Three-dimensional construction method of insulator appearance based on point cloud scanning
Through the combination of point cloud scanning and clustering algorithm, the problem of insulating unit shape insensitive noise recognition in complex environments is solved, and more accurate three-dimensional modeling of insulators and creepage distance calculation is achieved.
Patent Information
- Application Number
- CN202411026178.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-12-13
AI Technical Summary
The existing three-dimensional construction method of insulator shape is insensitive in noise recognition in complex external environments, resulting in large errors in insulator shape recognition, affecting the accuracy of creepage distance calculation.
Point cloud scanning technology is adopted to divide point cloud data into clusters through clustering algorithms, remove noise points, combine feature data and normalization processing, use K-Means clustering algorithm and preprocessing to remove outliers, and optimize the clustering algorithm Parameters to improve denoising accuracy.
It improves the accuracy of insulator three-dimensional modeling and enhances the accuracy of creepage distance calculation.
Smart Images
Figure CN119131233A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of image recognition, and in particular to a three-dimensional construction method of an insulator shape based on point cloud scanning. Background Art
[0002] Insulators are important supporting devices and insulation protection devices for power systems. In the digital measurement of the creepage distance of insulators, it is first necessary to perform image recognition on the insulators to determine their specific locations. Three-dimensional laser scanning technology, also known as real-scene replication technology, is a technological revolution in the field of surveying and mapping after GNSS technology. It breaks through the traditional single-point measurement method and has the unique advantages of high efficiency and high precision. For the calculation of the creepage distance of insulators, the following steps are generally taken: use a three-dimensional scanner to scan the insulator as a whole; transfer the coordinate point cloud data to the computer for transformation and storage; standard three-dimensional software to realize point cloud presentation and formatted file output; system software to realize contour curve extraction and calculate the shortest creepage distance, but the existing three-dimensional construction method of the insulator shape is not sensitive enough to the recognition of noise. In a complex external environment, the shape recognition error of the insulator is large, which in turn affects the accuracy of the calculation of the creepage distance of the insulator. Summary of the invention
[0003] The purpose of the present invention is to provide a method to overcome the shortcomings of the prior art three-dimensional construction method of the insulator shape, which is not sensitive enough to noise recognition, has a large error in the insulator shape recognition in a complex external environment, and thus affects the accuracy of the insulator creepage distance calculation.
[0004] The purpose of the present invention is achieved through the following technical solutions: A three-dimensional construction method of an insulator shape based on point cloud scanning comprises the following steps: Step 1, scanning the overall shape of the insulator from different angles through point cloud, and generating point cloud data after scanning; Step 2: The point cloud data is divided into different clusters through a clustering algorithm, and points that do not belong to any cluster are excluded to complete the point cloud data denoising process; Step 3: The point cloud data scanned from different angles are stitched and fused to form a complete point cloud model; Step 4: Use modeling software to perform entity modeling on the point cloud model to obtain a monomeric three-dimensional model of the insulator.
[0005] Preferably, the step 2 is specifically as follows: Extract feature data of point cloud data and normalize the feature data; The K-Means clustering algorithm divides the point cloud into different clusters based on the feature data of the point cloud data; If the number of point clouds in a cluster is less than the set threshold, or a point cloud does not belong to any cluster, these point cloud data are judged as noise or abnormal data, and these point cloud data are removed to complete the point cloud data denoising process.
[0006] Preferably, the feature data includes the spatial position, density and reflection intensity of the point cloud data.
[0007] Preferably, before extracting the feature data of the point cloud data, the point cloud data is preprocessed to remove outliers.
[0008] Preferably, the three-dimensional construction method of the insulator shape based on point cloud scanning also verifies the result of point cloud data denoising. If the error of the denoising result exceeds the set threshold, it is determined that the point cloud data denoising process needs to be optimized, and the parameters of the clustering algorithm are readjusted so that the error of the denoising result does not exceed the set threshold.
