A three-dimensional real-time map construction method for power transmission and transformation projects

Abstract

The present application relates to the technical field of engineering safety control, in particular to a three-dimensional real-time map construction method for power transmission and transformation engineering, which sequentially collects substation site clouds, determines initial voxel size according to target volume of drawings, creates voxel grid for thinning, retains any point in voxel space, constructs a three-dimensional map model in sequence according to regions, periodically detects the completeness of model features, if the accuracy is unqualified, determines the reason according to the variance of the difference between target size and drawings, and corrects the thinning mode, thinning parameters and data compression rate; if it is qualified, a three-dimensional real-time map is generated, the target live state, voltage level and safety distance are updated, and a safety operation virtual fence is constructed. The construction method can accurately thin the point cloud data according to the state of the point cloud data and the actual operation demand, thereby efficiently constructing a high-precision three-dimensional map, and providing more accurate and reliable safety protection for nearby live operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engineering safety control, and particularly relates to a three-dimensional real-time map construction method for power transmission and transformation engineering. BACKGROUND

[0002] In recent years, China has accelerated the construction of a new power system, and the total scale of power grid operation has steadily increased, with the level of power grid intelligence continuously improving. According to statistics from the China Electricity Council, since 2021, the capacity growth rate of 220kV and above power transformation equipment has maintained around 5%. As of the end of 2023, the capacity of 220kV and above power transformation equipment in the national power grid reached 5.4 billion kVA. At the same time, people's demand for power supply reliability is becoming increasingly high, and in order to reduce unnecessary power loss, the demand for eliminating power grid defects using near-live working methods is becoming more urgent, and the safety control of working sites is of great significance. With the continuous growth of power grid construction scale, the demand for improving work safety is also growing, and the application market for near-live equipment work safety warning technology will continue to expand.

[0003] Near-live work has always been a key area of infrastructure safety control and a difficult problem that needs to be overcome. In recent years, there have been several incidents of personnel electrocution, mechanical miscontact with operating equipment and operating lines within the system, causing serious damage to life and property. The main reason for these accidents is not a lack of safety vigilance on the part of personnel, but rather data problems that exist in the three-dimensional modeling process. Specifically, when using a three-dimensional laser scanner to obtain point cloud data of the work site for three-dimensional modeling, the collected point cloud data often contains noise points and outliers, and the data distribution is not uniform enough. In order to improve accuracy, local features are repeatedly scanned, resulting in a large amount of data and high data redundancy. Furthermore, current point cloud data thinning methods are relatively simple, and static parameters and methods are usually used for thinning, making it difficult to dynamically adjust according to point cloud data and actual conditions. Even after thinning, it is difficult to find a balance between accuracy and efficiency, which further makes it difficult to meet safety requirements.

[0004] A method for real-time monitoring of the safety distance of a live body in live-line work is disclosed in Chinese patent CN114757992A, which specifically comprises the following steps: obtaining a three-dimensional model of each live device in the work area and the safety distance of each live device; expanding the outer contour of the three-dimensional model of the target device based on the safety distance to obtain an expanded three-dimensional model, and determining the safety area of the target device according to the expanded three-dimensional model; obtaining a video image of the work area and constructing a three-dimensional map of the work area containing the target device based on the video image; inputting the video image into a target detection model to obtain the region of interest of the human body and the target device; obtaining the position information of the human body and the target device in the three-dimensional map of the work area and mapping it to a two-dimensional map for distance calculation; Although this method can overcome the defect that the existing technology cannot accurately monitor the safety distance of the live device in all directions in real time, its detection and implementation are still based on the constructed three-dimensional map, and the patent lacks optimization of the precision of the three-dimensional map, making it still difficult to achieve accurate monitoring of the safety distance of the live body. Therefore, at present, a three-dimensional real-time map construction method for power transmission and transformation projects is urgently needed to solve the problem of single point cloud data processing method and difficulty in targeted adjustment according to actual conditions, so as to provide more accurate and reliable safety protection for live-line work. SUMMARY

[0005] The present application provides a three-dimensional real-time map construction method for power transmission and transformation projects to overcome the problem of insufficient precision of three-dimensional maps caused by the single and fixed point cloud data processing method and the difficulty in targeted adjustment according to actual conditions.

[0006] Therefore, the present application provides a three-dimensional real-time map construction method for power transmission and transformation projects, which comprises,

[0007] Point cloud acquisition: sequentially acquire the point cloud compression return of the substation according to the order of the regions and determine the initial voxel size based on each target volume in the drawing;

[0008] Voxel thinning: create a voxel grid covering the point cloud data based on the initial voxel size and assign corresponding point positions, and reserve any point position in each voxel space based on the voxel space;

[0009] Model construction: sequentially construct a three-dimensional map model according to the order of the acquired regions based on the thinned point cloud;

[0010] The precision check periodically detects whether the three-dimensional map model feature completeness degree judgment is qualified, and when it is determined to be unqualified, determines the unqualified reason based on the variance of the difference between each target size and the corresponding target size in the drawing, and corrects the sparsification mode, sparsification parameters, and data compression parameters based on the unqualified reason, or when it is determined to be qualified, generates a three-dimensional real-time map, updates the target live state, voltage level, and safety distance in the three-dimensional map, and constructs a safety operation virtual fence based on the safety distance.

[0011] The target refers to an object constructed in the three-dimensional map model.

[0012] Further, the process of determining whether the three-dimensional map model precision is qualified based on the three-dimensional map model feature completeness degree judgment includes:

[0013] Determine the feature completeness rate based on the ratio of the number of correctly expressed features in the three-dimensional map model to the total number of features in the point cloud.

[0014] If the feature completeness rate is less than the set preset feature completeness rate, it is determined that the current precision is unqualified, and the reason for the unqualified precision is determined based on the variance of the absolute value of the difference between each target size and the target size in the drawing.

[0015] If the feature completeness rate is greater than or equal to the preset feature completeness rate, it is determined that the current precision is qualified, and the three-dimensional real-time map is generated.

