Method, system and equipment for classifying panoramic point cloud data of transformer substation

By constructing the mapping relationship between cloud data of substation sites and panoramic images, identifying equipment identification cards and texts, and establishing equipment management databases, the problem of slow equipment classification and recognition speed in the substation industry is solved, and fast and high-accuracy equipment recognition is achieved Related to the management library.

CN119006927AInactive Publication Date: 2024-11-22JIANGSU SHITONG HUANYU POWER TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411230990.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-04
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing point cloud data segmentation method is rarely used in the substation industry, and cannot achieve refined classification of equipment. It is slow to identify and has a large amount of calculation. It cannot be directly associated with the substation ledger, and it is difficult to locate the target equipment.

Method used

By constructing the mapping relationship between substation point cloud data and panoramic images, removing ground and power line point cloud data, using panoramic images to identify equipment signs and text, establishing equipment management libraries, and realizing refined classification of equipment clusters.

Benefits of technology

Rapidly establish a substation equipment management library to improve the equipment identification speed and accuracy, be able to identify new equipment without retraining, and classify them in a refined manner, which facilitates inspection of target equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119006927A_ABST
    Figure CN119006927A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of point cloud data processing, and discloses a substation panoramic point cloud data classification method, system and equipment, and the method comprises the following steps: S1, carrying out the mapping of the point cloud data and a panoramic image of a substation; preprocessing the point cloud data of the transformer substation; the preprocessing comprises removing ground and power lines in the point cloud data; s2, performing clustering analysis on the preprocessed point cloud data to obtain a device cluster; s3, identifying an equipment signboard by using the panoramic image, and identifying characters on the equipment signboard as an equipment name; s4, obtaining an equipment cluster of equipment corresponding to the equipment name according to a mapping relation between the point cloud data and the panoramic image; and S5, constructing an equipment management library between the panoramic image and the point cloud data according to the equipment name, the equipment cluster and the panoramic image pixel position of the equipment corresponding to the equipment name. The method has the advantage that the incidence relation among the substation equipment management library, the point cloud data and the panoramic image data is quickly established.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud data processing, and in particular to a method, system and device for classifying panoramic point cloud data of a substation. Background Art

[0002] Substations are an indispensable part of the power system, carrying multiple tasks such as power conversion, distribution, control, and protection.

[0003] In order to ensure the safety of substation equipment inspectors, it is necessary to strictly classify substation equipment according to its live properties and divide the equipment in a refined manner to achieve better and more efficient management of substations. Based on its simplicity of modeling and high precision requirements, point cloud is one of the important representation forms of 3D data.

[0004] However, existing point cloud segmentation methods include point cloud instance segmentation algorithms based on prior experience and point cloud segmentation methods based on deep learning. The above methods are mostly used in the power transmission industry and are rarely used in the substation industry. In addition, the characteristics of substation data and transmission line data are quite different, and the focus is also different, which leads to the point cloud data not being able to play a greater role in the substation industry.

[0005] In addition, the existing cloud device identification methods for substations mostly identify equipment categories such as main transformers, lightning arresters, etc., but do not specify which device or which line the lightning arrester is on, etc. They cannot be directly associated with the substation ledger, and cannot directly locate the target device during inspections.

[0006] In addition, the amount of data in point cloud artificial intelligence algorithms is much larger than that of image data, the recognition speed is slower, and the amount of calculation is large.

[0007] Therefore, it is urgent to develop a method to solve the above problems. Summary of the invention

[0008] In order to solve the above technical problems, the present invention provides a method, system and device for classifying panoramic point cloud data of a substation, which can quickly establish an association between the substation equipment management library and point cloud data and panoramic image data, and realize the refinement of equipment classification.

