Substation monomer vectorization modeling method, electronic equipment and storage medium

Abstract

The invention discloses a substation monomer vectorization modeling method, electronic equipment and a storage medium, and the method comprises the steps: obtaining to-be-processed point cloud data corresponding to a substation, carrying out the preprocessing of the to-be-processed point cloud data, and obtaining the preprocessed point cloud data; processing the preprocessed point cloud data based on a Hough voting algorithm to obtain an outline and a feature vector corresponding to the target equipment in the transformer substation; according to the contour and the feature vector, segmenting the preprocessed point cloud data to obtain device point cloud data of the target device; and identifying the equipment point cloud data through the target deep learning model to obtain component point cloud data of the target equipment, and performing vectorization processing on the component point cloud data to generate a single vectorization model of the target equipment. According to the technical scheme of the invention, vectorization modeling of the single equipment of the transformer substation can be realized, and a basis is provided for planning of an intelligent patrol path, so that intelligent construction of the transformer substation is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of substation monomer vectorization model construction, and in particular to a substation monomer vectorization modeling method, electronic equipment and storage medium. Background Art

[0002] With the continuous development of power grid technology, the automation level of substations has been continuously improved. At present, there are several significant characteristics of the development of domestic substations: the proportion of unmanned substations has increased; the deployment scope of robot systems in stations has expanded; and the number of stations responsible for operation and maintenance teams has increased. The problem that comes with it is that the number, scale and complexity of substations under the jurisdiction of each manager have doubled compared with the previous scope of responsibility. There are many substations with equipment, and there are deficiencies in many aspects such as asset management, inspection task planning, autonomous inspection, and intelligent machine inspection terminal applications, which make it difficult to meet the requirements of digital management and operation and maintenance of substation equipment.

[0003] According to relevant measures, it is proposed to promote the upgrade of substation video and environmental monitoring systems. The terminal cameras of the video monitoring system of the substation in operation will be installed or modified. Through fixed cameras, wireless cameras, robots, drones and other intelligent terminals, wireless communication technologies such as 5G and WAPI will be adopted to gradually realize the intelligent inspection mode integrating multiple technologies, and completely replace the daily manual inspections.

[0004] In order to facilitate the implementation of intelligent patrol mode, it is usually necessary to apply point cloud vectorization technology for equipment modeling. How to efficiently and accurately realize the modeling of complex scenes and promote the intelligent construction of substations is a technical problem that needs to be solved urgently. Summary of the invention

[0005] The present invention provides a substation monomer vector modeling method, electronic equipment and storage medium to realize the generation of a standard vector model. The equipment monomer model of the substation can realize the combination of point cloud data and three-dimensional model, facilitate the planning of intelligent patrol paths, explore the autonomous positioning and navigation of the substation, and promote the intelligent construction of the substation.

[0006] According to one aspect of the present invention, a method for vector modeling of a substation unit is provided, comprising:

[0007] Acquire the to-be-processed point cloud data corresponding to the substation, and pre-process the to-be-processed point cloud data to obtain pre-processed point cloud data;

[0008] Processing the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target equipment in the substation;

[0009] Segmenting the pre-processed point cloud data according to the outer contour and the feature vector to obtain device point cloud data of the target device;

[0010] The device point cloud data is identified through a target deep learning model to obtain component point cloud data of the target device, and the component point cloud data is vectorized to generate a single vectorized model of the target device.

[0011] In a possible implementation, the step of obtaining the to-be-processed point cloud data corresponding to the substation includes:

[0012] According to the layout and equipment distribution of the substation, a scanning path and site settings are formulated, and the point cloud data to be processed corresponding to the substation is obtained based on a high-precision three-dimensional laser scanner.

[0013] In a possible implementation, preprocessing the point cloud data to be processed to obtain preprocessed point cloud data includes:

[0014] Down-sampling is performed on the point cloud data to be processed to obtain simplified point cloud data, and noise reduction is performed on the simplified point cloud data to obtain the pre-processed point cloud data.