[0016] Further, the process of determining the unqualified precision reason based on the variance of the absolute value of the difference between each target size and the target size in the drawing includes:

[0017] Determine the size difference absolute value variance based on the variance of the absolute value of the difference between the target size in the three-dimensional map model and the target size in the corresponding drawing.

[0018] If the size difference absolute value variance is less than or equal to the set preset size difference absolute value variance, the reason for unqualification is determined based on the difference between the maximum transmission delay and the minimum delay.

[0019] If the size difference absolute value variance is greater than the preset size difference absolute value variance, it is determined that the node selection in the sparsification process does not meet the requirements, and the voxel size is corrected based on the ratio of the size difference absolute value variance to the preset size difference absolute value variance.

[0020] Further, the process of correcting the voxel size based on the ratio of the size difference absolute value variance to the preset size difference absolute value variance includes:

[0021] Determine the size variance ratio based on the ratio of the size difference absolute value variance to the preset size difference absolute value variance.

[0022] The size variance ratio is compared with a preset size variance ratio, and the voxel size is reduced based on the comparison result, and the reduction range of the voxel size is proportional to the size variance ratio.

[0023] Further, the process of determining the voxel size based on the point distribution uniformity includes:

[0024] The voxel point number variance is determined based on the variance of the number of points in each voxel.

[0025] If the voxel point number variance is less than or equal to a preset voxel point number variance, adjacent points are deleted for each point.

[0026] If the voxel point number variance is greater than the preset voxel point number variance, the voxel space is constructed based on the corrected voxel size, and any point in the voxel space is retained.

[0027] Further, the process of deleting adjacent points for each point includes:

[0028] A random point in the voxel space is selected as a representative point, points within a preset distance from the representative point are deleted, the closest point to the representative point after deletion is set as a new representative point, and the deletion is repeated until all points in the voxel space are traversed.

[0029] Further, a three-dimensional map model is constructed based on the point cloud after deleting adjacent points, and it is determined whether the construction time is greater than a preset construction time, and the process includes:

[0030] The construction time is determined based on the time of constructing a three-dimensional map model based on the point cloud after deleting adjacent points.

[0031] If the construction time is less than or equal to the preset construction time, it is determined whether the feature completeness of the three-dimensional map model is qualified.

[0032] If the construction time is greater than the preset construction time, a construction time ratio is determined based on the ratio of the construction time to the preset construction time, and the preset distance is increased based on the construction time ratio, and the increase range of the preset distance is proportional to the construction time ratio.

[0033] Further, the feature completeness is determined based on the three-dimensional map model constructed after the voxel size is corrected, and the process includes:

[0034] The feature completeness is determined based on the three-dimensional map model constructed after the voxel size is corrected.

[0035] If the feature completeness is less than the preset feature completeness, it is determined that the current accuracy is unqualified, and the thinning method is determined based on the coefficient of variation.

[0036] If the feature integrity rate is greater than or equal to the preset feature integrity rate, it is determined that the current accuracy is qualified, and the three-dimensional real-time map is generated.

[0037] The process of determining the thinning mode based on the coefficient of variation includes:

[0038] The coefficient of variation is determined based on the ratio of the standard deviation of the number of k nearest points to the variance of the points.

[0039] If the coefficient of variation is less than or equal to the preset coefficient of variation, the point reference value is set based on the uniform thinning of the point.

[0040] If the coefficient of variation is greater than the preset coefficient of variation, the unit evaluation value is set based on the uniform thinning of the target.

[0041] The k nearest point number of the point is the number of the k nearest points around each point in the point cloud.

[0042] Further, the process of determining the unqualified reason based on the difference between the maximum transmission delay and the minimum delay includes:

[0043] The absolute jitter is determined based on the difference between the maximum transmission delay and the minimum delay.

[0044] If the absolute jitter is less than or equal to the preset absolute jitter, an feature extraction failure warning is output.

[0045] If the absolute jitter is greater than the preset absolute jitter, it is determined that the network is not good, and the data compression rate is corrected based on the difference between the absolute jitter and the preset absolute jitter.

[0046] Further, the process of correcting the data compression rate based on the difference between the absolute jitter and the preset absolute jitter includes:

[0047] The jitter difference value is determined based on the difference between the absolute jitter and the preset absolute jitter.

[0048] The jitter difference value is compared with the preset jitter difference value, and the data compression rate is increased based on the comparison result, and the increase amplitude of the data compression rate is proportional to the jitter difference value.

[0049] Compared with the prior art, the beneficial effects of the present application are that, through the complete process of point cloud collection, voxel thinning, model construction and precision checking, efficient construction of the three-dimensional real-time map of the power transmission and transformation project is realized. The point cloud collection determines the initial voxel size according to the drawings, ensuring the pertinence of data collection; the voxel thinning effectively reduces data redundancy and improves modeling efficiency; the precision checking link can timely discover and correct precision problems by periodically detecting the completeness of model features, and the finally generated three-dimensional real-time map can update the live state, voltage level and safety distance of the target in real time, and construct a safety operation virtual fence, providing accurate and reliable map support for the safety operation of the power transmission and transformation project, effectively improving the safety guarantee level of the operation site and reducing the risk of accidents.

[0050] Further, compared with the traditional size error judgment method, the feature completeness rate can comprehensively evaluate the overall quality of the model, including the existence and correctness of all necessary features, rather than just the local size. This judgment method is more in line with the actual application requirements and can more accurately reflect the accuracy of the three-dimensional map model, providing a more reliable basis for the near live operation and avoiding safety hazards caused by the lack of overall features although the local size error is qualified.

[0051] Further, by analyzing the size difference absolute value variance to determine the reason for the unqualified precision, it can effectively distinguish whether the unqualified precision is caused by node selection problems in the thinning process or data transmission problems. When the size difference absolute value variance is greater than the preset value, it is clear that it is a thinning problem, and the voxel size is corrected based on the ratio of the size difference absolute value variance to the preset value. This targeted correction method can accurately adjust the voxel size, avoiding the waste of computing resources caused by excessive adjustment, ensuring the improvement of model precision, improving the scientificity and effectiveness of modeling, and further optimizing the precision and reliability of the three-dimensional map model.