[0009] In a first aspect, the first object of the present invention is to provide a method for classifying substation panorama cloud data, comprising the following steps: S1. Constructing a mapping relationship between the point cloud data of the substation and the panoramic image; preprocessing the point cloud data of the substation; the preprocessing includes removing the ground point cloud and the power line point cloud in the point cloud data; S2, clustering the preprocessed point cloud data to obtain device clusters; S3, using the panoramic image to identify the equipment identification plate, and then identifying the text on the equipment identification plate as the equipment name; S4, obtaining a device cluster of a device corresponding to a device name according to a mapping relationship between the point cloud data and the panoramic image; S5. Construct a device management library between the panoramic image and the point cloud data according to the device name, the device cluster, and the panoramic image pixel position of the device corresponding to the device name.

[0010] Further, the S3 includes: S31, extracting the device identification plate area in the panoramic image by using the color features of the identification plate in the panoramic image; S32: Recognize the text in the device identification plate area using text recognition technology as the device name.

[0011] Furthermore, it also includes: S33, obtaining a sign area M where the device name is a letter, and obtaining a sign area N where the device name includes the same letter, and if the minimum distance between M and N is less than a preset threshold, deleting M.

[0012] Further, the S31 includes: The S31 includes: S311, labeling equipment identification plates in the panoramic image to obtain a training set; S312, preprocessing the training set; the preprocessing includes geometric enhancement, color enhancement and fuzzy processing; S313, constructing a model to be trained, inputting the preprocessed training set into the model to be trained, and obtaining an equipment identification plate recognition model; S314: Use the equipment identification plate recognition model to perform equipment identification plate recognition on the panoramic image of the substation to be tested, and obtain the equipment identification plate area.

[0013] Further, the S4 includes: Map the equipment identification plate to the point cloud data to obtain the identification cluster; Calculating a first point set distance, where the first point set distance is a distance from the identification cluster to each device cluster; The correspondence between the identification cluster and the device cluster is determined based on the first point set distance, specifically: When the distance of the first point set is the smallest, if the device cluster and the identification cluster correspond one-to-one, it is determined that the device cluster and the identification cluster correspond correctly; if the device cluster and the identification cluster do not correspond one-to-one, it is determined that multiple device clusters are connected; if the device cluster does not correspond to the identification cluster, it is determined that the device corresponding to the device cluster is a support.

[0014] Furthermore, when the determination result is that multiple device clusters are connected, the following operations are performed: Acquire a device point group, wherein the device point group is point cloud data connected by multiple device clusters; Obtain a first undetermined cluster, where the first undetermined cluster is an identification cluster corresponding to the device point group when the distance of the first point set is the smallest; perform the following steps on each device point group and the corresponding first undetermined cluster respectively: Acquire a first data range, where the first data range is a maximum range of point cloud data of the first to-be-determined cluster in the vertical direction; Obtain a group of device points within the first data range, perform cluster analysis, and obtain a first cluster: If the number of the first cluster is one, the first cluster and the first undetermined cluster correspond correctly; If there are multiple first clusters, then calculate the second point set distance, where the second point set distance is the distance from the first undetermined cluster to each of the first clusters; When the distance of the second point set is the smallest, the corresponding first cluster and the first undetermined cluster are determined to be correctly corresponding.

[0015] Furthermore, when the distance of the second point set is the smallest, the corresponding first cluster and the first undetermined cluster are determined to correspond correctly, and further comprising: Acquire point cloud data that is outside the first data range in the device point group and has no corresponding relationship, perform cluster analysis to obtain a second cluster, and calculate a third point set distance, where the third point set distance is the distance from the second cluster to each of the first undetermined clusters; When the distance of the third point set is the smallest, if the corresponding first undetermined cluster is one, it is determined that the second cluster and the first undetermined cluster correspond correctly; if the corresponding first undetermined cluster is not one, it is determined that the second cluster includes multiple device structures.

