Transformer substation safety early warning method and device based on position positioning

Abstract

The invention relates to the technical field of power system automation, discloses a substation safety early warning method and device based on position positioning, and greatly improves the positioning precision by means of deep fusion of point cloud scanning and image technologies. By accurately collecting the second point cloud data and the second image data, the real-time position and posture orientation of the target person in the three-dimensional scene of the transformer substation can be accurately obtained, and high-precision positioning is achieved. According to the high-precision positioning technology, the situation that due to the fact that the precision of a traditional positioning mode is limited, position judgment of target personnel deviates, the target personnel accidentally intrude into a dangerous area, and finally serious potential safety hazards are brought to a transformer substation is avoided. Meanwhile, the movement of articles in the transformer substation is dynamically updated through the first point cloud data collected in real time, so that the accuracy and timeliness of the whole three-dimensional scene information are ensured, and the safe and stable operation of the transformer substation is comprehensively ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of power system automation, and in particular to a substation safety early warning method and device based on position positioning. Background Art

[0002] There are some dangerous areas in the substation that are prohibited for people to approach. At present, when the actual on-site work is carried out in the substation, the safety assurance work is achieved by fencing the on-site work area with safety fences, or by on-site supervision by full-time safety staff. In actual work, such measures have many disadvantages, such as some areas are not suitable for physical fences to facilitate the work of other departments, full-time safety staff cannot supervise in all directions, and the definition of working areas and non-working areas is unclear. These shortcomings have a great impact on the safety assurance work in the substation.

[0003] At present, digitally constructed substations have begun to use high-precision positioning devices combined with digital twin electronic fences to carry out safety management of personnel and engineering vehicles. However, the high-precision positioning devices of substations usually use three high-precision positioning systems: UWB, Bluetooth, and Beidou. The positioning accuracy of the three existing positioning systems is relatively low, making it difficult to accurately judge whether personnel are actually close to dangerous areas in the safety management of substation operations based on digital twin electronic fences. Misjudgments or missed judgments are prone to occur, which brings potential risks to the safe production of substations. More effective high-precision positioning technology is urgently needed to solve this problem. Summary of the invention

[0004] The present invention provides a substation safety early warning method and device based on position positioning, which solves the technical problem that the positioning accuracy of the prior art is insufficient, thereby causing potential safety hazards in the substation.

[0005] A first aspect of the present invention provides a substation safety early warning method based on location positioning, comprising:

[0006] Acquire first point cloud data and first image data of the substation, and construct a three-dimensional scene of the target substation;

[0007] Collecting second point cloud data and second image data of the surrounding environment of the target person in the substation;

[0008] Using the second point cloud data and the second image data for registration to determine the real-time position data and posture orientation data of the target person;

[0009] Acquire the Beidou position data of the target person, and determine the projection position of the target person in the three-dimensional scene of the target substation by combining the real-time position data and the posture orientation data;

[0010] When the projection position is close to the electronic fence of the preset dangerous area, a beeping prompt is given to the target person.

[0011] Optionally, it further includes:

[0012] Perform point cloud clustering analysis on the first point cloud data to obtain a target object clustering set;

[0013] Calculate the centroid position and principal component direction associated with each target object clustering in the target object clustering set;

[0014] Update the three-dimensional scene of the target substation by using the centroid position and principal component direction associated with each target object clustering to obtain an updated three-dimensional scene of the target substation;

[0015] Jump to execute the step of collecting the second point cloud data and the second image data of the environment around the target person in the substation.

[0016] Optionally, the performing point cloud clustering analysis on the first point cloud data to obtain a target object clustering set includes:

[0017] Perform hierarchical processing on the first point cloud data and construct multiple Gaussian mixture models;

[0018] Use the Gaussian mixture models to calculate the background probability values of each spatial point in the first point cloud data;

[0019] When the background probability value is less than a preset background probability threshold, mark the spatial point associated with the background probability value as a foreground point to obtain a foreground point set;

[0020] Perform clustering processing on the foreground point set by using a preset shape feature-based clustering algorithm to obtain an initial object clustering set;

[0021] Perform integrity analysis on each initial object clustering in the initial object clustering set, and retain the initial object clustering with the analysis result being complete to obtain a target object clustering set.

[0022] Optionally, the obtaining the first point cloud data and the first image data of the substation and constructing a three-dimensional scene of the target substation includes:

[0023] Collect the first point cloud data and the first image data of the substation;

[0024] Use the first point cloud data and the first image data to perform scene construction to construct an initial three-dimensional scene of the substation;

[0025] Surround the preset dangerous area in the initial three-dimensional scene of the substation with an electronic fence to obtain a three-dimensional scene of the target substation.

[0026] Optionally, registering using the second point cloud data and the second image data to determine the real-time position data and pose orientation data of the target person includes:

[0027] Performing grid division on the second point cloud data to obtain a plurality of voxel grids;

[0028] Calculating the comprehensive point cloud feature vector corresponding to each voxel grid;

[0029] Performing multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector;

[0030] Based on a plurality of the comprehensive point cloud feature vectors and the image comprehensive feature vector, determining a target matching set;

[0031] Based on the target matching set, determining the real-time position data and pose orientation data of the target person.

[0032] Optionally, performing multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector includes:

[0033] Performing multi-scale decomposition on the second image data to obtain a plurality of Gaussian pyramid images;

[0034] Performing feature extraction on each Gaussian pyramid image to obtain a plurality of depth feature maps;

[0035] Performing upsampling and fusion operations on each depth feature map to obtain an image comprehensive feature vector.

[0036] Optionally, based on a plurality of the comprehensive point cloud feature vectors and the image comprehensive feature vector, determining a target matching set includes:

[0037] Calculating the cosine similarity between a plurality of the comprehensive point cloud feature vectors and the image comprehensive feature vector respectively;

[0038] When the cosine similarity is greater than the preset similarity threshold, then matching the comprehensive point cloud feature vector and the image comprehensive feature vector to obtain an initial matching set;

[0039] Obtaining the spatial corresponding points of each pair of matching pairs in the initial matching set in a preset space;

[0040] Calculating the spatial distance value and angle deviation value corresponding to each spatial corresponding point;

[0041] When the spatial distance value and the angle deviation value do not meet the preset geometric constraint conditions, then removing the corresponding matching pair from the initial matching set to obtain a target matching set.

[0042] Optionally, determining the real-time position data and pose orientation data of the target person based on the target matching set includes:

[0043] Constructing a topological graph by using the matching pairs in the target matching set;

[0044] Taking the minimization of the total cost function of the topological graph as the objective, solving the total cost function to obtain a rough registration transformation matrix;

[0045] Performing fine registration operation on the rough registration transformation matrix to obtain a fine registration transformation matrix;

[0046] Decomposing the fine registration transformation matrix to determine the real-time position data and pose orientation data of the target person.

[0047] Optionally, obtaining the Beidou position data of the target person, and combining the real-time position data and the pose orientation data to determine the projection position of the target person in the three-dimensional scene of the target substation includes:

[0048] Obtaining the Beidou position data of the target person;

[0049] Generating position positioning information of the target person by using the Beidou position data, the real-time position data and the pose orientation data;

[0050] Mapping the position positioning information to the three-dimensional scene of the target substation to obtain the projection position of the target person in the three-dimensional scene of the target substation.