[0052] Further, the size variance ratio is compared with the preset size variance ratio, and the voxel size is proportionally reduced based on the comparison result, realizing adaptive adjustment of the voxel size. This method can flexibly adjust the voxel size according to the actual size deviation, ensuring the improvement of model precision, avoiding the problems of insufficient precision or waste of computing resources caused by fixed adjustment amplitude, optimizing the modeling efficiency, ensuring the scientificity and effectiveness of the correction measures, and further improving the precision and reliability of the three-dimensional map model.

[0053] Further, the voxel thinning process is determined according to the variance of the number of voxel points, which can dynamically adapt to different data states. When the variance of the number of voxel points is small, it indicates that the point cloud distribution is relatively uniform, and the adjacent point deletion method is used for thinning to improve efficiency. When the variance of the number of voxel points is large, it indicates that the distribution is uneven, and the thinning is performed based on the corrected voxel size to ensure the uniformity of the data and the model accuracy. This flexible thinning processing method effectively solves the precision problem caused by the single thinning method in the prior art, optimizes the modeling efficiency, and ensures the model accuracy.

[0054] Further, the random representative point extraction and adjacent point deletion method is used for thinning, which has randomness and step-by-step optimization characteristics. This method can effectively reduce redundant data, improve modeling efficiency, and enhance the adaptability and robustness of the algorithm, so that it can better cope with point cloud data of different densities and complexities, ensure efficient and accurate generation of three-dimensional map models in various scenarios, and further improve the flexibility and reliability of modeling.

[0055] Further, by judging whether the construction time is greater than the preset construction time and increasing the preset distance based on the construction time ratio when the construction time is too long, a dynamic balance between precision and efficiency is achieved. When the actual construction time exceeds the preset time, it indicates that the point cloud data is still too large, and increasing the preset distance can further reduce the point cloud data, thereby improving the construction efficiency. The increase in the preset distance is proportional to the construction time ratio, which can be flexibly adjusted according to the severity of the efficiency problem, avoiding excessive adjustment that causes precision loss. This dynamic adjustment strategy can adapt to different application scenarios, optimize modeling efficiency, and ensure efficient and accurate construction of three-dimensional map models.

[0056] Further, the thinning method is determined based on the coefficient of variation, which can dynamically select the appropriate thinning method according to the actual distribution characteristics of the point cloud data. When the coefficient of variation is small, it indicates that the point cloud data is uniformly distributed, and the uniform thinning method based on the point reference value is simple and efficient, suitable for scenarios with low precision requirements. When the coefficient of variation is large, it indicates that the point cloud data is unevenly distributed, and the classification thinning method based on the unit evaluation value is used to better preserve key feature points and avoid detail loss caused by excessive thinning, suitable for scenarios with high precision requirements. This dynamic selection of thinning methods improves modeling accuracy and efficiency and adapts to different application scenarios.

[0057] Furthermore, by analyzing absolute jitter to determine the cause of transmission failure, it is possible to effectively distinguish between accuracy issues caused by poor network conditions or feature extraction failures. When the absolute jitter exceeds a preset value, it is clearly a network problem, and the data compression rate is adjusted based on the difference between the absolute jitter and the preset value. This targeted correction method can optimize the data compression rate in real time according to changes in network conditions, reduce the amount of data transmitted, and lower transmission latency and packet loss rate. This improves the reliability of data transmission and the accuracy of 3D map model construction, effectively solving the modeling accuracy problem caused by unstable transmission in existing technologies, and ensuring the stability and reliability of the system.

[0058] Furthermore, the jitter difference is compared with a preset jitter difference, and the data compression ratio is increased proportionally based on the comparison result, achieving adaptive adjustment of the data compression ratio. This method can flexibly adjust the data compression ratio according to the actual network jitter situation, ensuring the reliability of data transmission while avoiding under-compression or over-compression problems caused by a fixed adjustment range. This further optimizes data transmission efficiency and the accuracy of 3D map model construction, improving the overall performance and stability of the system. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating the steps of a method for constructing a three-dimensional real-time map for power transmission and transformation projects in an embodiment of the present invention.

[0060] Figure 2 This is a logic diagram for determining the cause of accuracy failure in an embodiment of the present invention;

[0061] Figure 3 This is a logic decision diagram for correcting voxel size in an embodiment of the present invention;

[0062] Figure 4 This is a logic diagram for determining the thinning method based on the coefficient of variation in an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0064] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0065] Please see Figure 1 The diagram shows a flowchart of a method for constructing a three-dimensional real-time map for power transmission and transformation projects according to an embodiment of the present invention. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to an embodiment of the present invention includes:

[0066] S1: Point cloud acquisition, collecting substation point cloud data in the order of regions, compressing and transmitting back the data, and determining the initial voxel size based on the volume of each target in the drawing.

[0067] S2: Voxel thinning: Create a voxel grid covering the point cloud data based on the initial voxel size and assign corresponding points. Retain any point in each voxel space based on the voxel space.

[0068] S3: Model building, based on the thinned point cloud, constructs a 3D map model in sequence according to the collected areas;

[0069] S4: Accuracy verification, periodically checking the completeness of the features of the three-dimensional map model to determine whether the accuracy of the constructed three-dimensional map model is qualified, and, when it is determined to be unqualified, determining the reason for the unqualification based on the variance of the difference between each target size and each target size in the corresponding drawing, and correcting the thinning method, thinning parameters, and data compression parameters based on the reason for the unqualification, or, when it is determined to be qualified, generating a three-dimensional real-time map, updating the state in the three-dimensional map, and constructing a safe operation virtual fence based on the safety distance.

[0070] Specifically, the process of determining the initial voxel size based on each target volume in the drawing includes obtaining the sum V of each target volume in the drawing. 总 Set the voxel space volume V 体素 V is the sum of the volumes of all targets 总 One ten-thousandth of the original value; since the voxel is a cube, the initial voxel size is set accordingly.