[0015] In a possible implementation, downsampling the point cloud data to be processed to obtain simplified point cloud data includes:

[0016] Dividing the point cloud data to be processed into voxel grids of fixed size, and determining the non-empty body centroid corresponding to each voxel grid;

[0017] The point cloud data to be processed in the voxel grid is replaced by the centroid of the non-empty volume to obtain the simplified point cloud data.

[0018] In a possible implementation, the processing of the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target equipment in the substation includes:

[0019] Extracting local feature points of the preprocessed point cloud data and calculating local descriptors of the local feature points;

[0020] Matching the feature points in the preset model space with the local feature points based on the local descriptors to determine the target feature points;

[0021] The outer shape contour and feature vector corresponding to the target device are determined based on the target feature points.

[0022] In a possible implementation, matching the feature points in the preset model space with the local feature points based on the local descriptors to determine the target feature points includes:

[0023] Determine a descriptor to be used and a shape index value to be used corresponding to the feature point in the preset model space;

[0024] Calculating the Euclidean distance between the descriptor to be used and the local descriptor, and if the Euclidean distance satisfies a first preset threshold, determining the local feature point as a candidate feature point;

[0025] A target shape index value of the candidate feature point is determined, and if a difference between the target shape index value and the shape index value to be used satisfies a second preset threshold, the candidate feature point is determined as the target feature point.

[0026] In a possible implementation, segmenting the pre-processed point cloud data according to the outer contour and the feature vector to obtain device point cloud data of the target device includes:

[0027] According to the outer contour and the feature vector, the pre-processed point cloud data is segmented using a clustering algorithm or a graph cut algorithm to obtain device point cloud data of the target device.

[0028] In a possible implementation, identifying the device point cloud data through a target deep learning model to obtain component point cloud data of the target device, and vectorizing the component point cloud data to generate a single vectorized model of the target device includes:

[0029] Using the KPConv model, different parts of the device point cloud data are identified and split into component point cloud data;

[0030] The component point cloud data is converted into vector data using a vectorization algorithm, and the monomer vectorization model is constructed based on the vector data.

[0031] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0032] at least one processor;

[0033] and a memory communicatively connected to the at least one processor; wherein,

[0034] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the substation single-unit vector modeling method described in any embodiment of the present invention.

[0035] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the substation single-unit vector modeling method described in any embodiment of the present invention when executed.

[0036] The technical solution of the embodiment of the present invention includes: obtaining the to-be-processed point cloud data corresponding to the substation, pre-processing the to-be-processed point cloud data to obtain pre-processed point cloud data; processing the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target device in the substation; segmenting the pre-processed point cloud data according to the shape contour and feature vector to obtain the device point cloud data of the target device; identifying the device point cloud data through the target deep learning model to obtain the component point cloud data of the target device, and vectorizing the component point cloud data to generate a single vectorized model of the target device. The technical solution of the present invention can realize the vectorized modeling of single substation equipment, provide a basis for the planning of intelligent patrol routes, and promote the intelligent construction of substations.

[0037] In response to the needs of unmanned duty and autonomous inspection of substations, this paper studies the rapid individual modeling technology of substation equipment, which can associate the model with the ledger, give the model more attribute information, and establish a limited degree vector model library of the substation. It meets the requirements of efficient operation and maintenance work in multiple scenarios within the station, improves the digital management capabilities of substation assets, and empowers intelligent operation and maintenance.

[0038] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of a substation single-unit vector modeling method provided in Embodiment 1 of the present invention;

[0041] Figure 2 A flowchart of a substation single-unit vector modeling method provided in the second embodiment of the present invention;

[0042] Figure 3 A schematic diagram of feature point matching provided in Embodiment 2 of the present invention;

[0043] Figure 4 A schematic diagram of Hough voting provided in Embodiment 2 of the present invention;

[0044] Figure 5 is the depth image of the circuit breaker;

[0045] Figure 6 is the distance histogram;

[0046] Figure 7 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0048] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0049] Embodiment 1

[0050] Figure 1 This is a flowchart of a method for vector modeling of a substation unit provided in the first embodiment of the present invention. This embodiment is applicable to the case of establishing a single unit vector model of a substation device. The method can be executed by a substation unit vector modeling device. The device can be implemented in the form of hardware and / or software. The device can be configured in an electronic device. Figure 1 As shown, the method specifically comprises the following steps:

[0051] S110, obtaining the unprocessed point cloud data corresponding to the substation, and preprocessing the unprocessed point cloud data to obtain preprocessed point cloud data.