[0016] Further, when the determination result is that the second cluster includes multiple device structures, the following operations are performed: Acquire a connection point group, where the connection point group is point cloud data of multiple device structures; Obtain a second undetermined cluster, which is the first undetermined cluster corresponding to the connection point group when the distance between the third point set is the smallest; perform the following steps on each connection point group and the corresponding second undetermined cluster respectively: Acquire a second data range, where the second data range is the maximum range of point cloud data of the first cluster corresponding to the second undetermined cluster in the horizontal direction; Calculate the area of ​​the second data range and sort it from small to large; The connection point groups within each second data range are acquired in sequence according to the sorting order, and the connection point groups within the second data range and the first cluster are determined to belong to the same device.

[0017] In a second aspect, a second object of the present invention is to provide a system for classifying substation panorama point cloud data, using the method for classifying substation panorama point cloud data as described in the first aspect, specifically comprising: Data processing module: constructing a mapping relationship between the substation point cloud data and the panoramic image; preprocessing the substation point cloud data; the preprocessing includes removing the ground point cloud and the power line point cloud in the point cloud data; performing cluster analysis on the preprocessed point cloud data to obtain a device cluster; Corresponding module: connected to the data processing module, used to identify the equipment identification plate by using the panoramic image, and then identify the text on the equipment identification plate as the equipment name; obtain the equipment cluster of the equipment corresponding to the equipment name according to the mapping relationship between the point cloud data and the panoramic image; Generation module: connected to the corresponding module, used to build a device management library between panoramic images and point cloud data according to the device name, device cluster, and the panoramic image pixel position of the device corresponding to the device name.

[0018] In a third aspect, a third object of the present invention is to provide an electronic device, comprising: a memory, a processor and a computer program; wherein the computer program is stored in the memory and is configured to be executed by the processor to implement the method described in the first aspect.

[0019] The embodiments of the present invention have the following technical effects: This application first identifies the signboard in the panoramic image, then identifies the text in the signboard, that is, the equipment name, and then uses the mapping relationship between the panoramic image and the point cloud data to establish the corresponding relationship between the equipment name and the point cloud data. It can quickly establish the association relationship between the substation equipment management library and the point cloud data and the panoramic image data; The present application also provides a specific method for matching equipment and point cloud data, by performing cluster analysis and comparison on the point cloud position where the equipment identification plate is located, to obtain the corresponding relationship between the two; and continue to analyze and segment according to the corresponding relationship until all point cloud data are matched with the identification cluster; In addition, this application has the following advantages: (1) Identification through equipment identification plate, with obvious features, fast recognition speed and high accuracy; (2) It can identify new equipment that has not appeared in the existing database, without the need to establish new equipment samples for retraining; it is convenient to update the changed equipment without scanning the entire site for equipment updates; (3) The classification level is refined, from the general category identification in the existing technology to the identification of detailed equipment; it is convenient to view the target equipment and to inspect the designated target equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 It is a flow chart of a method for classifying substation panorama cloud data provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of deleting duplicate equipment identification plates provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0023] Point cloud data refers to a collection of a large number of discrete points in a three-dimensional coordinate system. These points are usually obtained through LiDAR, and each point contains three-dimensional coordinates (X, Y, Z) and may carry additional information such as color (RGB), reflectivity, intensity, etc. Point cloud data is a digital representation of the surface shape of an object in three-dimensional space. It has high-precision, high-resolution and high-dimensional geometric information, and can intuitively reflect the shape, surface and texture characteristics of an object.

[0024] Point cloud data can be used to reconstruct real-world scenes, including buildings, terrain, vegetation, etc., providing a realistic three-dimensional environment for virtual reality, game development, simulation training, etc.

[0025] Panoramic images are created by stitching together a panoramic camera or multiple ordinary cameras, providing 360-degree visual information within a specified range.

[0026] The panoramic image and point cloud data can be mapped through appropriate coordinate transformations to achieve accurate registration and fusion of the two. The specific methods are existing technologies and will not be described in detail in this application. Deep learning and other methods can be used to detect and classify the fused data, while analyzing the structure and features in the scene to achieve understanding and analysis of the scene. The mapping between the two combines the advantages of three-dimensional spatial information and two-dimensional visual information, providing strong support for three-dimensional modeling, object recognition, scene analysis and other fields.