[0071] Specifically, the voxel thinning process includes:

[0072] Obtain the initial voxel size L, and traverse the point cloud data to obtain the minimum coordinate value (X) of the point cloud in 3D space. min Y min Z min ) and maximum coordinate value (X) max Y max Z max To determine the overall bounding box of the point cloud;

[0073] Calculate the total number of rows, columns, and layers of the voxel mesh based on the initial voxel size L and the minimum and maximum coordinates of the point cloud in 3D space:

[0074]

[0075] Traverse the point cloud data, calculate the voxel index of the point based on its coordinates (x, y, z), and add the point to the corresponding voxel to complete the voxel creation.

[0076] in,

[0077]

[0078] Using voxel coordinates as boundaries, select any point within the voxel range to retain it, and delete all remaining points within the corresponding voxel range. Traverse the voxels to complete voxel thinning.

[0079] Furthermore, the process of determining whether the accuracy of the 3D map model is acceptable based on the completeness of its features includes:

[0080] The feature completeness rate is determined based on the ratio of the number of correctly represented features in the 3D map model to the total number of features in the point cloud.

[0081] If the feature completeness rate is less than the preset feature completeness rate, the current accuracy is determined to be unqualified, and the reason for the unqualified accuracy is determined based on the variance of the absolute value of the difference between each target size and each target size in the drawing.

[0082] If the feature completeness rate is greater than or equal to the preset feature completeness rate, the current accuracy is determined to be qualified, and the three-dimensional real-time map is generated.

[0083] The feature completeness rate refers to the ratio of the number of correctly expressed features in the 3D map model to the total number of features in the point cloud. Compared with directly using size error to judge the accuracy of the 3D map model, the feature completeness rate can comprehensively evaluate the overall quality of the model, including the existence and correctness of all necessary features, rather than just local size. This judgment method helps to better reflect the actual application needs, thereby better judging the accuracy of the 3D map model and providing a reliable basis for near-line work.

[0084] The calculation method for the feature completeness rate is not limited in principle. Technicians can automatically extract features from the 3D map model using tools such as SolidWorks, Siemens NX, FeatureCAM, and FeatureRecognition, or align point cloud data with the 3D map model and calculate the deviation to obtain the feature completeness rate.

[0085] Specifically, the process of determining whether the accuracy of the 3D map model is acceptable based on the completeness of its features includes:

[0086] The feature completeness rate W is determined based on the ratio of the number of correctly expressed features in the 3D map model to the total number of features in the point cloud. The feature completeness rate is then compared with the preset feature completeness rate W1, where the preset feature completeness rate W1 is set to [90%, 99%]. According to industry regulations, in order to meet safety requirements, the completeness rate of 3D map models constructed in the power sector is generally required to reach 90%-99%. Therefore, the preset feature completeness rate W1 is set to be selected from 90%-99%.

[0087] If the feature completeness rate W is less than the preset feature completeness rate W1, the current accuracy is determined to be unqualified, and the reason for the unqualified accuracy is determined based on the variance of the absolute value of the difference between each target size and each target size in the drawing.

[0088] If the feature completeness rate W is greater than or equal to the preset feature completeness rate W1, then the current accuracy is deemed acceptable, and a three-dimensional real-time map is generated.

[0089] Furthermore, the process of determining the cause of accuracy non-compliance based on the variance of the absolute value of the difference between each target dimension and each target dimension in the drawing includes:

[0090] The variance of the absolute value of the size difference is determined based on the variance of the absolute value of the difference between the target size in the 3D map model and the target size in the corresponding drawing.

[0091] If the absolute value variance of the size difference is less than or equal to the preset absolute value variance of the size difference, the reason for non-compliance is determined based on the difference between the maximum transmission delay and the minimum delay.

[0092] If the absolute value variance of the size difference is greater than the preset absolute value variance of the size difference, it is determined that the node selection during the thinning process does not meet the requirements, and the initial voxel size is corrected based on the ratio of the absolute value variance of the size difference to the preset absolute value variance of the size difference.

[0093] The absolute variance of the size difference refers to the variance of the absolute value of the difference between the target size in the 3D map model and the target size in the corresponding drawing. This indicator can quantify the fluctuation of the difference between the target size in the 3D map model and the target size in the drawing. In the 3D modeling process, point cloud data thinning is a key step. If the node selection is unreasonable during the thinning process, it may lead to excessive loss of point cloud data in some areas or uneven distribution, resulting in a large deviation between the size of the 3D map model and the actual size, which in turn leads to an increase in the absolute variance of the size difference. In addition, in this solution, the 3D map model is collected and constructed sequentially according to the region order. During this process, packet loss may occur due to network fluctuations, resulting in a relatively accurate 3D map model size but with feature loss. By calculating the absolute variance of the size difference, the specific reasons for the substandard accuracy of the 3D map model can be further effectively determined, thus providing clear guidance for subsequent optimization and correction measures.

[0094] Please see Figure 2 As shown, this is a logic diagram for determining the cause of accuracy non-compliance in an embodiment of the present invention. The process of determining the cause of accuracy non-compliance based on the variance of the difference between each target dimension and each target dimension in the drawing includes:

[0095] The variance of the absolute value of the size difference is determined based on the variance of the absolute value of the difference between the target size in the 3D map model and the target size in the corresponding drawing. The variance of the absolute value of the size difference is then compared with the preset variance of the absolute value of the size difference, A1, where the preset variance of the absolute value of the size difference is set to [0.01, 0.04], and the unit of the absolute value of the size difference is m.

[0096] If the absolute value variance A of the size difference is less than or equal to the preset absolute value variance A1 of the size difference, then the reason for non-compliance is determined based on the difference between the maximum transmission delay and the minimum delay.

[0097] If the absolute variance A of the size difference is greater than the preset absolute variance A1 of the size difference, then it is determined that the node selection during the thinning process does not meet the requirements, and the voxel size is corrected based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference.

[0098] Furthermore, the process of correcting the voxel size based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference includes:

[0099] The size variance ratio is determined based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference;

[0100] The size variance ratio is compared with the preset size variance ratio, and the voxel size is reduced based on the comparison result. The reduction in voxel size is proportional to the size variance ratio.