[0052] The point cloud data to be processed may be the collected point cloud data of the entire substation. The point cloud data to be processed has not been preprocessed. The point cloud data to be processed may be preprocessed to obtain preprocessed point cloud data.

[0053] In a possible implementation, the point cloud data to be processed corresponding to the substation is obtained, including: formulating a scanning path and site settings according to the layout and equipment distribution of the substation, and scanning the point cloud data to be processed corresponding to the substation based on a high-precision three-dimensional laser scanner.

[0054] Specifically, you can choose a high-precision 3D laser scanner to ensure that the acquired point cloud data has high resolution and accuracy. According to the layout and equipment distribution of the substation, formulate a reasonable scanning path and site settings to ensure that all key areas and equipment are covered.

[0055] In a possible implementation, the point cloud data to be processed is preprocessed to obtain preprocessed point cloud data, including: downsampling the point cloud data to be processed to obtain simplified point cloud data, and denoising the simplified point cloud data to obtain the preprocessed point cloud data.

[0056] Specifically, the collected original point cloud data to be processed usually contains a lot of noise and redundant information, so preprocessing is required, mainly including downsampling and denoising.

[0057] In a possible implementation, the point cloud data to be processed is downsampled to obtain simplified point cloud data, including: dividing the point cloud data to be processed into voxel grids of a fixed size, determining the centroid of a non-empty body corresponding to each voxel grid; and replacing the point cloud data to be processed in the voxel grid with the centroid of the non-empty body to obtain the simplified point cloud data.

[0058] Among them, downsampling refers to the use of voxel grid downsampling to reduce the amount of point cloud data and reduce the amount of calculation for subsequent processing.

[0059] Due to the huge amount of point cloud data in the substation, directly processing all points will lead to a waste of computing resources. The voxel grid downsampling method is used to divide the point cloud into a fixed-size voxel grid. The principle of body centroid downsampling is very simple. First, the point cloud is divided into voxels, and then the centroid of the non-empty voxel is calculated to replace all the points in the voxel to achieve point cloud downsampling. The calculation of body centroid is as follows:

[0060] Given a non-empty voxel V containing m points, we need to calculate the centroid Pcentroid (Xcentroid, Ycentroid, Zcentroid) of the voxel.

[0061] The centroid is calculated as follows: Xcentroid: Calculate the average x-coordinate value of all points. The formula is

[0062] Ycentroid: Calculates the average y coordinate value of all points. The formula is

[0063] Zcentroid: Calculates the average z-coordinate value of all points. The formula is:

[0064] Furthermore, point cloud denoising refers to eliminating or reducing the noise in point cloud data to improve the cleanliness and accuracy of the data. Statistical filters and voxel grid filters can be used to denoise point cloud data, remove isolated points and abnormal points, and improve data quality. The main idea of ​​statistical filtering is to remove outliers by judging the distance between each point in the point cloud and the k points adjacent to it. The specific algorithm steps are as follows:

[0065] 1. Read in the simplified 3D point cloud data;

[0066] 2. Build a kd-tree based on the read point cloud, establish the topological relationship between each point, and record the k nearest neighbors N(p) of pi;

[0067] 3. Calculate the average distance between each point in the point cloud and its k nearest neighbors, denoted as di. The average distance formula is:

[0068]

[0069] 4. Where pj is a point in the k-neighborhood of point pi;

[0070] Calculate the global average distance D and global standard deviation S:

[0071]

[0072] Where N is the total number of points contained in the point cloud;

[0073] 5. Set the threshold σ to determine whether the di of each point is greater than D+S*σ. If so, the point is considered an outlier and deleted.

[0074] S120. Process the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target equipment in the substation.

[0075] Specifically, in the substation cloud data, the Hough voting algorithm is used to identify the outlines of equipment such as transformers and switchgear. Through the voting mechanism, the algorithm can identify the area in the point cloud that matches the shape of the equipment, thereby determining the approximate location and shape of the equipment.