[0027] In order to facilitate the application of point cloud data, detailed equipment identification is performed on the substation site cloud data, and an association is established with the substation ledger data. A substation equipment management library is established for substations without equipment management.

[0028] The types of equipment to be identified include: main transformer, main transformer lightning arrester, main transformer fire sand box, main transformer on-load voltage regulator box, pan-tilt camera, equipment area maintenance power box, reactor rheology, switch, grounding switch, etc. The identification of the equipment will be specific to the equipment of each voltage level and specific line.

[0029] Figure 1 is a flow chart of a method for classifying substation panorama cloud data provided by an embodiment of the present invention. Figure 1 , specifically including: S1. Mapping the point cloud data of the substation and the panoramic image; preprocessing the point cloud data of the substation; the preprocessing includes removing the ground and power lines in the point cloud data.

[0030] Artificial intelligence methods can be used here to remove the ground and power lines from the point cloud data to prevent the ground and power lines from interfering with subsequent processing.

[0031] S2. Perform cluster analysis on the preprocessed point cloud data to obtain device clusters.

[0032] Cluster analysis can be performed based on spatial distribution. Euclidean clustering, density clustering, or superbody clustering can be used to group point cloud data according to differences or similarities to obtain clusters, which are defined here as device clusters, which are a preliminary distinction between point cloud data belonging to the same device and different devices.

[0033] S3. Use the panoramic image to identify the equipment identification plate, and then identify the text on the equipment identification plate as the equipment name.

[0034] The equipment in the substation is affixed with labels or identification plates, which record the equipment model and specification information, so that maintenance personnel can use the correct accessories to repair according to the equipment type to ensure the normal operation of the equipment. This application uses the equipment identification plate as the basis for identifying each device and makes relevant designs.

[0035] Specifically, the steps include: S31. Extracting a device identification plate region in the panoramic image by using color features of the identification plate in the panoramic image.

[0036] This step can be specifically implemented by extracting color features, and includes the following steps: S311. Label the equipment identification plates in the panoramic image to obtain a training set. The equipment identification plates in the substation are mainly red characters on a white background or black characters on a white background. The phase names mainly include "A", "B" and "C". Generally, the characters are white and the background color is yellow, red or green.

[0037] S312, preprocessing the panoramic image; the preprocessing includes geometric enhancement, color enhancement and blur processing; Among them, geometric enhancement is a method of changing the image geometry by translation, rotation, shearing, etc., which can enhance the generalization ability of the model; color enhancement refers to optimizing and adjusting the color information in the image through a series of image processing algorithms to enhance the visual effect of the image, such as contrast adjustment, saturation adjustment, color mapping, etc.; blur processing, such as Gaussian filtering, box filtering, median filtering, etc., can enhance the generalization ability of the model for blurred images.

[0038] S313, constructing a model to be trained, inputting the preprocessed training set into the model to be trained, and obtaining an equipment identification plate recognition model; For example, a YOLO model is established to train the preprocessed panoramic image, and the texture features of the equipment sign can be analyzed by the machine learning algorithm SVM classifier to identify the equipment sign. Among them, the YOLO model converts the target detection problem into a regression problem, and the recognition and positioning of objects in the image can be completed by viewing the image once; the SVM classifier finds an optimal hyperplane to separate data points of different categories and maximizes the interval from the hyperplane to the nearest data point, thereby realizing data classification.

[0039] S314, using the equipment identification plate recognition model to perform equipment identification plate recognition on the panoramic image of the substation to be tested, and obtain the equipment identification plate area.

[0040] By using the trained model for recognition, we can get the area where the equipment identification plate may exist, which is defined here as the equipment identification plate area.

[0041] S32. Recognize the text in the device identification plate area using text recognition technology as the device name.