[0101] The size variance ratio refers to the ratio of the absolute variance of the size difference to the absolute variance of the preset size difference. The size variance ratio reflects problems existing in the point cloud data thinning process. By making the reduction in voxel size proportional to the size variance ratio, adaptive adjustments to voxel sizes can be achieved. This ensures improved model accuracy while avoiding excessive computational resource waste caused by over-adjustment. This method not only optimizes modeling efficiency but also ensures the scientific validity and effectiveness of correction measures, thereby improving the accuracy and reliability of the 3D map model.

[0102] Please see Figure 3 As shown, it is a logic decision diagram for correcting voxel size in an embodiment of the present invention. The process of correcting the voxel size based on the ratio of the absolute value variance of the size difference to the preset absolute value variance of the size difference includes:

[0103] The size variance ratio B is determined based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference, and the size variance ratio B is compared with the set first preset size variance ratio B1 and second preset size variance ratio B2, wherein the first preset size variance ratio B1 is set to (1, 4) and the second preset size variance ratio B2 is set to [4, 8].

[0104] If the size variance ratio B is less than or equal to the first preset size variance ratio B1, then the voxel size L is corrected using the first voxel size correction threshold α1, and the corrected voxel size L' = L × α1, wherein the first voxel size correction threshold α1 is set to 0.98.

[0105] If the size variance ratio B is greater than the first preset size variance ratio B1 and less than or equal to the second preset size variance ratio B2, then the voxel size L is corrected using the second voxel size correction threshold α2, and the corrected voxel size L' = L × α2, wherein the second voxel size correction threshold α2 is set to 0.95.

[0106] If the size variance ratio B is greater than the second preset size variance ratio B2, then the voxel size L is corrected using the third voxel size correction threshold α3, and the corrected voxel size L' = L × α3, wherein the third voxel size correction threshold α3 is set to 0.91.

[0107] Furthermore, the process of determining the voxel thinning process based on the uniformity of point distribution includes:

[0108] The variance of the number of voxels is determined based on the variance of the number of points within each voxel.

[0109] If the variance of the number of voxels is less than or equal to the preset variance of the number of voxels, then adjacent voxels are deleted for each voxel.

[0110] If the variance of the number of voxels is greater than the preset variance of the number of voxels, then the voxel space is constructed based on the corrected voxel size and any point in each voxel space is retained;

[0111] The voxel variance refers to the variance of the number of points within each voxel. The voxel variance quantifies the distribution of point cloud data in the voxel grid. A smaller voxel variance indicates a more uniform point cloud distribution; conversely, a larger variance indicates an uneven distribution. By using the voxel variance, two different voxel thinning processes can be determined. When the point cloud distribution is uniform, a simple thinning method (deleting adjacent points) can be used to improve efficiency. When the distribution is uneven, a more complex thinning method (thinning based on the corrected voxel size) can be used to ensure data uniformity and model accuracy. This method can dynamically adapt to different data states, optimize modeling efficiency, and ensure model accuracy, effectively solving the accuracy problem caused by the single thinning method in the existing technology.

[0112] Specifically, the process of determining the voxel thinning process based on the uniformity of point distribution includes:

[0113] The variance of the number of voxels is determined based on the variance of the number of points within each voxel, and the variance of the number of voxels is compared with the preset variance of the number of voxels C1. The specific value of the preset variance of the number of voxels C1 is not limited in principle, and can be derived based on the actual situation and the patterns of historical data.

[0114] If the variance of the number of voxels C is less than or equal to the preset variance of the number of voxels C1, then adjacent voxels are deleted for each voxel.

[0115] If the variance C of the number of voxels is greater than the preset variance C1 of the number of voxels, then the voxel space is constructed based on the corrected voxel size and any point in each voxel space is retained.

[0116] Furthermore, the process of deleting adjacent points for each location includes:

[0117] Randomly select any point in the voxel space as a representative point, delete points whose distance from the representative point is less than or equal to a preset distance, find the point closest to the representative point after deletion and set it as a new representative point, repeat the deletion until all points in the voxel space are traversed, wherein the preset distance R is set to one-fifteenth of the voxel size.

[0118] By randomly selecting any point in the voxel space as a representative point and deleting points whose distance from the representative point is less than or equal to a preset distance, redundant data can be effectively reduced and modeling efficiency improved. In addition, the randomness and stepwise optimization characteristics of this method enhance the adaptability and robustness of the algorithm, enabling it to better cope with point cloud data of different densities and complexities, and ensuring that 3D map models can be generated efficiently and accurately in various scenarios.

[0119] Furthermore, a 3D map model is constructed based on the point cloud after deleting adjacent points, and it is determined whether the construction time exceeds the preset construction time. The process includes:

[0120] The construction time is determined based on the time required to build a 3D map model from the point cloud after deleting adjacent points.

[0121] If the construction time is less than or equal to the preset construction time, the parameter will continue to be monitored.

[0122] If the construction time is greater than the preset construction time, a construction time ratio is determined based on the ratio of the construction time to the preset construction time, and the preset distance is increased based on the construction time ratio, wherein the increase in the preset distance is proportional to the construction time ratio.

[0123] Determining whether the construction time meets the requirements aims to find a balance between accuracy and efficiency. If the construction time exceeds the preset time, it indicates that the amount of point cloud data is still too large. The amount of point cloud data can be further reduced by increasing the preset distance, thereby improving construction efficiency. The construction time ratio refers to the ratio of the construction time to the preset construction time. The construction time ratio can quantify modeling efficiency. When the actual construction time exceeds the preset time, it indicates that the modeling efficiency is low. It is necessary to increase the preset distance to reduce the amount of point cloud data and improve efficiency. The increase in the preset distance is proportional to the construction time ratio. This ensures that, while meeting accuracy requirements, adjustments can be made flexibly according to the severity of efficiency issues, avoiding over-adjustment that leads to accuracy loss. This dynamic adjustment strategy can adapt to different application scenarios, optimize modeling efficiency, and ensure that the construction of 3D map models is both efficient and accurate.