[0076] In one possible implementation, principal component analysis (PCA) is a statistical method for dimensionality reduction and data compression. By calculating the covariance matrix of point cloud data, the main components of the data (i.e., eigenvectors) are extracted to achieve a simplified representation of the data. When identifying substation equipment, using PCA to extract the eigenvectors of the equipment helps to further classify and identify the equipment.

[0077] S130. Segment the pre-processed point cloud data according to the outer contour and the feature vector to obtain device point cloud data of the target device.

[0078] Specifically, according to the identified device outline and feature vector, a clustering algorithm (such as K-means, DBSCAN) or a graph cut algorithm is used to segment the point cloud data into individual devices, thereby obtaining the device point cloud data of the target device.

[0079] S140. Identify the device point cloud data through a target deep learning model to obtain component point cloud data of the target device, and vectorize the component point cloud data to generate a single vectorized model of the target device.

[0080] Among them, the target deep learning model can be a KPConv model, which is a three-dimensional point cloud processing model, and the component point cloud data refers to the point cloud data corresponding to the key components of the target device.

[0081] Specifically, the KPConv model is used to identify different parts of the device point cloud (such as the housing, internal components, etc.) and split them into independent component point clouds. Then, the component point cloud data is vectorized to generate a single vectorized model of the target device.

[0082] The technical solution of the embodiment of the present invention includes: obtaining the to-be-processed point cloud data corresponding to the substation, pre-processing the to-be-processed point cloud data to obtain pre-processed point cloud data; processing the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target device in the substation; segmenting the pre-processed point cloud data according to the shape contour and feature vector to obtain the device point cloud data of the target device; identifying the device point cloud data through the target deep learning model to obtain the component point cloud data of the target device, and vectorizing the component point cloud data to generate a single vectorized model of the target device. The technical solution of the present invention can realize the vectorized modeling of single substation equipment, provide a basis for the planning of intelligent patrol routes, and promote the intelligent construction of substations.

[0083] In response to the needs of unmanned duty and autonomous inspection of substations, this paper studies the rapid individual modeling technology of substation equipment, which can associate the model with the ledger, give the model more attribute information, and establish a limited degree vector model library of the substation. It meets the requirements of efficient operation and maintenance work in multiple scenarios within the station, improves the digital management capabilities of substation assets, and empowers intelligent operation and maintenance.

[0084] Embodiment 2

[0085] Figure 2 This is a flowchart of a substation single vector modeling method provided in the second embodiment of the present invention. Based on the above embodiment, this embodiment further specifies the Hough voting algorithm and the vector modeling process. The specific implementation method can refer to the technical solution of this embodiment. Among them, the technical terms that are the same or corresponding to the above embodiment are not repeated here. Figure 2 As shown, the method specifically comprises the following steps:

[0086] S210, obtaining the unprocessed point cloud data corresponding to the substation, and preprocessing the unprocessed point cloud data to obtain preprocessed point cloud data.

[0087] S220: extracting local feature points of the preprocessed point cloud data, and calculating local descriptors of the local feature points.

[0088] Specifically, local features of preprocessed point cloud data may be extracted first. Extracting local features of point cloud may include two methods: extracting key points by calculating the density around each point in the point cloud or extracting key points by using features such as curvature and surface normal vectors in the point cloud.

[0089] In order to better describe the geometric information around the feature point and facilitate the matching of the feature point, the local descriptor of the feature point can be calculated. The descriptor is a vector that contains the geometric information around the feature point, such as curvature, normal direction, etc. By comparing the descriptors of two feature points, it can be determined whether they belong to the same device or shape.

[0090] S230 , matching feature points in a preset model space with the local feature points based on the local descriptors to determine target feature points.

[0091] The preset model space refers to a pre-set model space, which includes feature points of each type of equipment.

[0092] Specifically, feature points in the preset model space can be matched with local feature points based on local descriptors. If the local feature point matches the feature point in the preset model space appropriately, it means that the local feature point belongs to a feature point of a certain device, that is, the local feature point can be determined as a target feature point.