[0042] Text recognition technology, such as OCR technology, is an optical character recognition technology that determines the shape of characters by detecting dark and light patterns, and uses character recognition methods to translate the shapes into computer text. OCR technology is an existing technology and will not be described in detail here. It can also be recognized by deep learning or machine learning methods, and this application does not make specific limitations on this.

[0043] S33, obtaining a sign area M whose device name is a letter, and obtaining a sign area N whose device name includes the same letter, and if the minimum distance between M and N is less than a preset threshold, deleting M.

[0044] After identification, there will be some duplicate device names, so it is necessary to delete the duplicates to ensure that the device and the device name correspond one to one, so as to avoid interference with the subsequent correspondence between the device and the device point cloud. Figure 2 , usually the letters that appear on the equipment identification plate are "A", "B" and "C", which are used to indicate a phase in the three-phase power supply. For example, if both the "A" and "No. 2 Main Transformer 110kV Side Lightning Arrester Phase A" equipment identification plates have A, and both are pasted on the same device, and the distance between the two is less than the preset threshold, such as 3cm, then the two equipment identification plates are considered to be duplicated, and the single equipment identification plate "A" is removed.

[0045] S4. Obtain a device cluster of devices corresponding to the device names according to the mapping relationship between the point cloud data and the panoramic image.

[0046] Specifically, the acquisition of point cloud data of the device corresponding to the device name includes the following: The equipment identification plate is mapped to the point cloud data to obtain an identification cluster; the identification cluster here refers to the collection of all point cloud data corresponding to the equipment identification plate.

[0047] Calculate the first point set distance, which is the distance from the identification cluster to each device cluster. The distance between point sets is a prior art, and this application will not elaborate on it. Just select a suitable method according to the needs.

[0048] The correspondence between the identification cluster and the device cluster is determined based on the first point set distance. When the first point set distance is the smallest, it is divided into the following three cases: (1) If the device cluster and the identification cluster correspond one to one, it is determined that the device cluster and the identification cluster correspond correctly; That is, when the device cluster and the identification cluster are one-to-one, it is determined that the corresponding relationship between the two is correct.

[0049] (2) If the device cluster and the identification cluster do not correspond one to one, it is determined that multiple device clusters are connected; That is, when the relationship between the device cluster and the identification cluster is one-to-many or many-to-many, it is considered that the distance between multiple devices is too close to form a connection. There will be multiple groups of multiple devices connected, and each group needs to be analyzed separately.

[0050] (3) If the device cluster does not correspond to the identification cluster, it is determined that the device to which the device cluster corresponds is a support frame.

[0051] Structural supports refer to various support systems used to support building structures and frames. The equipment cluster of structural supports does not have a matching identification cluster.

[0052] Among them, in the case of multiple devices connected, the above process cannot accurately judge, and it is necessary to continue to judge each group of connected devices: First, the scope of the point cloud data that needs to be judged is limited, that is, the point cloud data and the associated identification clusters when the device clusters are connected are judged: Obtain device point groups, which are point cloud data connected by multiple device clusters; Obtain a first pending cluster, which is an identification cluster corresponding to the device point group when the distance of the first point set is the smallest; perform the following steps on each device point group and the corresponding first pending cluster respectively: Acquire a first data range, where the first data range is the maximum range of point cloud data of the first to-be-determined cluster in the vertical direction; A device point group located within a first data range is obtained, and cluster analysis is performed to obtain a first cluster.

[0053] Exemplarily, the first data range of the point cloud data of the equipment identification plate in the vertical plane is obtained, that is, the Z value range of the first pending cluster is obtained: Z_max, Z_min; under the limitation of the large range of the equipment point group, all point cloud data within the first data range are obtained, and cluster analysis is performed to analyze various situations to obtain the corresponding results between the first pending cluster and the equipment point cloud data.

[0054] The specific analysis is as follows: 1. If the number of the first cluster is one, the first cluster and the first undetermined cluster correspond correctly; If the number of the first cluster is only one, it means that it is uniquely corresponding to the first undetermined cluster, and there is a corresponding relationship between the two.