[0124] Specifically, the process of constructing a 3D map model based on the point cloud after deleting adjacent points, and determining whether the construction time exceeds the preset construction time, includes:

[0125] The construction time T is determined based on the time required to construct a 3D map model from the point cloud after deleting adjacent points, and the construction time T is compared with the preset construction time T1, where the preset construction time T1∈[0.5,24h].

[0126] If the construction time T is less than or equal to the preset construction time T1, the current construction time is deemed acceptable, and the completeness of the 3D map model features is then determined to be acceptable.

[0127] If the construction time T is greater than the preset construction time T1, then a construction time ratio is determined based on the ratio of the construction time to the preset construction time, and the preset distance is increased based on the construction time ratio, wherein the increase in the preset distance is proportional to the construction time ratio.

[0128] Specifically, the process of increasing the preset distance based on the construction time ratio includes,

[0129] The construction time ratio D is determined based on the ratio of the construction time to the preset construction time, and the construction time ratio D is compared with the set first preset construction time ratio D1 and second preset construction time ratio D2. The first preset construction time ratio D1 is set to [1, 3), and the second preset construction time ratio D2 is set to [3, 5].

[0130] If the construction time ratio D is less than or equal to the first preset construction time ratio D1, then the preset distance R is corrected using the first distance correction threshold β1, and the corrected preset distance R' = R × β1, where the first distance correction threshold β1 is set to 1.05;

[0131] If the construction time ratio D is greater than the first preset construction time ratio D1 and less than or equal to the second preset construction time ratio D2, then the preset distance R is corrected using the second distance correction threshold β2, and the corrected preset distance R' = R × β2, wherein the second distance correction threshold β2 is set to 1.11;

[0132] If the construction time ratio D is greater than the second preset construction time ratio D2, then the preset distance R is corrected using the third distance correction threshold β3, and the corrected preset distance R' = R × β3, where the third distance correction threshold β3 is set to 1.17.

[0133] Furthermore, the feature completeness rate is determined based on the 3D map model constructed after correcting the voxel size. The process includes:

[0134] The feature completeness rate is determined based on the 3D map model constructed after correcting the voxel size;

[0135] If the feature completeness rate is less than the preset feature completeness rate, the current accuracy is determined to be unqualified, and the thinning method is determined based on the coefficient of variation.

[0136] If the feature completeness rate is greater than or equal to the preset feature completeness rate, the current accuracy is determined to be qualified, and the three-dimensional real-time map is generated.

[0137] The process of determining the thinning method based on the aforementioned coefficient of variation includes:

[0138] The coefficient of variation is determined based on the ratio of the standard deviation to the variance of the number of k nearest neighbors of a point;

[0139] If the coefficient of variation is less than or equal to the preset coefficient of variation, then the points are uniformly thinned based on the set point reference values.

[0140] If the coefficient of variation is greater than the preset coefficient of variation, then uniform thinning is performed on the target based on the set unit evaluation value;

[0141] The coefficient of variation is the ratio of the standard deviation to the variance of the number of k nearest neighbors of a point. The number of k nearest neighbors is the number of the k nearest neighbors of each point in the point cloud. The coefficient of variation can effectively reflect the uniformity of the distribution of point cloud data. When the coefficient of variation is small, it indicates that the point cloud data is relatively uniformly distributed, and a uniform thinning method based on point reference values ​​is suitable. This method is simple and efficient and suitable for scenarios where the accuracy requirement is not extremely high. When the coefficient of variation is large, it indicates that the point cloud data is unevenly distributed, and a classification thinning method based on unit evaluation values ​​is suitable. This method can better preserve key feature points and avoid the loss of details due to excessive thinning, and is suitable for scenarios with high accuracy requirements. This method can dynamically select an appropriate thinning method according to the actual distribution characteristics of the point cloud data, thereby improving modeling accuracy and efficiency and adapting to different application scenarios.

[0142] Specifically, the feature completeness rate is determined based on the 3D map model constructed after correcting the voxel size. The process includes:

[0143] The feature completeness rate W' is determined based on the 3D map model constructed after correcting the voxel size, and the feature completeness rate W' is compared with the preset feature completeness rate W1, wherein the preset feature completeness rate W1 is set to [90%, 99%];

[0144] If the feature completeness rate W' is less than the preset feature completeness rate W1, the current accuracy is determined to be unqualified, and the thinning method is determined based on the coefficient of variation.

[0145] If the feature completeness rate W' is greater than or equal to the preset feature completeness rate W1, then the current accuracy is determined to be qualified, and the three-dimensional real-time map is generated.

[0146] Please see Figure 4 As shown, this is a logic decision diagram for determining the thinning method based on the coefficient of variation in an embodiment of the present invention. The process of determining the thinning method based on the coefficient of variation includes:

[0147] The coefficient of variation E is determined based on the ratio of the standard deviation to the variance of the number of nearest neighbors of point k. The coefficient of variation E is then compared with the preset coefficient of variation E1. The specific value of the preset coefficient of variation E1 is not limited in principle and can be derived based on the actual situation and the patterns of historical data.

[0148] If the coefficient of variation E is less than or equal to the preset coefficient of variation E1, then the points are uniformly thinned based on the set point reference values. The point reference values ​​effectively reflect the distribution and quantity of points. The point reference values ​​include the coordinates of reference points and the number of reference points. The existence of the number of reference points helps to divide the thinning space based on the number of points and the target volume, while the existence of the coordinates of reference points helps to provide key points. Thus, while significantly reducing the number of points, the accuracy of modeling is guaranteed. This avoids the problems of excessive data processing complexity and computation during modeling, and effectively improves the accuracy of the 3D map model.

[0149] Specifically, the process of uniformly thinning points based on the set point reference values ​​includes:

[0150] Obtain the sum of the volumes of all targets, V 总 and the number of reference points N 点位 The volume of the space after division

[0151] To ensure uniform division of the target volume, the resulting space is a cube; therefore, the side length of the resulting space is...

[0152] Obtain the minimum coordinate value (X) of the point cloud in 3D space. min Y min Z min ) and maximum coordinate value (X) max Y max Z max );

[0153]

[0154]

[0155] Traverse the point cloud data, calculate the voxel index of each point based on its coordinates (x, y, z), and add the point to the corresponding space.