[0093] In a possible implementation, matching the feature points in a preset model space with the local feature points based on the local descriptors to determine the target feature points includes: determining the descriptor to be used and the shape index value to be used corresponding to the feature points in the preset model space; calculating the Euclidean distance between the descriptor to be used and the local descriptor, and if the Euclidean distance satisfies a first preset threshold, determining the local feature point as a candidate feature point; determining the target shape index value of the candidate feature point, and if the difference between the target shape index value and the shape index value to be used satisfies a second preset threshold, determining the candidate feature point as the target feature point.

[0094] The descriptor to be used refers to the descriptor of the feature point in the preset model space, and the shape index value to be used refers to the shape index value of the feature point in the preset model space.

[0095] After having the descriptors, we can match the feature points. The basic matching process is to compare the descriptors of two feature points. If their descriptors are very similar, that is, the Euclidean distance is very small and meets the first preset threshold, then I think the two feature points are matched, that is, the local feature points are determined as candidate feature points. However, there may be mismatches based solely on the similarity of descriptors. Therefore, further verification is required, such as comparing the shape index values ​​around the feature points.

[0096] Exemplarily, the basic matching process includes the following steps:

[0097] Establish a local coordinate system: Establish a local coordinate system based on the feature points to better describe the geometric information around the feature points.

[0098] Extract features and calculate descriptors: Use a suitable feature detector to extract local features of feature points and calculate descriptors.

[0099] Calculate Euclidean distance: For each pair of feature points in model space and target space, calculate the Euclidean distance between their descriptors.

[0100] Determine matching points: Determine matching feature point pairs based on the size of the Euclidean distance. Usually a threshold is set, and only feature point pairs with a Euclidean distance less than the threshold are considered matched.

[0101] Verify the matching results: Verify the correctness of the matching results by further comparing the shape index values ​​or other information around the feature points.

[0102] In addition, the Hough voting mechanism can also be used to verify the correctness of the matching results and discard incorrect corresponding points.

[0103] In an optional embodiment, during the feature point matching process, when comparing the shape index values ​​around the feature points, the shape index values ​​around the feature points in the model space are usually compared with the shape index values ​​around the candidate matching points. By comparing the shape index values ​​around the two feature points, it can be further verified whether they are actually matched.

[0104] Specifically, the following steps can be included:

[0105] Calculate shape index values: For each feature point in the model space and target space, calculate the shape index values ​​around it respectively.

[0106] Shape index value comparison: For each candidate matching point, the shape index value around it is compared with the shape index value around the corresponding feature point in the model space, for example, the difference between the two is calculated.

[0107] Decision: Based on the comparison results of the shape index values, the candidate matching points can be further screened, and those points with higher similarity in shape index values ​​can be retained as the final matching points.

[0108] In a preferred embodiment, the basic steps of 3D target recognition based on Hough voting and 3D Hough voting algorithm include:

[0109] 1. Descriptor acquisition and coordinate conversion;

[0110] 2. Hough transform is to use local features to vote and identify the category to which the target belongs. Since the angle and distance from the target are different each time the scanner scans the point cloud in the actual recognition process, the local surface characteristics of the point cloud are used to make the recognition more robust. After the local coordinate system is established, the local descriptor can be established. The general descriptor acquisition process is: establish a local coordinate system based on feature points, extract features through a suitable feature detector, and then obtain a local descriptor.

[0111] The matching between feature points can be completed based on the descriptor of each feature point. The basic matching process is: first, find the index values ​​of all points within the radius r, and take the average value as the shape index value of the point when comparing. Then, calculate the Euclidean distance based on the descriptor, obtain the candidate points within the threshold, and compare the shape index values ​​of the two points starting from the most similar point. If the difference is within another threshold, the point is considered to be a matching point. Otherwise, continue searching. If no matching point is found among the candidate points, it is considered that no corresponding point can be found on the model, and the point is discarded. Figure 3 , which is a schematic diagram of feature point matching provided in Embodiment 2 of the present invention.