[0051] Optionally, it further includes:

[0052] Real-time collecting the external force data received by the target person;

[0053] When the external force data is greater than a preset external force threshold, an alarm is given.

[0054] A substation safety early warning device based on position positioning provided by the second aspect of the present invention includes:

[0055] A remote positioning and solving system, configured to obtain first point cloud data and first image data of a substation, and construct a three-dimensional scene of a target substation;

[0056] A safety helmet body, configured to collect second point cloud data and second image data of the environment around a target person in the substation;

[0057] The remote positioning and solving system is further configured to perform registration by using the second point cloud data and the second image data to determine the real-time position data and pose orientation data of the target person;

[0058] A Beidou receiving device is used to obtain the Beidou position data of the target person;

[0059] The remote positioning and resolution system is further configured to use the Beidou position data, and in combination with the real-time position data and the attitude and orientation data, to determine the projected position of the target person in the three-dimensional scene of the target substation;

[0060] A buzzer is used to give a buzzer prompt to the target person when the projected position is close to the electronic fence of a preset dangerous area.

[0061] Optionally, the remote positioning and resolution system is further configured to perform point cloud clustering analysis on the first point cloud data to obtain a target object clustering set;

[0062] Calculate the centroid position and the principal component direction associated with each target object clustering in the target object clustering set;

[0063] Use the centroid position and the principal component direction associated with each target object clustering to update the three-dimensional scene of the target substation to obtain an updated three-dimensional scene of the target substation;

[0064] Jump to execute the step of collecting the second point cloud data and the second image data of the environment around the target person in the substation.

[0065] Optionally, a device host is provided inside the safety helmet body;

[0066] The device host is provided with a 5G communication module;

[0067] The Beidou receiving device, the lidar and the micro camera are integrated on the device host;

[0068] The lidar is used to collect the second point cloud data of the environment around the target person in the substation;

[0069] The micro camera is used to collect the second image data of the environment around the target person in the substation;

[0070] The lidar and the micro camera are communicatively connected to the remote positioning and resolution system through the 5G communication module.

[0071] Optionally, a lining shell is provided inside the safety helmet body;

[0072] A central connecting column is fixedly connected between the lining shell and the safety helmet body;

[0073] A rotating cover body is rotatably provided outside the central connecting column;

[0074] The rotating cover body is located between the lining shell and the safety helmet body in the interlayer.

[0075] Optionally, a gear ring is fixedly arranged on the rotating cover body;

[0076] A servo driver is fixedly arranged on the safety helmet body;

[0077] A driving gear is arranged on the servo driver;

[0078] The driving gear meshes with the gear ring;

[0079] An arc-shaped connecting plate is fixedly arranged on the rotating cover body;

[0080] The arc-shaped connecting plate is fixedly installed with the buzzer;

[0081] A sound transmission groove is penetrated and opened on the inner lining shell;

[0082] The servo driver is used to collect the rotation resistance of the rotating cover body in real time as external force data;

[0083] The 5G communication module is used to give an alarm when the external force data is greater than a preset external force threshold.

[0084] Optionally, a top blind hole is opened inside the central connecting column;

[0085] A compressed gas cylinder is fixedly arranged in the top blind hole;

[0086] The air outlet of the compressed gas cylinder is closed by an electrofusion gas plug;

[0087] An air delivery flow channel is opened in the central connecting column.

[0088] Optionally, one end of the air delivery flow channel is communicated with the air outlet of the compressed gas cylinder;

[0089] The other end of the air delivery flow channel is communicated between the gear ring and the central connecting column.

[0090] An electronic device provided in the third aspect of the present invention includes a memory and a processor. When a computer program stored in the memory is executed by the processor, the processor executes the steps of the substation safety warning method based on position positioning as described in any one of the above.

[0091] A computer-readable storage medium provided in the fourth aspect of the present invention stores a computer program thereon. When the computer program is executed, it implements the substation safety warning method based on position positioning as described in any one of the above.

[0092] A computer program product provided by the fifth aspect of the present invention, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer is made to execute the substation safety warning method based on position positioning as described in any one of the above.

[0093] As can be seen from the above technical solutions, the present invention has the following advantages:

[0094] The present invention greatly improves the positioning accuracy by means of the deep integration of point cloud scanning and image technology. By accurately collecting the second point cloud data and the second image data, the real-time position and attitude orientation of the target person in the three-dimensional scene of the substation can be accurately obtained, realizing high-precision positioning. This high-precision positioning technology avoids the situation that the traditional positioning method has limited accuracy, resulting in deviation in the judgment of the target person's position, and then leading to its accidental intrusion into the dangerous area, ultimately bringing serious safety hazards to the substation. At the same time, the dynamic update of the movement of items in the substation is carried out through the first point cloud data collected in real time, ensuring the accuracy and timeliness of the entire three-dimensional scene information, and comprehensively guaranteeing the safe and stable operation of the substation. Description of the Drawings

[0095] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0096] Figure 1 It is a flowchart of the steps of a substation safety warning method based on position positioning provided by an embodiment of the present invention;

[0097] Figure 2 It is a schematic diagram of a substation safety warning device based on position positioning of the present invention.

[0098] Figure 3 It is another perspective schematic diagram of a substation safety warning device based on position positioning of the present invention.

[0099] Figure 4 It is a three-dimensional semi-sectional schematic diagram of a substation safety warning device based on position positioning of the present invention.

[0100] Figure 5 For Figure 4 The enlarged schematic diagram of area A in

[0101] Figure 6 It is a structural block diagram of a computer device provided by an embodiment of the present invention;

[0102] Among them, the meanings of the attached drawing reference numerals are as follows:

[0103] 1. Safety helmet body; 2. Device host; 3. Lidar; 4. Beidou receiving device; 5. Micro camera; 6. Buzzer; 101. Lining shell; 102. Central connecting column; 103. Rotating cover; 104. Gear ring; 105. Driving gear; 106. Servo driver; 107. Arc connecting plate; 108. Sound transmission groove; 109. Top blind hole; 110. Compressed gas cylinder; 111. Electric fusion gas plug; 112. Air supply flow channel. Detailed implementation manners

[0104] An embodiment of the present invention provides a substation safety warning method and device based on position positioning, which determines the safety status of staff in the substation through data positioning, and realizes automatic reminder when approaching a dangerous area, avoiding the inconvenience of setting up enclosures in the prior art, reducing the occupation of human resources for setting up full-time safety supervisors, and improving the safety guarantee ability of the substation.

[0105] In order to make the invention purpose, features, and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0106] Currently, high-precision positioning devices in substations usually adopt three high-precision positioning systems: UWB, Bluetooth, and Beidou. However, all three positioning devices have irreparable defects. First, the UWB positioning device has high input costs, high performance requirements for system positioning data calculation equipment, and limited signal frequency bands for occupying positioning data sources, making it impossible to solve the problem of large-scale use in substations. Second, the Bluetooth positioning accuracy is not high. Third, Beidou positioning cannot be applied indoors or under high equipment, and is greatly affected by factors such as weather.

[0107] Please refer to Figure 1 , Figure 1 , which is a step flowchart of a substation safety warning method based on position positioning provided by an embodiment of the present invention.

[0108] A substation safety warning method based on position positioning provided by the present invention includes:

[0109] S1. Obtain the first point cloud data and the first image data of the substation, and construct a three-dimensional scene of the target substation.