[0009] Preferably, the parameters of the clustering algorithm are readjusted as follows: The external environment when scanning the insulator is obtained, the external environment includes the time, weather and background scene category of the insulator scanning, and the type of the scanned insulator is obtained; In the historical data, matching data similar to the external environment and having the same or similar insulator type are found, the parameters of the clustering algorithm corresponding to the matching data are obtained, and the obtained parameters are applied to this clustering algorithm.
[0010] Preferably, the searching for matching data similar to the external environment and having the same or similar insulator type is specifically: The weight value of the external environment and the weight value of the type of insulator are set. The sum of the weight value of the external environment and the weight value of the type of insulator is 1. The weight values of the time, weather and background scene category of the scanned insulator are set. The sum of the weight values of the time, weather and background scene category of the scanned insulator is equal to the weight value of the external environment. The degree of matching between the external environment and the type of insulator in the historical data and the current scanned insulator is scored. The closer the time of scanning the insulator, the more similar the weather conditions, the more similar the background scene, and the more similar the type of insulator, the higher the scores. The respective scores are multiplied by the various weight values to obtain the respective weight scores. The sum of the weight scores represents the degree of matching. The historical data with the highest sum of weight scores is selected as the matching data.
[0011] The beneficial effect of the present invention is that the present invention denoises the point cloud data through a clustering algorithm, ensuring that the point cloud data is data obtained by scanning the insulator, and the individual three-dimensional modeling of the insulator is more accurate, thereby increasing the accuracy of the calculation of the creepage distance of the insulator. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0013] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.
[0014] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.
[0015] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions. A three-dimensional construction method of an insulator shape based on point cloud scanning comprises the following steps: Step 1, scanning the overall shape of the insulator from different angles through point cloud, and generating point cloud data after scanning; Step 2: The point cloud data is divided into different clusters through a clustering algorithm, and points that do not belong to any cluster are excluded to complete the point cloud data denoising process; Step 3: The point cloud data scanned from different angles are stitched and fused to form a complete point cloud model; Step 4: Use modeling software to perform entity modeling on the point cloud model to obtain a monomeric three-dimensional model of the insulator.
[0016] The step 2 is specifically as follows: Extract feature data of point cloud data and normalize the feature data; The K-Means clustering algorithm divides the point cloud into different clusters based on the feature data of the point cloud data; If the number of point clouds in a cluster is less than the set threshold, or a point cloud does not belong to any cluster, these point cloud data are judged as noise or abnormal data, and these point cloud data are removed to complete the point cloud data denoising process.
[0017] The K-means clustering algorithm is a widely used iterative clustering analysis algorithm. Its core idea is to divide the samples in the data set into K different clusters, where each cluster is represented by the average value of all samples in the cluster (i.e., the cluster center). The algorithm steps are as follows: Randomly select K objects as initial cluster centers.
[0018] The distance between each object and each cluster center is calculated, and each object is assigned to the cluster center that is closest to it.
[0019] Recalculate the cluster center of each cluster based on the current assigned objects.
[0020] Repeat 2 and 3 until a termination condition is met, such as no (or a minimum number of) objects are reallocated to different clusters, no (or a minimum number of) cluster centers change, or the sum of squared errors is locally minimized.
[0021] The K-Means algorithm is sensitive to the selection of initial center points. Different initial center points may lead to different clustering results. In order to optimize the clustering effect, you can use improved algorithms such as K-Means++ to initialize the center points so that the distance between the initial center points is as far as possible, thereby improving the stability and accuracy of clustering.
[0022] The feature data includes the spatial position, density and reflection intensity of the point cloud data.
[0023] Before extracting the feature data of the point cloud data, the point cloud data is also preprocessed to remove outliers. The preprocessing stage is the process of cleaning the point cloud data to remove obvious outliers and abnormal values. These points may be caused by equipment errors during the scanning process, environmental factors (such as strong winds), or attachments (such as dust) on the surface of the insulator.