[0055] 2. If there are multiple first clusters, the second point set distance is calculated, where the second point set distance is the distance from the first undetermined cluster to each first cluster; when the second point set distance is the smallest, the corresponding first cluster and the first undetermined cluster are judged to correspond correctly.

[0056] When the number of the first clusters is greater than 1, it is not possible to determine which one corresponds to the first undetermined cluster. In this case, the first cluster closest to the first undetermined cluster corresponds correctly.

[0057] Based on the previous segmentation conditions, that is, point cloud data within the Z value range is obtained, only the point cloud data of devices at the same level as the first pending cluster is obtained, while the point cloud data of other parts of each device are not segmented in, and those that are not segmented are not matched to the correct first pending cluster, so it is impossible to accurately identify which device it belongs to. For this situation, further judgment is needed.

[0058] Point cloud data that is outside the first data range in the device point group and has no corresponding relationship is obtained, and cluster analysis is performed to obtain a second cluster cluster, which is a collection of point cloud data that has not been determined.

[0059] Calculate the third point set distance, which is the distance from the second cluster to each first undetermined cluster. When the third point set distance is the smallest: 1) If the corresponding first undetermined cluster is one, it is determined that the second cluster and the first undetermined cluster correspond correctly; When the first undetermined cluster is unique, the relationship between the first undetermined cluster and the second clustering cluster may also correspond uniquely.

[0060] 2) If the corresponding first undetermined cluster is not one, it is determined that the second cluster includes multiple device structures.

[0061] When devices are adjacent, clustering may group parts of the structures between different devices into the same cluster, resulting in a large number of corresponding first pending clusters, and it is impossible to confirm the ownership of each part. In order to reduce this uncertainty, it is still necessary to make judgments and split them: When the determination result is that the second cluster includes multiple device structures, the following operations are performed: A connection point group is obtained, where the connection point group is point cloud data of multiple device structures. First, a range of point cloud data to be segmented is determined, that is, the connection point group is segmented to correspond to each second pending cluster.

[0062] Obtain the second undetermined cluster, which is the first undetermined cluster corresponding to the connection point group when the distance between the third point set is the smallest; perform the following steps on each connection point group and the corresponding second undetermined cluster respectively: Obtaining a second data range of the first cluster corresponding to the second undetermined cluster in the horizontal direction; Calculate the area of ​​the second data range and sort it from small to large; The connection point groups within each second data range are acquired in sequence according to the sorting order, and the connection point group corresponding to the second data range and the first cluster are determined to belong to the same device.

[0063] The second data range of the first cluster corresponding to the second pending cluster in the horizontal direction is obtained, that is, the XY range of the point cloud data that can be obtained when looking down, and the connection point group is segmented according to the second data range. Among them, the application is segmented according to the order of the area size of the second data range until all matches are completed; the ordering by area here is to obtain an order to maximize the accuracy of device point cloud separation and reduce the amount of calculation.

[0064] The above process can also be understood as: obtaining the XY range of the device part corresponding to the identification cluster, and then segmenting the remaining point cloud data that meets the range, and determining that the segmented point cloud data and the device part belong to the same device and have the same identification cluster.

[0065] In practice, for mismatched point cloud data, manual correction methods can be used to minimize errors and reduce the probability of errors.

[0066] It should be noted that the present application does not limit any cluster analysis methods, as long as they meet actual needs.

[0067] In summary, the point cloud data of the device corresponding to the device identification plate, that is, the device cluster, is obtained.

[0068] S5. Construct a device management library between the panoramic image and the point cloud data according to the device name, the device cluster, and the panoramic image pixel position of the device corresponding to the device name.

[0069] Establish an equipment management library, manage various data in the library, manually correct identification errors, and store the managed data as the basic data for subsequent queries and inspections.

[0070] It should be noted that after each device is matched with the point cloud data, the relevant attributes can also be calculated, including the range of the point cloud data, the center position and size of the device, etc., and all the information can be recorded in the device management library.