[0156]

[0157] Traverse all spaces and refer to the coordinates of reference points. When the coordinates of the reference point do not fall within the corresponding space, calculate the average value of all points in the corresponding space and record it as the centroid. Record the centroid as the point and delete all remaining points in the space. When the coordinates of the reference point fall within the corresponding space, use the coordinates of the reference point as the point and delete all remaining points in the space.

[0158] The source of the reference point coordinates is not limited in principle. Technicians can obtain the corresponding coordinates based on historical 3D map data or drawings. The specific value of the number of reference points is not limited in principle. Technicians can assign them independently or set them based on historical data according to their needs.

[0159] If the coefficient of variation E is greater than the preset coefficient of variation E1, then the target is uniformly thinned based on the set unit evaluation value. The unit evaluation value includes the equipment voltage and reference point coordinates. The unit evaluation value can effectively reflect the voltage and location of the energized equipment. Then, the volume of the cubic unit is determined according to the unit evaluation value, and the modeling accuracy requirements of different energized equipment and different locations are determined according to the equipment evaluation value. This ensures that the point cloud data of energized equipment with larger voltage and point reference values ​​with smaller reference values ​​can have more details after 3D modeling, thereby providing a reliable basis for power grid workers.

[0160] Specifically, the process of uniformly thinning the target based on the set cell evaluation values ​​includes:

[0161] Obtain the voltage of the target device and determine the number N of reference points based on the voltage level. 目标 In principle, there is no limit to the number of electrical levels and corresponding reference points. Technicians can set them according to their needs or based on historical data. This will not be elaborated further.

[0162] Obtain the corresponding target volume V 目标 And divide the space, and determine the volume of the target single space after the space is divided.

[0163] To ensure uniform division of the target volume, the resulting space is a cube; therefore, the side length of the resulting space is...

[0164] Obtain the minimum coordinate value (X) of the point cloud in 3D space. min Y min Z min ) and maximum coordinate value (X) max Y max Z max );

[0165]

[0166] Traverse the point cloud data, calculate the voxel index of each point based on its coordinates (x, y, z), and add the point to the corresponding space.

[0167]

[0168] Traverse all spaces and refer to the coordinates of reference points. When the coordinates of the reference point do not fall within the corresponding space, calculate the average value of all points in the corresponding space and record it as the centroid. Record the centroid as the point and delete all remaining points in the space. When the coordinates of the reference point fall within the corresponding space, use the coordinates of the reference point as the point and delete all remaining points in the space.

[0169] Furthermore, the process of determining the cause of non-compliance based on the difference between the maximum transmission delay and the minimum delay includes:

[0170] Absolute jitter is determined based on the difference between the maximum transmission delay and the minimum delay;

[0171] If the absolute jitter is less than or equal to the preset absolute jitter, a feature extraction failure warning will be output.

[0172] If the absolute jitter is greater than the preset absolute jitter, the network is determined to be poor, and the data compression rate is adjusted based on the difference between the absolute jitter and the preset absolute jitter.

[0173] The absolute jitter refers to the difference between the maximum transmission delay and the minimum delay. Absolute jitter can effectively reflect the instability and network conditions during data transmission. If the absolute jitter is greater than the preset absolute jitter, it indicates that the network conditions are poor, which may lead to abnormal data transmission. If the data transmission is normal and the absolute variance of the size difference is less than or equal to the preset absolute variance of the size difference, it indicates that the feature extraction may be insufficient, so a feature extraction failure warning is output.

[0174] Specifically, the process of determining the cause of non-compliance based on the difference between the maximum transmission delay and the minimum delay includes:

[0175] The absolute jitter F is determined based on the difference between the maximum transmission delay and the minimum delay, and then the absolute jitter F is compared with the preset absolute jitter F1, where the preset absolute jitter F1 is set to [100, 200ms].

[0176] If the absolute jitter F is less than or equal to the preset absolute jitter F1, then a feature extraction failure warning is output.

[0177] If the absolute jitter F is greater than the preset absolute jitter F1, the network is determined to be poor, and the data compression rate is adjusted based on the difference between the absolute jitter and the preset absolute jitter.

[0178] Furthermore, the process of correcting the data compression rate based on the difference between the absolute jitter and the preset absolute jitter includes:

[0179] The jitter difference is determined based on the difference between the absolute jitter and the preset absolute jitter;

[0180] The jitter difference is compared with a preset jitter difference, and the data compression ratio is increased based on the comparison result. The increase in the data compression ratio is proportional to the jitter difference.

[0181] The jitter difference is defined as the difference between the absolute jitter and a preset absolute jitter. This jitter difference quantifies the instability of network transmission. Through a proportional adjustment mechanism, it can optimize the data compression rate in real time according to changes in network conditions, reducing the amount of transmitted data, lowering transmission latency and packet loss rate, thereby improving the reliability of data transmission and the accuracy of 3D map model construction. This method effectively solves the modeling accuracy problem caused by unstable transmission in existing technologies, ensuring the stability and reliability of the system.

[0182] Specifically, the process of correcting the data compression rate based on the difference between the absolute jitter and the preset absolute jitter includes:

[0183] The jitter difference G is determined based on the difference between the absolute jitter and the preset absolute jitter, and the jitter difference G is compared with the set first preset jitter difference G1 and second preset jitter difference G2, wherein the first preset jitter difference G1 is set to [10, 100ms), and the second preset jitter difference G2 is set to [100, 200ms].

[0184] If the jitter difference G is less than or equal to the first preset jitter difference G1, then the data compression ratio Z is corrected using the first compression correction threshold μ1, and the corrected data compression ratio Z' = Z × μ1, where the first compression correction threshold μ1 is set to 1.03.

[0185] If the jitter difference G is greater than the first preset jitter difference G1 and less than or equal to the second preset jitter difference G2, then the data compression ratio Z is corrected using the second compression correction threshold μ2. The corrected data compression ratio Z' = Z × μ2, where the second compression correction threshold μ2 is set to 1.07.

