A near-electricity alarm method and device

Abstract

The present invention discloses a method and device for providing a near-electrical alarm. The method comprises: calculating a complete point cloud dataset of charged equipment based on laser radar point cloud data and video image data of an on-site operation area; constructing a spatial structure hierarchy of charged equipment based on the point cloud features in the complete point cloud dataset; monitoring the state of the point cloud data within the spatial structure hierarchy, and issuing a near-electrical alarm when, based on the monitoring results, it is determined that a person is approaching a charged equipment. The embodiments of the present invention can improve the efficiency of near-electrical alarms.

Description

Technical Field

[0001] The present invention relates to the field of electric shock prevention control technology, and in particular to a method and device for providing a near-electricity alarm. Background Art

[0002] Today, the power system has become the most crucial energy supply and transmission system. Transformer and distribution lines and equipment of varying specifications are ubiquitous in society. However, due to age, overload, and other factors, initially installed power equipment no longer meets basic requirements for safe operation and requires ongoing inspection, maintenance, and replacement. Furthermore, power equipment construction and renovation often presents challenges such as limited working space, complex on-site conditions, and a large number of construction personnel, making them difficult to manage. These issues pose safety risks to power operations, with the most prominent being electric shocks caused by workers accidentally touching or entering live compartments during construction. Once such accidents occur, they not only result in direct personal and economic losses, such as casualties and equipment damage, but also in significant indirect economic losses due to power outages.

[0003] Traditional near-electricity warning methods generally use high-voltage magnetic field induction to provide early warnings. However, in complex environments such as distribution rooms and substations, these methods suffer from low efficiency, scattered detection quality, and a single method. This makes it impossible to visually display the location of strong electric fields to operators and accurately measure the distance to charged objects, resulting in low warning efficiency. Currently, online monitoring methods such as lidar and cameras combine laser point clouds with image recognition to identify personnel locations and provide early warnings of proximity to charged objects. However, on-site monitoring devices require real-time identification and calculation, which leads to high energy consumption and complex calculations. This results in the inability to accurately and continuously monitor on-site personnel behavior during time-consuming, long-term continuous operations, resulting in low proximity warning efficiency.

[0004] As can be seen from the above, the existing method for detecting the proximity of electric shock has low detection accuracy in the process of detecting the positions of the operator and the charged object, resulting in low efficiency of the proximity of electric shock. Summary of the Invention

[0005] The embodiments of the present invention provide a method and device for a near-electrical alarm, which effectively improve the efficiency of the near-electrical alarm.

[0006] A first aspect of an embodiment of the present application provides a near-electricity alarm method, comprising:

[0007] Based on the LiDAR point cloud data and video image data of the on-site operation area, a complete point cloud dataset of the charged equipment is calculated;

[0008] Construct the spatial structure hierarchy of the charged body equipment based on the point cloud features in the complete point cloud dataset of the charged body equipment;

[0009] Monitor the point cloud data status within the spatial structure level, and issue a proximity alarm when it is determined based on the monitoring results that someone is approaching a charged device.

[0010] In one possible implementation of the first aspect, a complete point cloud dataset of the charged device is calculated based on the lidar point cloud data and video image data of the on-site operation area, specifically:

[0011] Based on the laser radar point cloud data and video image data of the on-site operation area, a three-dimensional point cloud original data set of the charged body area in the on-site operation area is obtained;

[0012] Density clustering is performed on the original three-dimensional point cloud data set of the charged body area to obtain a complete point cloud data set of the charged body equipment.

[0013] In a possible implementation of the first aspect, density clustering is performed on the original three-dimensional point cloud dataset of the charged body area to calculate a complete point cloud dataset of the charged body device, specifically:

[0014] Obtaining point cloud features of marked positions based on a three-dimensional point cloud original data set of the charged body area; wherein the marked positions are used to represent positions of charged body devices;

[0015] After feature extraction based on the point cloud features of the marked positions, density clustering is performed on the point cloud data within the preset neighborhood according to the extraction results to obtain a complete point cloud dataset of the charged equipment.

