Substitute the coordinates of adjacent points into the spherical linear equation to obtain the parameter x ct 、y ct、z ct The curvature is calculated with C. The specific calculation formula is as follows:

[0026] Among them, C 2 It represents the square of the distance error between the fitted spherical model and the point cloud data. After the curvature is obtained, it is compared with the preset threshold to determine whether it belongs to the point. After all points are judged, the simplification is completed.

[0027] S3. Build the point cloud model. The specific construction steps are as follows:

[0028] 2) Extract feature descriptors from the point cloud and select the neighborhood point set P = {p_1, p_2, p_3, ..., p_k} of point P. For each neighborhood point p_i, calculate its coordinate difference relative to point P. The specific calculation formula for the coordinate difference is: bc = p_i - p

[0029] Construct the normal equation. The specific calculation formula of the normal equation is: A*n=b

[0030] Where A represents a k×3 matrix, each row corresponds to the coordinate difference of a neighborhood point, n represents a 3×1 normal vector, b represents a k×1 zero vector, where k represents the number of neighborhood points. The normal vector n that minimizes the error is obtained by calculation. The specific calculation formula of n is: n=(A^T*A)^-1*A^T*b

[0031] Where A^T*A represents the matrix product of A^T and A, and A^T*b represents the matrix product of A^T and b;

[0032] 2) Match features from different perspectives, establish point correspondences, and build the first feature descriptor f_a and the second feature descriptor f_b. Calculate the distance between the two sets of feature descriptors using the Euclidean distance and select the closest match. The specific calculation formula for the Euclidean distance is as follows: d = |f_a - f_b|

[0033] 4) Based on the registered point cloud data, an implicit fitting function is used to approximate the actual surface near the point cloud data. A continuous surface model is reconstructed based on the implicit fitting function, and the surface model is converted into a voxel representation. The continuous geometric shape is discretized into a voxel grid. Specifically, the specific calculation method for establishing the surface model is as follows:

[0034] Among them, x represents any point, x_i represents the sampling point in the point cloud data, w_i represents the weight, Represents radial basis function, converts implicit function into voxel grid representation, divides the three-dimensional space into a uniform voxel grid, and judges whether each voxel is inside the surface according to the value of implicit function. The specific formula is: S(x) = sign(f(x))

[0035] Where S(x) is the label of point x in the voxel grid, f(x) is the value of the implicit function at point x, when f(x)>0, S(x)=1, indicating that point x is outside the surface, when f(x)<0, S(x)=-1, indicating that point x is inside the surface. A continuous surface model is extracted based on the interior and boundary of the voxel grid.

[0036] In the above-mentioned distribution network point cloud model construction system based on 3D visualization, the visualization module creates a visualization window for displaying the 3D model, loads the constructed 3D model data into the visualization window, parses the texture coordinate information in the 3D model, and sets the position, color and intensity of the model light source. The specific calculation formula for the specific lighting calculation is: Color = hg + mf + jm

[0037] Among them, hg represents the ambient lighting component, mf represents the diffuse reflection component, and jm represents the specular reflection component.

[0038] In the above-mentioned distribution network point cloud model construction system based on 3D visualization, the operation status assessment module collects fault data and historical operation records of the distribution network and synchronizes the collected data into the visual 3D model. The collected data includes information about the time of occurrence, duration, and type of fault. Based on the collected data, reliability indicators are calculated. Reliability indicators include availability, mean time between failures, and mean time to repair. The specific calculation formula for availability is: K = pz / (pz + pf)

[0039] Among them, pz represents the mean time between failures, pf represents the mean time to repair, and the specific calculation method of the mean time between failures is: G=Z / Gc

[0040] Where Z represents the total operating time, Gc represents the number of failures, and the specific calculation formula for the average repair time is: F = ZX / XC

[0041] Where ZX represents the total repair time, and XC represents the number of repairs.