[0112] When matching feature points, if the wrong corresponding points need to be discarded, Hough voting can effectively discard the wrong corresponding points. For example, if a point searches for an incorrect corresponding point in the model space, the centroid position obtained in the subsequent coordinate transformation and voting process will deviate from the true position, and the voting result will be considered invalid. Therefore, the voting process is actually a test of whether the previous feature point matching is correct or not.

[0113] The basic idea of ​​Hough transform is to transform target detection into the problem of finding the peak value in a small space. After the voting is over, according to the set threshold, if the number of votes is greater than the threshold, it means that the device exists in the scene, and the recognition result is output. If the number of votes is less than the threshold, it means that the device does not exist in the model library, and the recognition fails. Figure 4 FIG. 1 is a schematic diagram of Hough voting provided in Embodiment 2 of the present invention.

[0114] In another optional embodiment, in order to more efficiently determine the device point cloud data of the target device, a preliminary identification may be performed based on the pre-processed point cloud data before performing local feature point matching.

[0115] Specifically, a depth image can be generated based on the preprocessed point cloud data. When constructing a distance histogram based on the depth image, the distance histogram corresponding to the preset model space is matched with the distance histogram corresponding to the preprocessed point cloud data to perform preliminary recognition to determine which of the preprocessed point cloud data is the point cloud data corresponding to the device. The process of distance histogram and similarity discrimination can be specifically as follows:

[0116] The point cloud is converted into a depth image using a digital surface model (DSM). That is, when there are multiple points with the same x and y coordinate values, the point with the larger height coordinate value is selected, so that the distribution of elevation difference on the oxy plane is obtained. The process of obtaining the depth image of each device is relatively simple, but has good resolution. Among them, the depth image of the circuit breaker is as follows Figure 5 As shown:

[0117] The acquired depth image can be used to obtain the elevation distribution of the object. In actual scenes, the height of the equipment is fixed, so a distance histogram can be constructed based on the elevation difference distribution. When constructing the distance histogram, the maximum value of the distance is set to the maximum height value of all devices in the model library, and the minimum value is 0. The entire height range is divided into 36 levels, and the number of points in each interval is counted. In order to make the distance histogram more robust, the number of point clouds in each interval is divided by the total number of point clouds, that is, the percentage of each interval as a point cloud distribution is the final distance histogram. Among them, Figure 6 The distance histogram is shown.

[0118] The similarity can be determined using the following formula.

[0119]

[0120] Where cov(x,y) represents covariance, and var(x) represents variance. The value of corr represents the similarity between the two, and the larger the value, the higher the similarity.

[0121] S240: Determine an outline and a feature vector corresponding to the target device based on the target feature points.

[0122] S250. According to the outer contour and the feature vector, the pre-processed point cloud data is segmented by using a clustering algorithm or a graph cut algorithm to obtain device point cloud data of the target device.

[0123] Specifically, based on the identified shape contours and feature vectors, a clustering algorithm (such as K-means, DBSCAN) or a graph cut algorithm is used to segment the point cloud data into individual devices.

[0124] S260: Use the KPConv model to identify different parts in the device point cloud data and split them into component point cloud data.

[0125] S270, converting the component point cloud data into vector data using a vectorization algorithm, and constructing the monomer vectorization model based on the vector data.

[0126] Specifically, the KPConv model is used to identify different parts of the device point cloud (such as the shell, internal components, etc.) and split them into independent key component point clouds. Vectorization algorithms (such as polygon approximation and mesh generation) are used to convert point cloud data into vector data to construct a single vectorization model.

[0127] The technical solution of the embodiment of the present invention includes: obtaining the to-be-processed point cloud data corresponding to the substation, pre-processing the to-be-processed point cloud data to obtain pre-processed point cloud data; processing the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target device in the substation; segmenting the pre-processed point cloud data according to the shape contour and feature vector to obtain the device point cloud data of the target device; identifying the device point cloud data through the target deep learning model to obtain the component point cloud data of the target device, and vectorizing the component point cloud data to generate a single vectorized model of the target device. The technical solution of the present invention can realize the vectorized modeling of single substation equipment, provide a basis for the planning of intelligent patrol routes, and promote the intelligent construction of substations.