[0024] The three-dimensional construction method of the insulator shape based on point cloud scanning also verifies the results of point cloud data denoising. If the error of the denoising result exceeds the set threshold, it is judged that the point cloud data denoising process needs to be optimized, and the parameters of the clustering algorithm are readjusted so that the error of the denoising result does not exceed the set threshold.
[0025] The parameters of the clustering algorithm are readjusted as follows: The external environment when scanning the insulator is obtained, the external environment includes the time, weather and background scene category of the insulator scanning, and the type of the scanned insulator is obtained; In the historical data, matching data similar to the external environment and having the same or similar insulator type are found, the parameters of the clustering algorithm corresponding to the matching data are obtained, and the obtained parameters are applied to this clustering algorithm.
[0026] In this scheme, the influence of the external environment and the type of insulator on the clustering algorithm when scanning the insulator is considered. Time represents the change of light intensity. When the light intensity changes, some areas on the surface of the insulator may be densely scanned due to high reflection intensity, while other areas may be sparsely scanned due to low reflection intensity. Therefore, choosing the parameters of the clustering algorithm at the same time as much as possible can improve the accuracy of recognition. Similarly, the weather will also affect the recognition effect. For example, there will be interference from raindrops on rainy days; when the background scene is a variety of types such as cities, mountains or fields, the change of background will also affect the effect of insulator recognition. Therefore, through the scheme of this scheme, finding the historical data closest to the current insulator scan can improve the accuracy of the clustering algorithm.
[0027] The searching for matching data similar to the external environment and having the same or similar insulator type is specifically as follows: setting the weight value of the external environment and the weight value of the insulator type, the sum of the weight value of the external environment and the weight value of the insulator type is 1, setting the weight value of the time, weather and background scene category of the scanned insulator, the sum of the weight values of the time, weather and background scene category of the scanned insulator is equal to the weight value of the external environment, and scoring the degree of matching between the external environment and the type of the insulator scanned this time in the historical data, the closer the time of scanning the insulator, the more similar the weather conditions, the more similar the background scene, the more similar the type of the insulator, the higher the various scores, and multiplying each score by each weight value to obtain each weight score, the sum of the weight scores represents the degree of matching, and selecting the historical data with the highest sum of weight scores as the matching data.
[0028] The specific implementation of this plan is as follows: 1. Obtain the external environment when scanning the insulator Time to scan insulators: April 15, 2023 at 3:00 pm Weather: Sunny, light wind Background scene category: Outdoor substation 2. Get the type of insulator scanned Insulator Type: Ceramic Insulator 3. Find matching data in historical data Set the weight value: External environment weight: 0.6 Scanning time for insulator: 0.2 Weather: 0.3 Background scene category: 0.1 Weight value of insulator type: 0.4 Scoring historical data: Assume the following historical data: Data 1: Time (4:00 p.m., April 14, 2023), Weather (clear, no wind), Background (outdoor substation), Insulator type (ceramic insulator) Data 2: Time (3:00 PM, May 1, 2023), Weather (Cloudy, Light Wind), Background (Outdoor Power Pole), Insulator Type (Glass Insulator) Data 3: Time (10:00 am, April 15, 2023), Weather (clear, light wind), Background (outdoor substation), Insulator type (ceramic insulator) Weight score of data 1: Time: 0.2*(closeness, assuming 0.9) = 0.18 Weather: 0.3*(similarity, assumed to be 0.95) = 0.285 Background: 0.1*(similarity, assumed to be 1) = 0.1 Insulator type: 0.4*(similarity, assumed to be 1) = 0.4 Total score: 0.18+0.285+0.1+0.4=0.965 Weight score of data 2: Time: 0.2*(degree of proximity, assuming 0.5) = 0.1 Weather: 0.3*(similarity, assuming 0.8) = 0.24 Background: 0.1*(similarity, assumed to be 0.5) = 0.05 Insulator type: 0.4*(similarity, assumed to be 0.5) = 0.2 Total score: 0.1+0.24+0.05+0.2=0.59 Weight score of data 3: Time: 0.2*(degree of proximity, assumed to be 1) = 0.2 Weather: 0.3*(similarity, assumed to be 0.95) = 0.285 Background: 0.1*(similarity, assumed to be 1) = 0.1 Insulator type: 0.4*(similarity, assumed to be 1) = 0.4 Total score: 0.2+0.285+0.1+0.4=0.985 Select matching data: Data 3 has the highest weight score (0.985), so data 3 is selected as the matching data.