[0016] In a possible implementation of the first aspect, a spatial structure hierarchy of the charged body device is constructed based on point cloud features in a complete point cloud dataset of the charged body device, specifically:

[0017] According to the point cloud features in the complete point cloud data set of the charged body equipment, the corresponding normal vector is calculated;

[0018] According to the normal vector and the preset safety distance, the data set corresponding to the preset safety distance is calculated;

[0019] Based on the data set corresponding to the preset safety distance, the spatial structure hierarchy of the charged equipment is constructed.

[0020] In one possible implementation of the first aspect, the point cloud data status within the spatial structure hierarchy is monitored, and when it is determined based on the monitoring results that a person is approaching a charged device, a near-charge alarm is issued, specifically:

[0021] Monitor the state of point cloud data within the spatial structure hierarchy. When the state of point cloud data within the spatial structure hierarchy fluctuates and there is dense, continuous directional movement, it is determined that an intruding object is approaching the charged device.

[0022] Real-time video image data is recognized through a convolutional neural network algorithm. When the intruding object is determined to be a human based on the recognition results, it is determined that someone is approaching a charged device and a proximity alarm is issued.

[0023] In a possible implementation of the first aspect, the lidar point cloud data of the on-site operation area is collected by a lidar; and the video image data of the on-site operation area is collected by a surveillance camera.

[0024] In a possible implementation of the first aspect, the method further includes:

[0025] Performing noise reduction processing on spatial three-dimensional point cloud data corresponding to different spatial structure levels to generate a first processing result;

[0026] Performing hierarchical segmentation and cropping processing on the first processing result according to a preset safety distance, filtering out irrelevant point cloud data outside the spatial structure hierarchy, and generating a second processing result;

[0027] Record the static point cloud features of the second processing result.

[0028] In one possible implementation of the first aspect, determining whether the state of point cloud data within the spatial structure level fluctuates and whether there is continuous directional motion of high-density point clouds is specifically as follows:

[0029] The point cloud features within the spatial structure hierarchy are compared with the static point cloud features. Based on the comparison results, it is determined that the point cloud data state within the spatial structure hierarchy fluctuates and that there is continuous directional motion of high-density point clouds.

[0030] In a possible implementation manner of the first aspect, the point cloud features include: point cloud quantity data, point cloud density data, point cloud distribution data, and point cloud range data.

[0031] A second aspect of an embodiment of the present application provides a near-electricity alarm device, comprising: a calculation module, a construction module, and an alarm module;

[0032] The calculation module is used to calculate the complete point cloud data set of the charged equipment based on the lidar point cloud data and video image data of the on-site operation area;

[0033] The construction module is used to construct the spatial structure hierarchy of the charged body device based on the point cloud features in the complete point cloud data set of the charged body device;

[0034] The alarm module is used to monitor the status of point cloud data within the spatial structure level. When it is determined based on the monitoring results that someone is approaching a charged device, a proximity alarm is issued.

[0035] Compared with the existing technology, the embodiments of the present invention provide a near-electricity alarm method and device, which include: calculating a complete point cloud data set of charged equipment based on the lidar point cloud data and video image data of the on-site operation area; constructing a spatial structure hierarchy of the charged equipment based on the point cloud features in the complete point cloud data set of the charged equipment; monitoring the point cloud data status within the spatial structure hierarchy, and issuing a near-electricity alarm when it is determined that someone is approaching the charged equipment based on the monitoring results.

[0036] Its beneficial effect is that after the embodiment of the present invention calculates the complete point cloud data set of the charged body equipment based on the laser radar point cloud data and video image data of the on-site operation area, it constructs the spatial structure hierarchy of the charged body equipment based on the point cloud features in the complete point cloud data set of the charged body equipment, and finally monitors the point cloud data status within the spatial structure hierarchy. When it is determined that someone is approaching the charged body equipment based on the monitoring results, a near-electricity alarm is issued. The embodiment of the present invention performs a near-electricity alarm based on laser point cloud status monitoring, and only monitors the point cloud data status within the spatial structure hierarchy around the charged body. After identification and detection, a near-electricity alarm is issued, avoiding the problem of low alarm efficiency caused by the real-time tracking and real-time measurement of the operating personnel in the existing technology. The embodiment of the present invention can save the computing pressure of the edge computing unit to the maximum extent, without the need for real-time calculation, and improves the detection accuracy in the process of detecting the position of the operating personnel and the charged body, thereby improving the efficiency of the near-electricity alarm. At the same time, it can save energy and reduce consumption and effectively extend the monitoring time.