[0128] In response to the needs of unmanned duty and autonomous inspection of substations, this paper studies the rapid individual modeling technology of substation equipment, which can associate the model with the ledger, give the model more attribute information, and establish a limited degree vector model library of the substation. It meets the requirements of efficient operation and maintenance work in multiple scenarios within the station, improves the digital management capabilities of substation assets, and empowers intelligent operation and maintenance.

[0129] Embodiment 3

[0130] Figure 7 A schematic diagram of the structure of an electronic device provided for Embodiment 3 of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0131] like Figure 7 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0132] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0133] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a substation single-unit vector modeling method.

[0134] In some embodiments, a substation single vector modeling method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the substation single vector modeling method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute a substation single vector modeling method in any other appropriate manner (for example, by means of firmware).

[0135] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0136] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0137] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0138] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0139] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0140] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0141] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0142] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A substation single-unit vector modeling method, characterized in that: include: Acquire the to-be-processed point cloud data corresponding to the substation, and pre-process the to-be-processed point cloud data to obtain pre-processed point cloud data; Processing the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target equipment in the substation; Segmenting the pre-processed point cloud data according to the outer contour and the feature vector to obtain device point cloud data of the target device; The device point cloud data is identified through a target deep learning model to obtain component point cloud data of the target device, and the component point cloud data is vectorized to generate a monomer vectorized model of the target device.

2. The method according to claim 1, characterized in that The step of obtaining the to-be-processed point cloud data corresponding to the substation includes: According to the layout and equipment distribution of the substation, a scanning path and site settings are formulated, and the point cloud data to be processed corresponding to the substation is obtained based on a high-precision three-dimensional laser scanner.

3. The method according to claim 1, characterized in that The preprocessing of the to-be-processed point cloud data to obtain preprocessed point cloud data includes: Down-sampling is performed on the point cloud data to be processed to obtain simplified point cloud data, and noise reduction is performed on the simplified point cloud data to obtain the pre-processed point cloud data.

4. The method according to claim 3, characterized in that The downsampling process is performed on the point cloud data to be processed to obtain simplified point cloud data, including: Dividing the point cloud data to be processed into voxel grids of fixed size, and determining the non-empty body centroid corresponding to each voxel grid; The point cloud data to be processed in the voxel grid is replaced by the centroid of the non-empty volume to obtain the simplified point cloud data.

5. The method according to claim 1, characterized in that The processing of the pre-processed point cloud data based on the Hough voting algorithm to obtain the shape contour and feature vector corresponding to the target equipment in the substation includes: Extracting local feature points of the preprocessed point cloud data and calculating local descriptors of the local feature points; Matching the feature points in the preset model space with the local feature points based on the local descriptors to determine the target feature points; The outer shape contour and feature vector corresponding to the target device are determined based on the target feature points.

6. The method according to claim 5, characterized in that The matching between the feature points in the preset model space and the local feature points based on the local descriptor to determine the target feature points includes: Determine a descriptor to be used and a shape index value to be used corresponding to the feature point in the preset model space; Calculating the Euclidean distance between the descriptor to be used and the local descriptor, and if the Euclidean distance satisfies a first preset threshold, determining the local feature point as a candidate feature point; A target shape index value of the candidate feature point is determined, and if a difference between the target shape index value and the shape index value to be used satisfies a second preset threshold, the candidate feature point is determined as the target feature point.

7. The method according to claim 5, characterized in that The step of segmenting the pre-processed point cloud data according to the outer contour and the feature vector to obtain device point cloud data of the target device includes: According to the outer contour and the feature vector, the pre-processed point cloud data is segmented using a clustering algorithm or a graph cut algorithm to obtain device point cloud data of the target device.

8. The method according to claim 1, characterized in that The identifying the device point cloud data by the target deep learning model to obtain the component point cloud data of the target device, and performing vectorization processing on the component point cloud data to generate a single vectorized model of the target device includes: Using the KPConv model, different parts of the device point cloud data are identified and split into component point cloud data; The component point cloud data is converted into vector data using a vectorization algorithm, and the monomer vectorization model is constructed based on the vector data.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the substation single vector modeling method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the substation single vector modeling method according to any one of claims 1 to 8 when executed.

Images