[0110] Further, S1 may include the following sub-steps:

[0111] S11. Collect the first point cloud data and the first image data of the substation.

[0112] S12. Use the first point cloud data and the first image data to construct a scene, and construct an initial 3D scene of the substation.

[0113] S13. Set an electronic fence around the preset dangerous area in the initial 3D scene of the substation to obtain the target 3D scene of the substation.

[0114] In the embodiment of the present invention, first, the first point cloud data and the first image data of the substation are collected. The collection of the first point cloud data this time is to perform high-precision scanning on the first point cloud data of the substation through a professional 3D scanner, and perform 3D modeling on the substation in combination with the obtained first point cloud data; and the collected first image data is used as a texture map to be correspondingly covered on the 3D model to complete the construction of the initial 3D scene of the substation. An electronic fence is set around the dangerous area in the initial 3D scene of the substation as needed to complete the construction of the target 3D scene of the substation.

[0115] S2. Collect the second point cloud data and the second image data of the surrounding environment of the target person in the substation.

[0116] In the embodiment of the present invention, the second point cloud data and the second image data of the surrounding environment of the target person in the substation are scanned. It should be noted that the target person refers to the person in the substation, including but not limited to the staff.

[0117] S3. Use the second point cloud data and the second image data for registration to determine the real-time position data and the attitude and orientation data of the target person.

[0118] Further, S3 may include the following sub-steps:

[0119] S21. Perform grid division on the second point cloud data to obtain a plurality of voxel grids.

[0120] S22. Calculate the comprehensive point cloud feature vector corresponding to each voxel grid.

[0121] In the embodiment of the present invention, a multi-level voxelization method is used to extract the multi-scale geometric features of the second point cloud data. The second point cloud data is divided into voxel grids of different scales. For each scale , the voxel size is , and within each voxel, the statistical features of the points are calculated as the comprehensive point cloud feature vector.

[0122] Let the point in the second point cloud data be within the voxel of scale , and the point set within this voxel is , the calculation point The eigenvector at this scale , which contains the coordinate mean of the in-voxel points , covariance matrix and the relative position vector of the point to the voxel center .

[0123] That is ; in the way of weighted summation, corresponding weights are assigned according to the importance of different scales , then the statistical features of the point are used as the comprehensive feature vector of the point cloud :

[0124] .

[0125] S23. Perform multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector.

[0126] Furthermore, S23 may include the following sub-steps:

[0127] S231. Perform multi-scale decomposition on the second image data to obtain multiple Gaussian pyramid images.

[0128] In the embodiment of the present invention, the Gaussian pyramid is used to perform multi-scale decomposition on the second image data to construct layers of Gaussian pyramid, and the scale factor of the image of each layer of Gaussian pyramid is , where is a fixed scaling ratio, , to obtain multiple Gaussian pyramid images.

[0129] S232. Extract features from each Gaussian pyramid image to obtain multiple depth feature maps.

[0130] In the embodiment of the present invention, on each layer of Gaussian pyramid image, a feature extractor based on a convolutional neural network is used to obtain the depth features of the Gaussian pyramid image. Let the second image data at the layer of Gaussian pyramid, the extracted depth feature map is .

[0131] S233. Perform upsampling and fusion operations on each depth feature map to obtain an image comprehensive feature vector.

[0132] In the embodiment of the present invention, by performing upsampling and fusion on the depth feature maps of different layers, the image comprehensive feature vector of the second image data is obtained .

[0133] S24. Based on multiple point cloud comprehensive feature vectors and image comprehensive feature vectors, determine the target matching set.

[0134] Further, S24 may include the following sub-steps:

[0135] S241. Calculate the cosine similarity between the comprehensive feature vectors of multiple point clouds and the comprehensive feature vector of the image respectively.

[0136] In the embodiment of the present invention, for the comprehensive feature vector of the point cloud and the comprehensive feature vector of the image , a matching method based on distance measurement is adopted. Here, the improved cosine similarity is used as the distance measurement. Let the point cloud feature vector in the comprehensive feature vector of the point cloud be and the image feature vector in the comprehensive feature vector of the image be , and its improved cosine similarity calculation formula is:

[0137] ;

[0138] wherein, is a small positive number used to avoid the case where the denominator is zero.

[0139] S242. When the cosine similarity is greater than the preset similarity threshold, match the comprehensive feature vector of the point cloud and the comprehensive feature vector of the image to obtain an initial matching set.

[0140] In the embodiment of the present invention, by setting the similarity threshold , the feature pairs with cosine similarity higher than this similarity threshold are initially matched to obtain an initial matching set composed of multiple matching pairs .

[0141] Each pair of matching pairs in this matching set represents a potential correspondence between a point in the point cloud space and a point in the image space.

[0142] For example, if the feature vector of the point cloud point is successfully matched with the feature vector of the image point , it indicates that and may be the corresponding representations of the same object or scene point in the two spaces.

[0143] S243. Obtain the spatial corresponding points of each pair of matching pairs in the preset space.

[0144] It should be noted that the preset space includes the point cloud space and the image space.

[0145] In the embodiments of the present invention, the initial matching set provides the preliminary correspondence of points in the point cloud space and the image space. These correspondences are the basis for further screening and optimization. For example, the spatial corresponding points of the matching pairs in the initial matching set , etc., are the direct inputs for subsequent screening of geometric constraint conditions.

[0146] Considering the geometric correspondence between the point cloud and the image, the initial matching set is further screened. For the points in the point cloud space and the corresponding points in the image space , according to their geometric layout in the three-dimensional scene, certain distance and angle constraints should be satisfied.

[0147] S244. Calculate the spatial distance value and the angle deviation value corresponding to each spatial corresponding point.

[0148] In the embodiments of the present invention, let the point cloud point and and their corresponding image points and . Calculate the distance between the point cloud points and the distance between the image points, as well as the angle deviation between the point cloud point and the corresponding image point.

[0149] S245. When the spatial distance value and the angle deviation value do not meet the preset geometric constraint conditions, the corresponding matching pairs are removed from the initial matching set to obtain the target matching set.

[0150] In the embodiments of the present invention, the geometric constraint conditions are defined as follows:

[0151]

[0152] Among them, is the distance constraint threshold, is the angle constraint threshold. Only the matching pairs that meet the above preset geometric constraint conditions are retained in the target matching set .

[0153] S25. Based on the target matching set, determine the real-time position data and the attitude orientation data of the target person.

[0154] Furthermore, S25 may include the following sub-steps:

[0155] S251. Use the matching pairs in the target matching set to construct a topological map.

[0156] In the embodiments of the present invention, the target matching set The point cloud points and image points in it are constructed into a topological graph structure, where the point cloud points and image points are respectively used as the nodes of the graph, and the corresponding relationship between the matching pairs is used as the edges of the graph.

[0157] S252. Taking the minimization of the total cost function of the topological graph as the goal, solve the total cost function to obtain the rough registration transformation matrix.

[0158] In the embodiment of the present invention, define the cost function of the topological graph, comprehensively consider the distance error and geometric consistency error between the matching pairs, and set the edge The distance error of is The geometric consistency error is Then the cost function of the edge is:

[0159]

[0160] Where, and are weight coefficients used to balance the influence of the two errors. By solving the graph optimization problem and minimizing the total cost function of the graph, the rough registration transformation matrix is obtained.