[0071] Therefore, this application establishes a corresponding relationship between the device name and the point cloud data by first identifying the signboard in the panoramic image, then identifying the text in the signboard, that is, the equipment name, and then using the mapping relationship between the panoramic image and the point cloud data. This can quickly establish an association relationship between the substation equipment management library and the point cloud data and panoramic image data.

[0072] Example 2 Based on Example 1, the present application provides a system for classifying substation panoramic cloud data, including: Data processing module: constructing a mapping relationship between the substation point cloud data and the panoramic image; preprocessing the substation point cloud data; the preprocessing includes removing the ground point cloud and the power line point cloud in the point cloud data; performing cluster analysis on the preprocessed point cloud data to obtain a device cluster; Corresponding module: connected to the data processing module, used to identify the equipment identification plate by using the panoramic image, and then identify the text on the equipment identification plate as the equipment name; obtain the equipment cluster of the equipment corresponding to the equipment name according to the mapping relationship between the point cloud data and the panoramic image; Generation module: connected to the corresponding module, used to build a device management library between panoramic images and point cloud data according to the device name, device cluster, and the panoramic image pixel position of the device corresponding to the device name.

[0073] For other contents, please refer to Example 1 and will not be elaborated here.

[0074] Example 3 The present application provides an electronic device, comprising: a memory, a processor, and a computer program; wherein the computer program is stored in the memory and is configured to be executed by the processor to implement the method in Example 1.

[0075] Based on the above technical solution, the present application uses equipment identification tags for identification, which has obvious features, fast recognition speed and high accuracy; it can identify new equipment that has not appeared in the existing database without the need to establish samples of new equipment for retraining.

[0076] It should be noted that the terms used in the present invention are only for describing specific embodiments, rather than limiting the scope of the present application. As shown in the present specification, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of more restrictions, the elements defined by the sentence "include one..." do not exclude the presence of other identical elements in the process, method or device including the elements.

[0077] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for classifying substation panorama cloud data, characterized in that: The steps include: S1. Constructing a mapping relationship between the point cloud data of the substation and the panoramic image; preprocessing the point cloud data of the substation; the preprocessing includes removing the ground point cloud and the power line point cloud in the point cloud data; S2, clustering the preprocessed point cloud data to obtain device clusters; S3, using the panoramic image to identify the equipment identification plate, and then identifying the text on the equipment identification plate as the equipment name; S4, obtaining a device cluster of a device corresponding to a device name according to a mapping relationship between the point cloud data and the panoramic image; S5. Construct a device management library between the panoramic image and the point cloud data according to the device name, the device cluster, and the panoramic image pixel position of the device corresponding to the device name.

2. The method for classifying substation panorama cloud data according to claim 1, characterized in that: The S3 includes: S31, extracting the device identification plate area in the panoramic image by using the color features of the identification plate in the panoramic image; S32: Recognize the text in the device identification plate area using text recognition technology as the device name.

3. The method for classifying substation panorama cloud data according to claim 2, characterized in that: Also includes: S33, obtaining a sign area M where the device name is a letter, and obtaining a sign area N where the device name includes the same letter, and if the minimum distance between M and N is less than a preset threshold, deleting M.

4. The method for classifying substation panorama cloud data according to claim 2, characterized in that: The S31 includes: S311, labeling equipment identification plates in the panoramic image to obtain a training set; S312, preprocessing the training set; the preprocessing includes geometric enhancement, color enhancement and fuzzy processing; S313, constructing a model to be trained, inputting the preprocessed training set into the model to be trained, and obtaining an equipment identification plate recognition model; S314: Use the equipment identification plate recognition model to perform equipment identification plate recognition on the panoramic image of the substation to be tested, and obtain the equipment identification plate area.