[0037] Furthermore, the embodiments of the present invention are not limited to power operation environments. The principles of this method can be applied to mobile deployments within local area networks. Unlike other fixed-mount monitoring and measurement technologies, which require multi-layered analysis of complex surrounding environments, this method is highly portable and enables full, transparent, and visualized near-power warning monitoring.

[0038] Finally, the embodiment of the present invention obtains complete three-dimensional point cloud data of the charged body equipment by manually marking the charged body equipment, automatically divides the spatial hierarchical structure around the charged body for inductive alarm, and the overall operation is simple and easy to use, which is convenient for training and promotion. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a flow chart of a method for providing a near-electricity alarm according to an embodiment of the present invention;

[0040] Figure 2 This is a schematic structural diagram of an overall system for near-electricity alarm provided by an embodiment of the present invention;

[0041] Figure 3 The figure is a schematic structural diagram of a near-electricity warning device provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0043] Reference Figure 1 , is a flow chart of a method for providing a near-electricity alarm according to an embodiment of the present invention, including S101-S103:

[0044] S101: Calculate and obtain a complete point cloud data set of the charged equipment based on the LiDAR point cloud data and video image data of the on-site operation area.

[0045] The laser radar point cloud data of the on-site operation area is collected by the laser radar, and the video image data of the on-site operation area is collected by the surveillance camera. Specifically, the laser radar and the surveillance camera are aimed at the on-site operation area to obtain the laser radar point cloud data and video image data of the on-site operation area, and the laser radar point cloud data and video image data are matched and fused in the edge computing unit. The fused data is sent to the mobile terminal receiving device in the form of a video image for display, realizing the visualization of the operation scene monitoring.

[0046] In this embodiment, the complete point cloud data set of the charged device is calculated based on the laser radar point cloud data and video image data of the on-site operation area, specifically:

[0047] Acquire a three-dimensional point cloud original data set of the charged body area in the on-site operation area based on the laser radar point cloud data and video image data of the on-site operation area;

[0048] Density clustering is performed on the original three-dimensional point cloud data set of the charged body area to calculate and obtain a complete point cloud data set of the charged body device.

[0049] In a specific embodiment, the density clustering process is performed on the original three-dimensional point cloud data set of the charged body area to calculate and obtain the complete point cloud data set of the charged body device, specifically:

[0050] Obtaining point cloud features of marked positions based on the three-dimensional point cloud original data set of the charged body area; wherein the marked positions are used to represent the positions of the charged body devices;

[0051] After feature extraction is performed based on the point cloud features of the marked positions, density clustering is performed on the point cloud data within a preset neighborhood based on the extraction results to obtain a complete point cloud data set for the charged body device.

[0052] Specifically, the mobile terminal receives the display screen of the device, and according to the actual distribution of charged objects in the on-site operation area, clicks on the display screen to select the core position of the charged object equipment, and obtains the original 3D point cloud data set of the marked charged object area and the object plane area (that is, the original 3D point cloud data set of the charged object area in the on-site operation area) through the matching mapping relationship between the video image data and the 3D point cloud data. s :

[0053] P s {(x1,y1,z1),(x2,y2,z2),…(x n ,y n , z n )};

[0054] Obtain the original 3D point cloud dataset P of the charged body area in the field operation area s , according to P s The point cloud features (including color, material, shape, size, etc.) of the marked positions are extracted, and based on the extraction results, the point cloud data in the preset neighborhood is subjected to density clustering (DBSCAN) processing (setting the density threshold M, selecting P s Get the points in the preset neighborhood ε), and obtain the complete point cloud data set of the charged body equipment from the plane area to the three-dimensional space area (that is, the complete point cloud data set of the charged body equipment) P a :

[0055] P a {(x1,y1,z1),(x2,y2,z2),…(x k ,y k , z k )};

[0056] S102: Constructing a spatial structure hierarchy of the charged body device based on point cloud features in the complete point cloud data set of the charged body device.