[0042] The beneficial effects of the present invention are:

[0043] The present invention collects model data of the distribution network by receiving signals reflected by a laser radar scanner, denoising the collected model data to eliminate data errors caused by noise, simplifying the denoised point cloud data, screening important features in the distribution network, constructing a point cloud model of the screened data, and presenting the constructed model using visualization technology to achieve interactive visualization operations. Based on the fault data and historical operation records of the distribution network, the safety performance of the distribution network is evaluated. Through the three-dimensional visualization interface, the present invention can intuitively understand the operation of the distribution network, improve operation and maintenance efficiency, and provide an intuitive understanding of the structure and operation of the power grid, providing accurate data support for the management and operation of the distribution network.

[0044] The three-dimensional visualization problem of existing technologies has been solved, but the visualization effect is often not ideal. The expression of the model often only stays at the visualization level. It lacks the analysis and evaluation function of the distribution network operation status and cannot provide effective decision-making support for technical personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG1 is a schematic diagram of the process of constructing a distribution network point cloud model based on three-dimensional visualization proposed by the present invention;

[0046] FIG2 is a flow chart of the structure of the distribution network point cloud model construction system based on three-dimensional visualization proposed by the present invention; DETAILED DESCRIPTION

[0047] 1-2 , a distribution network point cloud model construction system based on 3D visualization includes a data acquisition module, a data denoising module, a point cloud reconstruction module, a visualization module, and an operation status assessment module;

[0048] Data acquisition module: collects model data of the distribution network by receiving the signal reflected by the lidar scanner;

[0049] Data denoising module: denoises the collected model data to eliminate data errors caused by noise;

[0050] Point cloud reconstruction module: simplifies the denoised point cloud data, filters important features in the distribution network, and constructs a point cloud model based on the filtered data;

[0051] Visualization module: Use visualization technology to present the constructed model and realize interactive visualization operation;

[0052] Operation status assessment module: Evaluate the safety performance of the distribution network based on its fault data and historical operation records.

[0053] In this embodiment, what needs to be specifically explained is the data acquisition module. The data acquisition module scans the distribution network data through a laser radar scanner deployed in the target area of ​​the distribution network. By starting the laser radar scanner and receiving the signal reflected back by the laser radar scanner, the texture and material information of the object surface are collected, and the collected data information is sent to the data denoising module.

[0054] In this embodiment, the data denoising module specifically needs to be explained. The data denoising module analyzes the spatial distribution of discrete points. The discrete point cloud is far away from the subject point cloud, while the subject point cloud has a high density and is relatively concentrated. The obtained point cloud dataset with noise points is represented as D = {D i ,i=1,2,...,n},D i Represents any point cloud in the data set, assuming that the average distance from all points in D to the remaining sub-neighborhoods obeys Gaussian distribution, that is, P i It obeys a Gaussian distribution and the shape of the distribution is determined by the mean and standard deviation.

[0055] The specific calculation formulas for the mean and standard deviation are:

[0056] Among them, Q z represents the mean, P i It represents the average distance to k points in the field. The maximum distance threshold for data denoising can be expressed as: max =d′+n*B z

[0057] Among them, d' represents the average distance from the point cloud to the k points in the neighborhood, n represents the standard deviation coefficient, and when the distance exceeds d max When the distance is less than d max It is determined to be a non-noise point. Specifically, data denoising includes the following steps:

[0058] S1. Input point cloud data D with noise points. For any point cloud, find its k nearest points through nearest neighbor search and calculate the average distance from it to each point in its neighborhood.

[0059] S2, calculate the distance between the point cloud and k adjacent points and each point cloud and calculate the average distance, and get the set of average distances {eq1,eq2,eq 3,... eq n};

[0060] S3. Calculate the mean and standard deviation, and obtain the maximum threshold d based on the Gaussian distribution according to the mean and standard deviation. max =d′+n*B z ;

[0061] S4, compare the mean of the calculated point cloud and its k nearest neighbors with the maximum threshold d max In comparison, the noise point removal is finally completed.