[0186] If the jitter difference G is greater than the second preset jitter difference G2, then the data compression ratio Z is corrected using the third compression correction threshold μ3. The corrected data compression ratio Z' = Z × μ3, where the third compression correction threshold μ3 is set to 1.12.

[0187] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for constructing a three-dimensional real-time map for power transmission and transformation projects, characterized in that, include, Point cloud acquisition: Substation site cloud data is collected sequentially according to the region order, compressed and transmitted back, and the initial voxel size is determined based on the volume of each target in the drawing. Voxel thinning involves creating a voxel grid covering the point cloud data based on the initial voxel size and assigning corresponding points, while retaining any point within each voxel space based on the voxel space. Model building: Based on the thinned point cloud, a 3D map model is constructed sequentially according to the collected areas. Accuracy verification involves periodically checking the completeness of the features of the 3D map model to determine whether the accuracy of the completed 3D map model is up to standard. If the model is deemed unqualified, the reason for the unqualification is determined based on the variance of the difference between each target size and the target size in the corresponding drawing. The thinning method, thinning parameters, and data compression parameters are corrected based on the reason for the unqualification. Alternatively, if the model is deemed qualified, a real-time 3D map is generated, the status in the 3D map is updated, and a virtual fence for safe operation is constructed based on the safe distance. The target refers to the object constructed in the three-dimensional map model.

2. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 1, characterized in that, The process of determining whether the accuracy of the 3D map model is acceptable based on the completeness of its features includes: The feature completeness rate is determined based on the ratio of the number of correctly represented features in the 3D map model to the total number of features in the point cloud. If the feature completeness rate is less than the preset feature completeness rate, the current accuracy is determined to be unqualified. The reason for the unqualified accuracy is determined based on the variance of the absolute value of the difference between each target size and each target size in the drawing. If the feature completeness rate is greater than or equal to the preset feature completeness rate, the current accuracy is determined to be qualified, and the three-dimensional real-time map is generated.

3. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 2, characterized in that, The process of determining the reasons for accuracy non-compliance based on the variance of the absolute values ​​of the differences between each target dimension and each target dimension in the drawing includes: The variance of the absolute value of the size difference is determined based on the variance of the absolute value of the difference between the target size in the 3D map model and the target size in the corresponding drawing. If the absolute value variance of the size difference is less than or equal to the preset absolute value variance of the size difference, the reason for non-compliance is determined based on the difference between the maximum transmission delay and the minimum delay. If the absolute variance of the size difference is greater than the preset absolute variance of the size difference, it is determined that the node selection during the thinning process does not meet the requirements, and the initial voxel size is corrected based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference.

4. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 3, characterized in that, The process of correcting the voxel size based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference includes: The size variance ratio is determined based on the ratio of the absolute variance of the size difference to the preset absolute variance of the size difference; The size variance ratio is compared with a preset size variance ratio, and the voxel size is reduced based on the comparison result. The reduction in voxel size is proportional to the size variance ratio.

5. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 4, characterized in that, The process of determining the voxel thinning process based on the uniformity of point distribution includes: The variance of the number of voxels is determined based on the variance of the number of points within each voxel. If the variance of the number of voxels is less than or equal to the preset variance of the number of voxels, then adjacent voxels are deleted for each voxel. If the variance of the number of voxels is greater than the preset variance of the number of voxels, then the voxel space is constructed based on the corrected voxel size and any point in each voxel space is retained.

6. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 5, characterized in that, The process of deleting adjacent points for each point includes: Randomly select any point in the voxel space as a representative point, delete points whose distance from the representative point is less than or equal to a preset distance, find the point closest to the deleted representative point and set it as the new representative point, repeat the deletion until all points in the voxel space have been traversed.

7. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 6, characterized in that, A 3D map model is constructed based on the point cloud after deleting adjacent points. The process involves determining whether the construction time exceeds the preset construction time, including: The construction time is determined based on the time required to build a 3D map model from the point cloud after deleting adjacent points. If the construction time is less than or equal to the preset construction time, then it is determined whether the completeness of the 3D map model features is qualified. If the construction time is greater than the preset construction time, a construction time ratio is determined based on the ratio of the construction time to the preset construction time, and the preset distance is increased based on the construction time ratio, wherein the increase in the preset distance is proportional to the construction time ratio.

8. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 4, characterized in that, The process of determining the feature completeness rate based on the 3D map model constructed after correcting the voxel size includes: The feature completeness rate is determined based on the 3D map model constructed after correcting the voxel size; If the feature completeness rate is less than the preset feature completeness rate, the current accuracy is determined to be unqualified, and the thinning method is determined based on the coefficient of variation. If the feature completeness rate is greater than or equal to the preset feature completeness rate, the current accuracy is determined to be qualified, and the three-dimensional real-time map is generated. The process of determining the thinning method based on the aforementioned coefficient of variation includes: The coefficient of variation is determined based on the ratio of the standard deviation to the variance of the number of k nearest neighbors of a point; If the coefficient of variation is less than or equal to the preset coefficient of variation, then the points are uniformly thinned based on the set point reference values. If the coefficient of variation is greater than the preset coefficient of variation, then uniform thinning is performed on the target based on the set unit evaluation value; Wherein, the number of k nearest neighbors of a point is the number of the k nearest neighbors around each point in the point cloud.

9. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 3, characterized in that, The process of determining the cause of non-compliance based on the difference between the maximum transmission delay and the minimum delay includes: Absolute jitter is determined based on the difference between the maximum transmission delay and the minimum delay; If the absolute jitter is less than or equal to the preset absolute jitter, a feature extraction failure warning will be output. If the absolute jitter is greater than the preset absolute jitter, the network is determined to be poor, and the data compression rate is adjusted based on the difference between the absolute jitter and the preset absolute jitter.

10. The method for constructing a three-dimensional real-time map for power transmission and transformation projects according to claim 9, characterized in that, The process of correcting the data compression rate based on the difference between absolute jitter and preset absolute jitter includes: The jitter difference is determined based on the difference between the absolute jitter and the preset absolute jitter; The jitter difference is compared with a preset jitter difference, and the data compression ratio is increased based on the comparison result. The increase in the data compression ratio is proportional to the jitter difference.

Images