[0161] S253. Perform fine registration operation on the rough registration transformation matrix to obtain the fine registration transformation matrix.

[0162] In the embodiment of the present invention, based on the rough registration transformation matrix obtained by rough registration, the adaptive ICP algorithm is used for fine registration. Let the current point cloud data be The point cloud after rough registration is The target scene point cloud is In each adaptive ICP iteration: First, divide the point cloud into different regions according to the local density information of the point cloud , and set different matching weights for each region. The region with high density has a high matching weight, and the region with low density has a low matching weight; for each point in the point cloud , find the nearest neighbor point in the target scene point cloud . At the same time, adjust the search range and matching priority of the nearest neighbor point according to the matching weight, and update the transformation matrix by solving the following least squares problem: Where, is the matching weight of the point . After multiple iterations until convergence or reaching the maximum number of iterations, the final fine registration transformation matrix is obtained.

[0163] S254. Decompose the fine registration transformation matrix to determine the real-time position data and attitude orientation data of the target person.

[0164] In the embodiment of the present invention, when decomposing the fine registration transformation matrix, this matrix can also be decomposed into rotation and translation parts to determine the real-time position and attitude orientation. Assuming the matrix is a 4×4 matrix, the upper-left 3×3 sub-matrix of the matrix is the rotation matrix, and the rotation matrix describes the attitude orientation of the object in three-dimensional space. It can interpret the attitude in various ways, such as Euler angles or quaternions. Taking Euler angles as an example, the rotation angles of the object around the x, y, and z axes can be calculated through a specific conversion formula from the rotation matrix to Euler angles, so as to determine the attitude orientation data. And the elements in the fourth column of the first three rows of the matrix form the translation vector, and this translation vector represents the displacement of the point cloud or object in three-dimensional space relative to the origin of the reference coordinate system, that is, the real-time position data.

[0165] S4. Obtain the Beidou position data of the target person, and combine the real-time position data and attitude orientation data to determine the projection position of the target person in the three-dimensional scene of the target substation.

[0166] Further, S4 may include the following sub-steps:

[0167] S31. Obtain the Beidou position data of the target person.

[0168] S32. Use the Beidou position data, real-time position data, and attitude orientation data to generate the position location information of the target person.

[0169] In the embodiment of the present invention, the Beidou positioning system is a satellite navigation and positioning system, which can obtain geographical coordinate information such as longitude and latitude of a certain position on the earth through satellite signals. When determining the Beidou position data of the target person, the real-time position data calculated based on the second point cloud data and the second image data is compared with the absolute geographical position coordinate information (i.e., the Beidou position data) provided by the Beidou positioning system. The absolute geographical position coordinate information provided by the Beidou positioning system is used to determine the different regions and the macroscopic positions of different substations where the staff in the substation are located. The real-time position data and attitude orientation data calculated based on the second point cloud data and the second image data are used to accurately determine the real-time position and attitude orientation of the staff in the substation. When combining, only the macroscopic position of the Beidou positioning needs to be obtained.

[0170] S33. Map the position location information onto the three-dimensional scene of the target substation to obtain the projection position of the target person in the three-dimensional scene of the target substation.

[0171] In the embodiment of the present invention, map the position location information onto the three-dimensional scene of the target substation to obtain the projection position of the target person in the three-dimensional scene of the target substation.

[0172] S5. When the projection position is close to the electronic fence of the preset dangerous area, a beeping prompt is given to the target person.

[0173] In the embodiment of the present invention, when the projection mapped in the three-dimensional scene of the substation is close to the electronic fence of the dangerous area, a beeping prompt is given to the target person.

[0174] S6. Perform point cloud clustering analysis on the first point cloud data to obtain a target item clustering set.

[0175] Further, S6 may include the following sub-steps:

[0176] S41. Perform hierarchical processing on the first point cloud data and construct multiple Gaussian mixture models.

[0177] In the embodiment of the present invention, hierarchical processing is performed on the first point cloud data of the target substation three-dimensional scene to establish a multi-layer Gaussian mixture model for background modeling, and the three-dimensional space is divided into layers, and independent are established for the first point cloud data of each layer For each layer , let the point coordinates in the first point cloud data of this layer be , and use from Gaussian distributions to fit the distribution of points within this layer, The probability density function is the same as the Gaussian mixture model in the basic algorithm

[0178]

[0179] where is the parameter set of this layer , represents the mean vector of the th Gaussian distribution in the Gaussian mixture model of the th layer, which describes the central position of this Gaussian distribution in the three-dimensional space. For the first point cloud data, its components correspond to the means in the directions of the three-dimensional coordinates; the parameters of each layer are estimated through the expectation maximization algorithm.

[0180] S42. Use the Gaussian mixture model to calculate the background probability values of each spatial point in the first point cloud data.

[0181] In the embodiment of the present invention, in the subsequently scanned point cloud data, segmentation is performed according to the established multi-layer For a newly scanned spatial point , first determine the level it belongs to , and then calculate the probability of each spatial point belonging to the background of this layer through the above formula .

[0182] S43. When the background probability value is less than the preset background probability threshold, mark the spatial point associated with the background probability value as a foreground point to obtain a foreground point set.

[0183] In the embodiment of the present invention, when the background probability value is less than the preset background probability threshold, mark the spatial point associated with the background probability value as a foreground point to obtain a foreground point set.

[0184] S44. Cluster the foreground point set using a preset shape feature-based clustering algorithm to obtain an initial item clustering set.

[0185] It should be noted that the shape feature-based clustering algorithm includes, but is not limited to, the shape clustering algorithm based on geometric divergence, the clustering algorithm based on multi-scale shape analysis, etc.

[0186] Among them, the clustering algorithm of multi-scale shape analysis is specifically to analyze and extract features of the shape at different scales, and comprehensively use the information of multiple scales to judge the similarity of shapes and perform clustering. For example, in remote sensing image analysis, clustering and recognition of geographical shapes at different scales are considered, taking into account the features of shapes at different resolutions, and can better handle shapes in complex scenes.

[0187] The shape clustering algorithm based on geometric divergence is specifically to use geometric divergence to measure the difference between different shapes, and geometric divergence reflects the deviation degree of two shapes in geometric features. Clustering is performed according to the value of geometric divergence, so that the geometric divergence of shapes within the same cluster is small, and shapes with different geometric features can be effectively distinguished.

[0188] In the embodiment of the present invention, for the set of points marked as foreground, a shape analysis-based clustering algorithm is used for finer division.

[0189] S45. Analyze the integrity of each initial item clustering in the initial item clustering set, and retain the initial item clustering with the analysis result of integrity to obtain a target item clustering set.

[0190] In the embodiment of the present invention, let the initial item clustering set after clustering be , for each initial item clustering , further confirm whether it is a complete item or a part of an item by analyzing its shape features; the integrity analysis specifically includes geometric size measurement, convex hull analysis and symmetry detection, as follows:

[0191] For example, geometric dimension measurement is performed to calculate the length, width, and height dimensions of the point cloud set. For example, for the point cloud clustering of a suspected cuboid item, if the ratio of its length, width, and height conforms to that of a common cuboid item, it may be a complete item; if the dimension of a certain dimension is much smaller than expected, it may be a part of the item.