5. The method for classifying substation panorama cloud data according to claim 1, characterized in that: The S4 includes: Map the equipment identification plate to the point cloud data to obtain the identification cluster; Calculating a first point set distance, where the first point set distance is a distance from the identification cluster to each device cluster; The correspondence between the identification cluster and the device cluster is determined based on the first point set distance, specifically: When the distance of the first point set is the smallest, if the device cluster and the identification cluster correspond one-to-one, it is determined that the device cluster and the identification cluster correspond correctly; if the device cluster and the identification cluster do not correspond one-to-one, it is determined that multiple device clusters are connected; if the device cluster does not correspond to the identification cluster, it is determined that the device corresponding to the device cluster is a support.

6. The method for classifying substation panorama cloud data according to claim 5, characterized in that: When the result is that multiple device clusters are connected, perform the following operations: Acquire a device point group, wherein the device point group is point cloud data connected by multiple device clusters; Obtain a first undetermined cluster, where the first undetermined cluster is an identification cluster corresponding to the device point group when the distance of the first point set is the smallest; perform the following steps on each device point group and the corresponding first undetermined cluster respectively: Acquire a first data range, where the first data range is a maximum range of point cloud data of the first to-be-determined cluster in the vertical direction; Obtain a group of device points within the first data range, perform cluster analysis, and obtain a first cluster: If the number of the first cluster is one, the first cluster and the first undetermined cluster correspond correctly; If there are multiple first clusters, then calculate the second point set distance, where the second point set distance is the distance from the first undetermined cluster to each of the first clusters; When the distance of the second point set is the smallest, the corresponding first cluster and the first undetermined cluster are determined to be correctly corresponding.

7. The method for classifying substation panorama cloud data according to claim 6, characterized in that: When the distance of the second point set is the smallest, the corresponding first cluster and the first undetermined cluster are determined to correspond correctly, and the method further includes: Acquire point cloud data that is outside the first data range in the device point group and has no corresponding relationship, perform cluster analysis to obtain a second cluster, and calculate a third point set distance, where the third point set distance is the distance from the second cluster to each of the first undetermined clusters; When the distance of the third point set is the smallest, if the corresponding first undetermined cluster is one, it is determined that the second cluster and the first undetermined cluster correspond correctly; if the corresponding first undetermined cluster is not one, it is determined that the second cluster includes multiple device structures.

8. The method for classifying substation panorama cloud data according to claim 7, characterized in that: When the determination result is that the second cluster includes multiple device structures, the following operations are performed: Acquire a connection point group, where the connection point group is point cloud data of multiple device structures; Obtain a second undetermined cluster, which is the first undetermined cluster corresponding to the connection point group when the distance between the third point set is the smallest; perform the following steps on each connection point group and the corresponding second undetermined cluster respectively: Acquire a second data range, where the second data range is the maximum range of point cloud data of the first cluster corresponding to the second undetermined cluster in the horizontal direction; Calculate the area of ​​the second data range and sort it from small to large; The connection point groups within each second data range are acquired in sequence according to the sorting order, and the connection point groups within the second data range and the first cluster are determined to belong to the same device.

9. A system for classifying substation panorama cloud data, characterized in that: The method for classifying substation panoramic cloud data according to any one of claims 1 to 8 specifically comprises: Data processing module: constructing a mapping relationship between the substation point cloud data and the panoramic image; preprocessing the substation point cloud data; the preprocessing includes removing the ground point cloud and the power line point cloud in the point cloud data; performing cluster analysis on the preprocessed point cloud data to obtain a device cluster; Corresponding module: connected to the data processing module, used to identify the equipment identification plate by using the panoramic image, and then identify the text on the equipment identification plate as the equipment name; obtain the equipment cluster of the equipment corresponding to the equipment name according to the mapping relationship between the point cloud data and the panoramic image; Generation module: connected to the corresponding module, used to build a device management library between panoramic images and point cloud data according to the device name, device cluster, and the panoramic image pixel position of the device corresponding to the device name.

10. An electronic device, characterized in that: include: A memory, a processor, and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method according to any one of claims 1 to 8.