[0062] In this embodiment, the point cloud reconstruction module specifically needs to be explained. The point cloud reconstruction module reduces the redundancy of point cloud data, extracts important features from the point cloud, simplifies the point cloud, filters the curvature values, and constructs a point cloud model using the filtered point cloud data. The specific method of point cloud simplification is as follows:

[0063] S1. When n point clouds P on a sphere int (x i ,y i ,z i )(i=1,2,3,...,n), its center coordinates can be expressed as ct(x ct ,y ct ,z ct ), the radius can be expressed as: R=(xx ct ) 2 +(yy ct ) 2 +(zz ct ) 2

[0064] Where C represents the spherical fitting error. From the above calculation, the spherical linear equation can be expressed as: 2 +y 2 +z 2 -2(x*x ct +y*y ct +z*z ct )+C=0

[0065] Among them, x 2 、y 2 、z 2 Represents the square of the distance between the coordinates of the point and the origin. By substituting each point in the point cloud data into the equation, if the calculated result is equal to 0, it means that the point is located on the fitted sphere. A local coordinate system is established and P is int As the coordinate far point, the X-axis, Y-axis and Z-axis directions are the same as the absolute coordinate system directions. int The points near the point are translated to obtain their spatial coordinates in the newly created local coordinate system;

[0066] S2. Substitute the coordinates of adjacent points into the spherical linear equation to obtain the parameter x ct 、y ct 、z ct The curvature is calculated with C. The specific calculation formula is as follows:

[0067] Among them, C 2 It represents the square of the distance error between the fitted spherical model and the point cloud data. After the curvature is obtained, it is compared with the preset threshold to determine whether it belongs to the point. After all points are judged, the simplification is completed.

[0068] S3. Build the point cloud model. The specific construction steps are as follows:

[0069] 3) Extract feature descriptors from the point cloud and select the neighborhood point set P = {p_1, p_2, p_3, ..., p_k} of point P. For each neighborhood point p_i, calculate its coordinate difference relative to point P. The specific calculation formula for the coordinate difference is: bc = p_i - p

[0070] Construct the normal equation. The specific calculation formula of the normal equation is: A*n=b

[0071] Where A represents a k×3 matrix, each row corresponds to the coordinate difference of a neighborhood point, n represents a 3×1 normal vector, b represents a k×1 zero vector, where k represents the number of neighborhood points. The normal vector n that minimizes the error is obtained by calculation. The specific calculation formula of n is: n=(A^T*A)^-1*A^T*b

[0072] Where A^T*A represents the matrix product of A^T and A, and A^T*b represents the matrix product of A^T and b;

[0073] 2) Match features from different perspectives, establish point correspondences, and build the first feature descriptor f_a and the second feature descriptor f_b. Calculate the distance between the two sets of feature descriptors using the Euclidean distance and select the closest match. The specific calculation formula for the Euclidean distance is as follows: d = |f_a - f_b|

[0074] 5) Based on the registered point cloud data, an implicit fitting function is used to approximate the actual surface near the point cloud data. A continuous surface model is reconstructed based on the implicit fitting function, and the surface model is converted into a voxel representation. The continuous geometric shape is discretized into a voxel grid. Specifically, the specific calculation method for establishing the surface model is as follows:

[0075] Among them, x represents any point, x_i represents the sampling point in the point cloud data, w_i represents the weight, Represents radial basis function, converts implicit function into voxel grid representation, divides the three-dimensional space into a uniform voxel grid, and judges whether each voxel is inside the surface according to the value of implicit function. The specific formula is: S(x) = sign(f(x))

[0076] Where S(x) is the label of point x in the voxel grid, f(x) is the value of the implicit function at point x, when f(x)>0, S(x)=1, indicating that point x is outside the surface, when f(x)<0, S(x)=-1, indicating that point x is inside the surface. A continuous surface model is extracted based on the interior and boundary of the voxel grid.

[0077] In this embodiment, the visualization module specifically needs to be explained. The visualization module creates a visualization window for displaying the three-dimensional model, loads the constructed three-dimensional model data into the visualization window, parses the texture coordinate information in the three-dimensional model, and sets the position, color, and intensity of the model light source. The specific calculation formula for the specific lighting calculation is: Color = hg + mf + jm

[0078] Among them, hg represents the ambient lighting component, mf represents the diffuse reflection component, and jm represents the specular reflection component.