[0029] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.
[0030] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A three-dimensional construction method of an insulator shape based on point cloud scanning, characterized in that: The following steps are involved: Step 1, scanning the overall shape of the insulator from different angles through point cloud, and generating point cloud data after scanning; Step 2: The point cloud data is divided into different clusters through a clustering algorithm, and points that do not belong to any cluster are excluded to complete the point cloud data denoising process; Step 3: The point cloud data scanned from different angles are stitched and fused to form a complete point cloud model; Step 4: Use modeling software to perform entity modeling on the point cloud model to obtain a monomeric three-dimensional model of the insulator.
2. The three-dimensional construction method of the insulator shape based on point cloud scanning according to claim 1 is characterized in that: The step 2 is specifically as follows: Extract feature data of point cloud data and normalize the feature data; The K-Means clustering algorithm divides the point cloud into different clusters based on the feature data of the point cloud data; If the number of point clouds in a cluster is less than the set threshold, or a point cloud does not belong to any cluster, these point cloud data are judged as noise or abnormal data, and these point cloud data are removed to complete the point cloud data denoising process.
3. The three-dimensional construction method of the insulator shape based on point cloud scanning according to claim 2 is characterized in that: The feature data includes the spatial position, density and reflection intensity of the point cloud data.
4. The three-dimensional construction method of the insulator shape based on point cloud scanning according to claim 2 is characterized in that: Before extracting the feature data of the point cloud data, the point cloud data is preprocessed to remove outliers.
5. The three-dimensional construction method of the insulator shape based on point cloud scanning according to claim 2 is characterized in that: The results of point cloud data denoising are also verified. If the error of the denoising result exceeds the set threshold, it is judged that the point cloud data denoising process needs to be optimized, and the parameters of the clustering algorithm are readjusted so that the error of the denoising result does not exceed the set threshold.
6. The three-dimensional construction method of the insulator shape based on point cloud scanning according to claim 5 is characterized in that: The parameters of the clustering algorithm are readjusted as follows: The external environment when scanning the insulator is obtained, the external environment includes the time, weather and background scene category of the insulator scanning, and the type of the scanned insulator is obtained; In the historical data, matching data similar to the external environment and having the same or similar insulator type are found, the parameters of the clustering algorithm corresponding to the matching data are obtained, and the obtained parameters are applied to this clustering algorithm.
7. The method for three-dimensionally constructing the insulator shape based on point cloud scanning according to claim 6 is characterized in that: The searching for matching data similar to the external environment and having the same or similar insulator type is specifically as follows: The weight value of the external environment and the weight value of the type of insulator are set. The sum of the weight value of the external environment and the weight value of the type of insulator is 1. The weight values of the time, weather and background scene category of the scanned insulator are set. The sum of the weight values of the time, weather and background scene category of the scanned insulator is equal to the weight value of the external environment. The degree of matching between the external environment and the type of insulator in the historical data and the current scanned insulator is scored. The closer the time of scanning the insulator, the more similar the weather conditions, the more similar the background scene, and the more similar the type of insulator, the higher the scores. The respective scores are multiplied by the various weight values to obtain the respective weight scores. The sum of the weight scores represents the degree of matching. The historical data with the highest sum of weight scores is selected as the matching data.