[0057] In this embodiment, the spatial structure hierarchy of the charged body device is constructed based on the point cloud features in the complete point cloud data set of the charged body device, specifically:

[0058] Calculating corresponding normal vectors according to point cloud features in a complete point cloud data set of the charged body device;

[0059] According to the normal vector and the preset safety distance, a data set corresponding to the preset safety distance is calculated;

[0060] According to the data set corresponding to the preset safety distance, the spatial structure hierarchy of the charged body equipment is constructed.

[0061] Specifically: Based on the complete point cloud dataset P of the charged body equipment a The point cloud features in the data are used to construct the spatial structure hierarchy of the charged body equipment. The preset safety distances include: the first safety distance d, the second safety distance 2d, and the third safety distance 4d. d is the minimum standard for the safety distance between the staff and the charged body; 2d is 2 times the minimum standard for the safety distance; 4d is 4 times the minimum standard for the safety distance; the use of a hierarchical structure can realize the near-electricity alarm at different levels of distance between the staff and the charged body, and prevent the staff from electric shock to the greatest extent by early warning. Specifically for the dataset P a The steps for expanding the data points outward by distances d, 2d, and 4d are as follows:

[0062] (1) Calculate the data set P a The normal vector of a point in , for each point there exists a plane cosine:

[0063] cosα i ·x i +cosβ i ·y i +cosγ i ·z i +p i =0;

[0064] Among them, cosα i 、cosβ i 、cozy i is a point on the plane (x i ,y i , z i ) direction cosine of the normal vector at |p i | is the distance from the origin to the plane;

[0065] (2)Point(x i ,y i , z i ) The point at a distance d outward in the direction of the normal vector is:

[0066] (x i +d·cosα i ,y i +d·cosβ i , z i +d·cosγ i );

[0067] (3) According to (2), the dataset P a The data set at the distance d in the normal vector direction is obtained for each point in the

[0068] (4) Accordingly, the data set at a distance of 2d in the normal vector direction (i.e., the data set corresponding to the second safety distance 2d) P a "for:

[0069] {(x i +2d·cosα i ,y i +2d·cosβ i , z i +2d·cosγ i ), i=1, 2, 3...k};

[0070] (5) Accordingly, the data set at a distance of 4d in the normal vector direction (i.e., the data set corresponding to the third safety distance 4d) P a "'for:

[0071] {(x i +4d·cosα i ,y i +4d·cosβ i , z i +4d·cosγ i ), i=1, 2, 3, ... k};

[0072] Among them, (x i ,y i , z i )∈P a .

[0073] S103: Monitor the point cloud data status within the spatial structure level, and issue a proximity alarm when it is determined based on the monitoring results that someone is approaching a charged device.

[0074] In this embodiment, the point cloud data status within the spatial structure level is monitored, and when it is determined based on the monitoring results that someone is approaching the charged device, a near-electricity alarm is issued, specifically:

[0075] monitoring the state of the point cloud data within the spatial structure hierarchy, and determining that an intruding object is approaching the charged device when the state of the point cloud data within the spatial structure hierarchy fluctuates and there is intensive and continuous directional movement;

[0076] The real-time video image data is identified by a convolutional neural network algorithm. When the intruding object is determined to be a human being based on the identification result, it is determined that someone is approaching the charged device and a proximity alarm is issued.

[0077] In this embodiment, it also includes:

[0078] Performing noise reduction processing on the spatial three-dimensional point cloud data corresponding to different spatial structure levels to generate a first processing result;

[0079] Performing hierarchical segmentation and cropping processing on the first processing result according to a preset safety distance, filtering out irrelevant point cloud data outside the spatial structure hierarchy, and generating a second processing result;

[0080] Record the static point cloud features of the second processing result.