[0079] In this embodiment, the operating status evaluation module specifically needs to be explained. The operating status evaluation module collects fault data and historical operating records of the distribution network and synchronizes the collected data into a visual three-dimensional model. The collected data includes the time of occurrence, duration, and type of fault. Based on the collected data, reliability indicators are calculated. Reliability indicators include availability, mean time between failures, and mean time to repair. The specific calculation formula for availability is: K = pz / (pz+pf)

[0080] Among them, pz represents the mean time between failures, pf represents the mean time to repair, and the specific calculation method of the mean time between failures is: G=Z / Gc

[0081] Where Z represents the total operating time, Gc represents the number of failures, and the specific calculation formula for the average repair time is: F = ZX / XC

[0082] Where ZX represents the total repair time, and XC represents the number of repairs.

Claims

1. A distribution network point cloud model construction system based on 3D visualization, Features: The system comprises: Data acquisition module: collects the model data of the distribution network by receiving the signal reflected by the lidar scanner; Data denoising module: denoises the collected model data to eliminate data errors caused by noise; Point cloud reconstruction module: simplifies the denoised point cloud data, screens the important features in the distribution network, and constructs a point cloud model with the screened data; Visualization module: Use visualization technology to present the constructed model and realize interactive visualization operation; Operation status evaluation module: Evaluate the safety performance of the distribution network based on the fault data and historical operation records of the distribution network.

2. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: The data acquisition module scans the distribution network data by deploying a laser radar scanner in the target area of ​​the distribution network, starts the laser radar scanner, receives the signal reflected by the laser radar scanner, collects the texture and material information of the object surface, and sends the collected data information to the data denoising module.

3. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: The data denoising module analyzes the spatial distribution of discrete points and represents the acquired point cloud data set with noise points as D = {D i ,i=1,2,...,n},D i Represents any point cloud in the data set. When the average distance from all points in the point cloud data set to the remaining sub-neighborhoods follows a Gaussian distribution, that is, P i It obeys Gaussian distribution and the shape of the distribution is determined by the mean and standard deviation. The specific calculation formula of the mean and standard deviation is: Among them, Q z represents the mean, P i represents the average distance to k points in the field. The maximum distance threshold for data denoising can be expressed as: d max =d′+n*B z Where d' represents the average distance from the point cloud to the k points in the neighborhood, and n represents the standard deviation coefficient.

4. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 3, Features: Data denoising includes the following steps: S1. Input point cloud data D with noise points. For any point cloud, find its k nearest points through nearest neighbor search and calculate the average distance from each point in its neighborhood. S2, calculate the distance between the point cloud and k adjacent points and each point cloud and calculate the average distance to obtain the set of average distances {eq 1 ,eq 2 ,eq 3 ,...eq n }; S3. Calculate the mean and standard deviation, and obtain the maximum threshold d based on the Gaussian distribution according to the mean and standard deviation. max =d′+n*B z ; S4, compare the mean of the calculated point cloud and its k nearest neighbors with the maximum threshold d max In comparison, the noise point removal is finally completed.

5. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: The point cloud reconstruction module reduces the redundancy of point cloud data, extracts important features in the point cloud, simplifies the point cloud, filters the curvature values, and constructs a point cloud model using the filtered point cloud data.

6. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: Point cloud simplification methods include: S1. When n point clouds P on a sphere int (x i ,y i ,z i )(i=1,2,3,...,n), its center coordinates can be expressed as ct(x ct ,y ct ,z ct ), the radius can be expressed as: R=(x-x ct ) 2 +(y-y ct ) 2 +(z-z ct ) 2 Among them, C represents the spherical fitting error. From the above calculation, the spherical linear equation can be expressed as: x 2 +y 2 +z 2 -2(x*x ct +y*y ct +z*z ct )+C=0 Among them, x 2 ,y 2 、z 2 Represents the square of the distance of the coordinates of the point relative to the origin. By substituting each point in the point cloud data into the equation, if the calculated result is equal to 0, it means that the point is located on the fitted sphere. The local coordinate system is established and P int As the coordinate far point, the directions of the X-axis, Y-axis and Z-axis are the same as those of the absolute coordinate system. int The points near the point are translated to obtain their spatial coordinates in the newly created local coordinate system; S2. Substitute the coordinates of the adjacent points into the spherical linear equation to obtain the parameter x ct ,y ct 、z ct And C, the curvature is calculated, the specific calculation formula is as follows: Among them, C 2 It represents the square of the distance error between the fitted spherical model and the point cloud data. After the curvature is obtained, it is compared with the preset threshold to determine whether it belongs to the point. After all points are judged, the simplification is completed.

7. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: S3. Construct a point cloud model. The construction steps include: 1) Extract feature descriptors from the point cloud and select the neighborhood point set of point P P={p_1,p_2,p_3,...,p_k}, for each neighborhood point p_i, calculate its coordinate difference relative to point P. The specific calculation formula of the coordinate difference is: bc=p_i-p Construct the normal equation. The specific calculation formula of the normal equation is: A*n=b Where A represents a k×3 matrix, each row corresponds to the coordinate difference of a neighborhood point, n represents a 3×1 normal vector, b represents a k×1 zero vector, where k represents the number of neighborhood point sets, and the normal vector n that minimizes the error is obtained by calculation. The specific calculation formula of n is: n=(A^T*A)^-1*A^T*b Among them, A^T*A represents the matrix product of A^T and A, and A^T*b represents the matrix product of A^T and b; 2) Match the features of different perspectives, establish point correspondence, establish the first feature descriptor f_a and the second feature descriptor f_b, calculate the distance between the two sets of feature descriptors through Euclidean distance, and select the closest match. The specific calculation formula of Euclidean distance is as follows: d=|f_a-f_b| 3) According to the registered point cloud data, an implicit fitting function is used to approximate the actual surface near the point cloud data, and a continuous surface model is reconstructed according to the implicit fitting function. The surface model is converted into a voxel representation, and the continuous geometric shape is discretized into a voxel grid. Specifically, the specific calculation method for establishing the surface model is as follows: Among them, x represents any point, x_i represents the sampling point in the point cloud data, w_i represents the weight, Represents radial basis function, converts implicit function into voxel grid representation, divides three-dimensional space into uniform voxel grid, and judges whether each voxel is inside the surface according to the value of implicit function. The specific formula is for: S(x)=sign(f(x)) Among them, S(x) is the label of point x in the voxel grid, f(x) is the value of the implicit function at point x, when f(x)>0, S(x)=1, indicating that point x is outside the surface, when f(x)<0, S(x)=-1, indicating that point x is inside the surface. A continuous surface model is extracted based on the interior and boundary of the voxel grid.

8. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: The visualization module creates a visualization window for displaying the 3D model, loads the constructed 3D model data into the visualization window, parses the texture coordinate information in the 3D model, sets the position, color and intensity of the model light source, and the specific calculation formula for the specific illumination calculation is: Color = hg + mf + jm Among them, hg represents the ambient lighting component, mf represents the diffuse reflection component, and jm represents the specular reflection component.

9. The distribution network point cloud model construction system based on three-dimensional visualization according to claim 1, Features: The operation status evaluation module collects the fault data and historical operation records of the distribution network, and synchronizes the collected data to the visual three-dimensional model. The collected data includes the time of occurrence, duration and type of fault. Based on the collected data, the reliability index is calculated. The reliability index includes availability, mean time between failures and mean time to repair. The specific calculation formula of availability is: K = pz / (pz+pf) Among them, pz represents the mean time between failures, pf represents the mean time to repair, and the specific calculation method of the mean time between failures is: G=Z / Gc Among them, Z represents the total running time, Gc represents the number of failures, and the specific calculation formula of the mean repair time is: F=ZX / XC Among them, ZX represents the total repair time, and XC represents the number of repairs.

Images