[0192] For example, convex hull analysis is performed to calculate the convex hull of the point cloud. The convex hull is the smallest convex polyhedron that contains all the point clouds. By comparing the differences between the point cloud and the convex hull, if the gap between the point cloud and the convex hull is large, it may mean that the point cloud is only a part of the item.

[0193] For example, symmetry detection is performed to check whether the point cloud has symmetry. Many items have a certain degree of symmetry, such as the axial symmetry of a cylinder and the central symmetry of a cube. By finding the axis of symmetry or the plane of symmetry and calculating the index of the degree of symmetry, it is determined whether the point cloud conforms to the symmetry characteristics of a complete item. Through comprehensive analysis from multiple aspects.

[0194] Through integrity analysis, the initial item clusters with the analysis result of being complete are retained to obtain the target item cluster set.

[0195] S7. Calculate the centroid position and the principal component direction associated with each target item cluster in the target item cluster set.

[0196] In the embodiment of the present invention, for each target item cluster whose integrity is confirmed , in addition to calculating its centroid position to track its position change in three-dimensional space, its attitude change is also considered. Let , its centroid The calculation formula, that is , this formula calculates the centroid of the target item cluster , in the formula represents the point in the target item cluster , is the number of points in this cluster. The centroid position is obtained by averaging the coordinates of all points to track the position change of the item in three-dimensional space.

[0197] To determine the attitude change of the item, the principal component direction of the item cluster is calculated. Calculating the principal component direction of the item cluster is achieved by performing principal component analysis (PCA) on the point cloud of the item cluster, including subtracting each point in the target item cluster point cloud from the centroid The centered point cloud data is obtained, and the covariance matrix of the centered point cloud data is calculated. The covariance matrix can reflect the correlation between data in each dimension and the data distribution. Perform eigenvalue decomposition (EVD) or singular value decomposition (SVD) on the covariance matrix. Through these two decomposition methods, the eigenvalues and eigenvectors of the covariance matrix can be obtained. The eigenvalues are sorted from largest to smallest, and the corresponding eigenvectors are the principal component vectors. The eigenvector corresponding to the largest eigenvalue is the first principal component direction, which reflects the direction of the largest change in the point cloud data, and so on. In practical applications, usually the eigenvectors corresponding to the first few larger eigenvalues are selected to describe the principal component directions of item clustering, and these principal component vectors can reflect the change in the pose direction of the items.

[0198] By performing principal component analysis on the item clustering point cloud, the principal component vectors are obtained. According to the principal component direction, the principal component vectors are obtained. First, calculate the covariance matrix of the point cloud data, then perform eigenvalue decomposition on the covariance matrix, and finally sort the eigenvalues from largest to smallest. The corresponding eigenvectors are the principal component vectors, which will not be elaborated here. This vector can reflect the change in the pose direction of the items.

[0199] S8. Update the three-dimensional scene of the target substation by using the centroid position and the principal component direction associated with each target item clustering to obtain the updated three-dimensional scene of the target substation.

[0200] In the embodiment of the present invention, by comparing the centroid positions and the principal component directions of the target item clustering at different times, the position and pose information of the items in the three-dimensional scene of the target substation are comprehensively updated to realize the update of the item movement.

[0201] S9. Jump to execute the step of collecting the second point cloud data and the second image data of the environment around the target person in the substation.

[0202] In the embodiment of the present invention, jump to execute the step of collecting the second point cloud data and the second image data of the environment around the target person in the substation.

[0203] Further, the following steps are also included:

[0204] S10. Real-time collect the external force data received by the target person. When the external force data is greater than the preset external force threshold, an alarm is given.

[0205] In the embodiments of the present invention, real-time collection of external force data of a target person refers to relevant information of various forces acting on the target person, such as impact force, pressure, etc. In the scenario of a safety helmet, it is data such as the magnitude of the external force borne when the safety helmet is hit and deformed or pierced by a heavy object. The preset external force threshold refers to a preset standard value, which is determined based on experience, experiments, safety requirements, etc. It represents a safety limit. When the external force data exceeds this limit, it indicates that the target person may be facing a dangerous situation. For example, through the analysis of a large amount of experimental data, it is determined that when the impact force on the safety helmet exceeds a certain specific value, it may cause serious harm to the head of the wearer, and this specific value is the preset external force threshold. Once the external force data is greater than the preset external force threshold, the system will immediately trigger an alarm mechanism.

[0206] A substation safety warning device based on position positioning provided by the present invention includes a remote positioning and resolution system, a safety helmet body 1, a Beidou receiving device 4, and a buzzer 6;

[0207] Both the Beidou receiving device 4 and the buzzer 6 are installed on the safety helmet body 1;

[0208] The safety helmet body 1, the Beidou receiving device 4, and the buzzer 6 are all communicatively connected to the remote positioning and resolution system;

[0209] The remote positioning and resolution system is used to obtain the first point cloud data and the first image data of the substation and construct a three-dimensional scene of the target substation;

[0210] The safety helmet body 1 is used to collect the second point cloud data and the second image data of the surrounding environment of the target person in the substation;

[0211] The remote positioning and resolution system is further used to perform registration using the second point cloud data and the second image data to determine the real-time position data and the attitude and orientation data of the target person;

[0212] The Beidou receiving device 4 is used to obtain the Beidou position data of the target person;

[0213] The remote positioning and resolution system is further used to use the Beidou position data and, in combination with the real-time position data and the attitude and orientation data, determine the projection position of the target person in the three-dimensional scene of the target substation;

[0214] The buzzer 6 is used to give a buzzer prompt to the target person when the projection position is close to the electronic fence of the preset dangerous area.

[0215] Further, the remote positioning and resolution system is further used to perform point cloud clustering analysis on the first point cloud data to obtain a target item clustering set;

[0216] Calculate the centroid positions and principal component directions associated with each target object cluster within the target object cluster set;

[0217] Use the centroid positions and principal component directions associated with each target object cluster to update the 3D scene of the target substation, obtaining the updated 3D scene of the target substation;

[0218] Jump to execute the step of collecting the second point cloud data and the second image data of the environment around the target personnel within the substation.

[0219] Furthermore, the remote positioning and solution system includes:

[0220] A Gaussian mixture model module for performing hierarchical processing on the first point cloud data and constructing multiple Gaussian mixture models;

[0221] A background probability value module for calculating the background probability values of each spatial point within the first point cloud data using the Gaussian mixture model;

[0222] A foreground point set module for, when the background probability value is less than a preset background probability threshold, marking the spatial points associated with the background probability value as foreground points to obtain a foreground point set;

[0223] An initial object cluster set module for performing clustering processing on the foreground point set using a preset shape feature-based clustering algorithm to obtain an initial object cluster set;

[0224] A target object cluster set module for performing integrity analysis on each initial object cluster within the initial object cluster set and retaining the initial object clusters with complete analysis results to obtain a target object cluster set.

[0225] Furthermore, the remote positioning and solution system further includes:

[0226] A data acquisition module for acquiring the first point cloud data and the first image data of the substation;

[0227] An initial substation 3D scene module for performing scene construction using the first point cloud data and the first image data to construct an initial substation 3D scene;

[0228] A target substation 3D scene module for surrounding a preset dangerous area in the initial substation 3D scene with an electronic fence to obtain a target substation 3D scene.

[0229] Furthermore, an internal device host 2 is provided in the safety helmet body 1;

[0230] The device host 2 is provided with a 5G communication module;

[0231] The Beidou receiving device 4, the lidar 3, and the micro camera 5 are integrated on the device host 2;

[0232] A lidar 3 for collecting second point cloud data of the environment around a target person in a substation;

[0233] A micro camera 5 for collecting second image data of the environment around a target person in a substation;

[0234] The lidar 3 and the micro camera 5 are communicatively connected to a remote positioning and solving system through a 5G communication module.

[0235] Furthermore, the remote positioning and solving system further includes:

[0236] A grid division module for dividing the second point cloud data into grids to obtain a plurality of voxel grids;

[0237] A point cloud comprehensive feature vector module for calculating the point cloud comprehensive feature vectors corresponding to each voxel grid;

[0238] A multi-scale feature fusion module for performing multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector;

[0239] A first data processing module for determining a target matching set based on a plurality of point cloud comprehensive feature vectors and the image comprehensive feature vector;

[0240] A second data processing module for determining the real-time position data and the attitude orientation data of the target person based on the target matching set.

[0241] Furthermore, the multi-scale feature fusion module includes:

[0242] A multi-scale decomposition sub-module for performing multi-scale decomposition on the second image data to obtain a plurality of Gaussian pyramid images;

[0243] A feature extraction sub-module for extracting features from each Gaussian pyramid image to obtain a plurality of depth feature maps;

[0244] An upsampling and fusion operation sub-module for performing upsampling and fusion operations on each depth feature map to obtain an image comprehensive feature vector.

[0245] Furthermore, the first data processing module includes:

[0246] A cosine similarity sub-module for calculating the cosine similarities between a plurality of point cloud comprehensive feature vectors and the image comprehensive feature vector respectively;

[0247] An initial matching set sub-module for matching the point cloud comprehensive feature vector and the image comprehensive feature vector when the cosine similarity is greater than a preset similarity threshold to obtain an initial matching set;

[0248] A spatial corresponding point sub-module, configured to obtain spatial corresponding points of each pair of matching pairs in the initial matching set in a preset space;

[0249] A spatial distance value sub-module, configured to calculate spatial distance values and angular deviation values corresponding to the respective spatial corresponding points;

[0250] A target matching set sub-module, configured to, when the spatial distance value and the angular deviation value do not satisfy a preset geometric constraint condition, remove the corresponding matching pair from the initial matching set to obtain a target matching set.

[0251] Further, the second data processing module includes:

[0252] A topology graph sub-module, configured to construct a topology graph by using the matching pairs in the target matching set;

[0253] A rough registration transformation matrix sub-module, configured to solve the total cost function with the goal of minimizing the total cost function of the topology graph to obtain a rough registration transformation matrix;

[0254] A fine registration operation sub-module, configured to perform a fine registration operation on the rough registration transformation matrix to obtain a fine registration transformation matrix;

[0255] A decomposition sub-module, configured to decompose the fine registration transformation matrix to determine real-time position data and attitude orientation data of the target person.

[0256] Further, the remote positioning and calculation system further includes:

[0257] A position positioning information module, configured to generate position positioning information of the target person by using Beidou position data, real-time position data, and attitude orientation data;

[0258] A projected position module, configured to map the position positioning information onto a three-dimensional scene of the target substation to obtain a projected position of the target person in the three-dimensional scene of the target substation.

[0259] Further, a lining shell 101 is arranged inside the safety helmet body 1, and a central connecting column 102 is fixedly connected between the lining shell 101 and the safety helmet body 1;

[0260] A rotating cover body 103 is rotatably arranged outside the central connecting column 102, and the rotating cover body 103 is located between the lining shell 101 and the safety helmet body 1;

[0261] A gear ring 104 is fixedly arranged on the rotating cover body 103;

[0262] A servo driver 106 is fixedly arranged on the safety helmet body 1, and a driving gear 105 is arranged on the servo driver 106;

[0263] The driving gear 105 meshes with the gear ring 104;

[0264] An arc-shaped connecting plate 107 is fixedly arranged on the rotating cover body 103, and the arc-shaped connecting plate 107 is fixedly installed with the buzzer 6;

[0265] A sound-transmitting groove 108 is penetratingly opened on the inner lining housing 101;

[0266] A top blind hole 109 is opened inside the central connecting column 102, and a compressed gas cylinder 110 is fixedly arranged in the top blind hole 109;

[0267] The air outlet of the compressed gas cylinder 110 is closed by an electrofusion gas plug 111, and an air delivery flow channel 112 is opened in the central connecting column 102;

[0268] One end of the air delivery flow channel 112 is communicated with the air outlet of the compressed gas cylinder 110, and the other end is communicated between the gear ring 104 and the central connecting column 102;

[0269] When it is detected that the rotation of the rotating cover body 103 is restricted, the electrofusion gas plug 111 is electrified, so that the electrofusion gas plug 111 melts and releases the compressed air in the compressed gas cylinder 110;

[0270] The servo driver 106 is used to collect the rotation resistance of the rotating cover body 103 in real time as external force data;

[0271] The 5G communication module is used to give an alarm when the external force data is greater than a preset external force threshold.

[0272] Please refer to Figures 2 - 5 , a substation safety warning device based on position positioning, including: a lidar 3, which is used to scan and collect the point cloud data of the surrounding environment of the staff in the substation; a micro camera 5, which is used to scan and collect the image of the surrounding environment of the staff in the substation;

[0273] The Beidou receiving device 4 is used to compare and further determine the position data of the staff in combination with Beidou positioning. The significance of the establishment of the Beidou receiving device 4 is to distinguish the indoor substations with similar environments. Through rough positioning by the Beidou receiving device 4, it can be determined which indoor substation the staff is in, and then the fine position can be determined;

[0274] A buzzer 6, which is used to give a buzzer prompt to the staff; and a safety helmet body 1, which the staff wears through the safety helmet body 1. A device host 2 is fixedly arranged on the safety helmet body 1. The lidar 3, the Beidou receiving device 4 and the micro camera 5 are integrated on the device host 2. A 5G communication module is arranged in the device host 2. The lidar 3 and the micro camera 5 upload the collected data to the remote positioning and resolution system through the 5G communication module. When the remote positioning and resolution system detects an electronic fence near a dangerous area, it feeds back an alarm signal through the 5G communication module, so that the buzzer 6 works.

[0275] Inside the safety helmet body 1, there is a lining shell 101. A central connecting column 102 is fixedly connected between the lining shell 101 and the safety helmet body 1. A rotating cover body 103 is rotatably arranged outside the central connecting column 102, and the rotating cover body 103 is located between the sandwich layers of the lining shell 101 and the safety helmet body 1.

[0276] A gear ring 104 is fixedly arranged on the rotating cover body 103. A servo driver 106 is fixedly arranged on the safety helmet body 1. A driving gear 105 is arranged on the servo driver 106, and the driving gear 105 meshes with the gear ring 104. An arc-shaped connecting plate 107 is fixedly arranged on the rotating cover body 103, and the arc-shaped connecting plate 107 is fixedly installed with a buzzer 6. A sound transmission groove 108 is penetrated and opened on the lining shell 101.

[0277] A top blind hole 109 is opened inside the central connecting column 102. A compressed gas cylinder 110 is fixedly arranged in the top blind hole 109. The air outlet of the compressed gas cylinder 110 is closed by an electric melting gas plug 111. An air delivery flow channel 112 is opened in the central connecting column 102.

[0278] One end of the air delivery flow channel 112 is communicated with the air outlet of the compressed gas cylinder 110, and the other end is communicated between the gear ring 104 and the central connecting column 102; when it is detected that the rotation of the rotating cover body 103 is restricted, the electric melting gas plug 111 is electrified, so that the electric melting gas plug 111 melts and releases the compressed air in the compressed gas cylinder 110.

[0279] In the embodiment of the present invention, the second point cloud data and the second image data of the surrounding environment of the target person in the substation are scanned, and the scanned second point cloud data and second image data are transmitted to the remote positioning and solving system through the 5G communication module. The remote positioning and solving system is a remote server platform, which has a 5G communication module and a computing function;

[0280] The remote positioning and solving system calculates the real-time position data and attitude orientation data mapped by the target person in the three-dimensional scene of the target substation according to the second point cloud data and the second image data in combination with the algorithm, and at the same time can use the first point cloud data to update the item model in the three-dimensional scene of the target substation in real time; combined with the Beidou positioning for comparison to further determine the position positioning information of the target person.

[0281] During use, when the remote positioning and solving system detects an electronic fence near the dangerous area, it feeds back an alarm signal through the 5G communication module. The servo driver 106 drives the driving gear 105 to rotate, driving the rotating cover body 103 to rotate, so that the buzzer 6 rotates to the azimuth where the electronic fence is located. Subsequently, the buzzer 6 alarms. At this time, the staff can determine the direction of the dangerous area according to the prompt azimuth of the buzzer 6.

[0282] The rotating cover 103 conducts a timed inspection and rotates one full circle. When the safety helmet body 1 is hit by a heavy object and deformed, or punctured, the inspection rotation of the rotating cover 103 will be hindered. The servo driver 106 can detect the change in rotational resistance for the servo motor system. When resistance is detected, the electrofusion gas plug 111 is energized, causing the electrofusion gas plug 111 to melt and release the compressed air in the compressed gas cylinder 110. The compressed air flow enters between the gear ring 104 and the central connecting column 102 to wash and clean the gear ring 104. After the cleaning is completed, if the resistance still exists, an alarm is sent through the 5G communication module, enabling the cloud supervision personnel to timely check the status of the corresponding staff member and reducing the probability of the target person suffering a head injury and missing the first aid time.

[0283] Due to the opening of windows such as the sound transmission groove 108, dust and foreign objects will inevitably enter the gear ring 104 during the use of the device. By pre-washing and cleaning the gear ring 104, the false alarm probability of the system can be significantly reduced.

[0284] The present invention can determine the status of the target person in the substation through point cloud scanning combined with image technology. When approaching a dangerous area, it realizes automatic reminder, avoiding the inconvenience of setting up fences in traditional technologies and reducing the human resource occupation of setting up full-time safety supervisors. Compared with technologies such as UWB, Bluetooth, and Beidou, it has the characteristics of high precision, low cost, and wide coverage, and can be accurate to the posture orientation of the target person.

[0285] The remote positioning and solution system in the present invention can determine the real-time position data and posture orientation data in the three-dimensional scene of the corresponding target substation based on the currently collected second point cloud data and second image data, so as to timely remind the target person of approaching the electronic fence area; at the same time, it can use the currently collected first point cloud data to update the movement of items in the three-dimensional scene of the target substation in real time, ensuring the accuracy of the three-dimensional scene of the target substation in the remote positioning and solution system.

[0286] In the substation safety warning device based on position positioning in the present invention, through the cooperation of structures such as the rotating cover 103, the arc-shaped connecting plate 107, and the servo driver 106, the position of the buzzer 6 can be changed, so that when the buzzer 6 emits a reminder buzzer, the warning sound comes from the location where the electronic fence is set up, enabling the target person to perceive the location of the dangerous area.

[0287] Through the cooperation of structures such as the provided compressed gas cylinder 110 and the air supply flow channel 112, when it is detected that the rotation of the rotary cover 103 is blocked, the gear ring 104 and the driving gear 105 can be cleaned by air pressure flushing. When the blockage still exists, it indicates that the safety helmet body 1 may be punctured at this time, and remote alarm is carried out, so that the cloud supervision personnel can view the status of the corresponding target personnel in time, reducing the probability that the target personnel's head is injured and missing the first aid time; the above-mentioned active cleaning function can reduce the probability of false alarms caused by foreign object jamming.

[0288] Please refer to Figure 6 , Figure 6 which is a structural block diagram of a computer device provided by an embodiment of the present invention.

[0289] An electronic device according to an embodiment of the present invention, the electronic device includes: a memory 201 and a processor 202, and a computer program is stored in the memory 201; when the computer program is executed by the processor 202, the processor 202 is caused to execute the substation safety warning method based on position positioning as described in any of the above embodiments.

[0290] The memory 201 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. The memory 201 has a storage space 203 for program codes 213 for executing any method steps in the above methods. For example, the storage space 203 for program codes may include respective program codes 213 for implementing various steps in the above methods. These program codes may be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact discs (CDs), memory cards, or floppy disks. The program codes may be compressed in an appropriate form. When these codes are run by a computing processing device, the computing processing device is caused to execute the respective steps in the method described above. These program codes may be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact discs (CDs), memory cards, or floppy disks. The program codes may be compressed in an appropriate form. When these codes are run by a computing processing device, the computing processing device is caused to execute the respective steps in the substation safety warning method based on position positioning described above.

[0291] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the substation safety warning method based on position positioning as described in any of the above embodiments.

[0292] An embodiment of the present invention also provides a computer program product. The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer is caused to execute the substation safety warning method based on position positioning as described in any of the above embodiments.

[0293] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0294] In several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, indirect couplings or communication connections of devices or units, and can be in electrical, mechanical, or other forms.

[0295] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0296] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0297] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0298] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A substation safety early warning method based on location positioning, characterized in that: include: Acquire first point cloud data and first image data of the substation, and construct a three-dimensional scene of the target substation; Collecting second point cloud data and second image data of the surrounding environment of the target person in the substation; Using the second point cloud data and the second image data for registration to determine the real-time position data and posture orientation data of the target person; Acquire the Beidou position data of the target person, and determine the projection position of the target person in the three-dimensional scene of the target substation by combining the real-time position data and the posture orientation data; When the projection position is close to the electronic fence of the preset dangerous area, a buzzer prompt is given to the target person.

2. The substation safety early warning method based on location positioning according to claim 1 is characterized in that: Also includes: Performing point cloud clustering analysis on the first point cloud data to obtain a target object cluster set; Calculating the centroid position and principal component direction of each target item cluster association in the target item cluster set; The target substation three-dimensional scene is updated using the centroid position and the principal component direction associated with each cluster of the target objects to obtain an updated target substation three-dimensional scene; Jump to the step of collecting the second point cloud data and the second image data of the surrounding environment of the target person in the substation.

3. The substation safety early warning method based on location positioning according to claim 2 is characterized in that: The performing point cloud cluster analysis on the first point cloud data to obtain a target object cluster set includes: Performing layered processing on the first point cloud data and constructing multiple Gaussian mixture models; Calculating the background probability value of each spatial point in the first point cloud data using the Gaussian mixture model; When the background probability value is less than a preset background probability threshold, the spatial point associated with the background probability value is marked as a foreground point to obtain a foreground point set; Using a preset clustering algorithm based on shape features to cluster the foreground point set to obtain an initial object cluster set; An integrity analysis is performed on each initial item cluster in the initial item cluster set, and the initial item clusters with complete analysis results are retained to obtain a target item cluster set.

4. The substation safety early warning method based on location positioning according to claim 1 is characterized in that: The step of acquiring the first point cloud data and the first image data of the substation and constructing a three-dimensional scene of the target substation includes: Collecting first point cloud data and first image data of the substation; Using the first point cloud data and the first image data to construct a scene, to construct an initial substation three-dimensional scene; An electronic fence is set to surround the preset dangerous area in the initial substation three-dimensional scene to obtain a target substation three-dimensional scene.

5. The substation safety early warning method based on location positioning according to claim 1 is characterized in that: The registering the second point cloud data and the second image data to determine the real-time position data and posture orientation data of the target person includes: Meshing the second point cloud data to obtain a plurality of voxel grids; Calculating the point cloud comprehensive feature vector corresponding to each of the voxel grids; Performing multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector; Determining a target matching set based on a plurality of the point cloud comprehensive feature vectors and the image comprehensive feature vector; Based on the target matching set, the real-time position data and posture orientation data of the target person are determined.

6. The substation safety early warning method based on location positioning according to claim 5 is characterized in that: The performing multi-scale feature fusion on the second image data to obtain an image comprehensive feature vector includes: Performing multi-scale decomposition on the second image data to obtain a plurality of Gaussian pyramid images; Performing feature extraction on each of the Gaussian pyramid images to obtain a plurality of depth feature images; Upsampling and fusion operations are performed on each of the depth feature maps to obtain an image comprehensive feature vector.

7. The substation safety early warning method based on location positioning according to claim 5 is characterized in that: The determining of a target matching set based on the plurality of point cloud comprehensive feature vectors and the image comprehensive feature vector comprises: Calculating cosine similarities between a plurality of the point cloud comprehensive feature vectors and the image comprehensive feature vector respectively; When the cosine similarity is greater than the preset similarity threshold, the point cloud comprehensive feature vector is matched with the image comprehensive feature vector to obtain an initial matching set; Obtaining spatial corresponding points of each pair of matching pairs in the initial matching set in a preset space; Calculate the spatial distance value and the angle deviation value corresponding to each of the spatial corresponding points; When the spatial distance value and the angle deviation value do not satisfy the preset geometric constraint condition, the corresponding matching pair is removed from the initial matching set to obtain a target matching set.

8. The substation safety early warning method based on location positioning according to claim 5 is characterized in that: The step of determining the real-time position data and posture orientation data of the target person based on the target matching set includes: constructing a topological graph using the matching pairs in the target matching set; With the goal of minimizing the total cost function of the topological graph, the total cost function is solved to obtain a coarse registration transformation matrix; Performing a fine registration operation on the coarse registration transformation matrix to obtain a fine registration transformation matrix; The precise registration transformation matrix is ​​decomposed to determine the real-time position data and posture orientation data of the target person.

9. The substation safety early warning method based on location positioning according to claim 1 is characterized in that: The obtaining of the Beidou position data of the target person, and combining the real-time position data and the attitude and orientation data to determine the projection position of the target person in the three-dimensional scene of the target substation, includes: Obtaining Beidou location data of the target person; Generate the position information of the target person by using the Beidou position data, the real-time position data and the attitude and orientation data; The position positioning information is mapped into the three-dimensional scene of the target substation to obtain the projection position of the target person in the three-dimensional scene of the target substation.

10. The substation safety early warning method based on location positioning according to claim 1, characterized in that: Also includes: Collecting the external force data on the target person in real time; When the external force data is greater than a preset external force threshold, an alarm is triggered.

11. A substation safety early warning device based on location positioning, based on the substation safety early warning method based on location positioning according to any one of claims 1 to 10, characterized in that: include: A remote positioning and solving system is used to obtain the first point cloud data and the first image data of the substation and construct a three-dimensional scene of the target substation; A safety helmet body, used to collect second point cloud data and second image data of the surrounding environment of the target person in the substation; The remote positioning and solving system is further used to use the second point cloud data and the second image data for registration to determine the real-time position data and posture orientation data of the target person; A Beidou receiving device, used to obtain Beidou location data of the target person; The remote positioning and solving system is further used to use the Beidou position data and combine the real-time position data and the attitude and orientation data to determine the projection position of the target person in the three-dimensional scene of the target substation; The buzzer is used to buzz the target person when the projection position is close to the electronic fence of the preset dangerous area.

12. The substation safety early warning device based on location positioning according to claim 11, characterized in that: The remote positioning and solving system is further used to perform point cloud clustering analysis on the first point cloud data to obtain a target object cluster set; Calculating the centroid position and principal component direction of each target item cluster association in the target item cluster set; The target substation three-dimensional scene is updated using the centroid position and the principal component direction associated with each cluster of the target objects to obtain an updated target substation three-dimensional scene; Jump to the step of collecting the second point cloud data and the second image data of the surrounding environment of the target person in the substation.

13. The substation safety early warning device based on location positioning according to claim 11, characterized in that: The helmet body is provided with an internal device host; The device host is provided with a 5G communication module; The Beidou receiving device, laser radar and micro camera are integrated on the device host; The laser radar is used to collect second point cloud data of the surrounding environment of the target person in the substation; The micro camera is used to collect second image data of the surrounding environment of the target person in the substation; The laser radar and the micro camera are communicatively connected to the remote positioning solution system through the 5G communication module.

14. The substation safety early warning device based on location positioning according to claim 13, characterized in that: The helmet body is provided with an inner lining shell; A central connecting column is connected and fixed between the liner shell and the helmet body; The outer rotation of the central connecting column is provided with a rotating cover; The rotating cover body is located between the inner lining shell and the sandwich layer of the safety helmet body.

15. The substation safety early warning device based on location positioning according to claim 14, characterized in that: A gear ring is fixedly arranged on the rotating cover body; A servo driver is fixedly arranged on the safety helmet body; The servo driver is provided with a driving gear; The driving gear and the ring gear are meshed with each other; An arc-shaped connecting plate is fixedly arranged on the rotating cover body; The arc-shaped connecting plate is fixedly installed with the buzzer; The inner lining shell is provided with a sound-permeable groove; The servo driver is used to collect the rotational resistance of the rotating cover in real time as external force data; The 5G communication module is used to alarm when the external force data is greater than a preset external force threshold.

16. The substation safety early warning device based on location positioning according to claim 15, characterized in that: A top blind hole is provided inside the central connecting column; A compressed gas cylinder is fixedly arranged in the top blind hole; The gas outlet of the compressed gas cylinder is sealed by an electric fused gas plug; An air supply passage is provided in the central connecting column.

17. The substation safety early warning device based on location positioning according to claim 16, characterized in that: One end of the air delivery channel is in communication with the air outlet of the compressed gas cylinder; The other end of the air supply passage is communicated with the gear ring and the center connecting column.

18. An electronic device, characterized in that: It comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the substation safety warning method based on location positioning as described in any one of claims 1-10.

19. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the substation safety early warning method based on location positioning as described in any one of claims 1 to 10 is implemented.

20. A computer program product, characterized in that The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer executes the substation safety warning method based on location positioning as described in any one of claims 1-10.

Images