[0081] Specifically:

[0082] The point cloud data in the spatial range d closest to the charged body is named S1:

[0083] S1{(x′1,y′1,z′1), (x′2,y′2,z′2),…(x′k,y′ k , z′ n )};

[0084] The point cloud data within the 2D space close to the charged body is named S2:

[0085] S2{(x″1,y″1,z″1), (x″2,y″2,z″2),…(x″ k , y″ k , z″ n )};

[0086] The point cloud data within the 4D space close to the charged body is named S3:

[0087] S3{(x″′1, y″′1, z″′1), (x″′2, y″′2, z″′2),…(x″′ k , y″′ k , z″′ n )};

[0088] The static point cloud features of the S1-S3 point cloud datasets (mainly including the number, density, distribution, range, etc. of point clouds) are recorded and stored as constants in the edge computing unit.

[0089] In a specific embodiment, it is determined that the state of the point cloud data within the spatial structure level fluctuates and there is continuous directional movement of high-density point clouds, specifically:

[0090] The point cloud features within the spatial structure level are compared with the static point cloud features, and based on the comparison results, it is determined whether the point cloud data state within the spatial structure level fluctuates and whether there is continuous directional motion of high-density point clouds. The continuous directional motion of high-density point clouds can be distinguished from discrete, low-frequency point clouds.

[0091] The LiDAR monitors changes in the point cloud state within the three levels S1-S3. When it senses fluctuations in the static point cloud data within the surrounding spatial structure, and detects multiple consecutive frames of high-density point cloud with continuous directional motion, it sends a recognition request to the edge computing unit's image recognition service. Upon receiving the recognition service request, the edge computing unit uses synchronized surveillance video images to identify the scene within the image and uses a convolutional neural network (CNN) algorithm to identify intruding objects. If a person is detected as the intruding object, the detection result is sent to a mobile receiving device based on the spatial structure level surrounding the intruding charged object, triggering an alarm to alert the on-site work supervisor and promptly remind workers to maintain a safe distance while working to prevent electric shock.

[0092] The point cloud features include: point cloud quantity data, point cloud density data, point cloud distribution data and point cloud range data.

[0093] In a preferred embodiment, the near-point warning method protected by this application is composed of two parts: a mobile integrated monitoring device composed of a laser radar, a monitoring camera, and an edge computing terminal, and a mobile terminal receiving device. Please refer to Figure 2 , Figure 2 It is a structural diagram of an overall system for near-electricity alarm provided by an embodiment of the present invention, including: a three-dimensional point cloud 201 of charged equipment, a mobile integrated monitoring device 202 and a mobile receiving device 203.

[0094] The mobile integrated monitoring device and the mobile receiving device use Wi-Fi signals for local data transmission. The LiDAR within the mobile integrated monitoring device communicates with the edge computing terminal using a standard RJ45 network cable, while the surveillance camera uses a USB data cable for data transmission. Due to the different fields of view and imaging principles of the laser radar and surveillance camera, the mobile integrated monitoring device requires alignment of the 3D point cloud data with the field of view of the RGB pixel format to achieve overlap and establish a mapping relationship. The mobile integrated monitoring device is installed and deployed by workers in the power operation environment using a tripod or other support structure. The LiDAR and surveillance camera must be oriented toward the power operation area to ensure full coverage of the core operation area. The surveillance camera captures images of the power operation area and transmits the relevant images to the mobile receiving device via the Wi-Fi signal of the edge computing unit.

[0095] The on-site supervisor holds a mobile receiving device, confirms the camera feed of the work area, and then selects the live devices in the feed on the receiving device's display. The mobile receiving device, based on the pixel count and resolution selected by the supervisor, transmits the relevant data back to the edge computing unit via Wi-Fi.

[0096] The edge computing unit obtains the set of pixels selected from the display screen, and finds the three-dimensional point cloud plane data set corresponding to the set of selected pixels through the mapping relationship between the three-dimensional point cloud data and the RGB plane pixel data. The edge computing unit traverses the three-dimensional point cloud plane data of the selected charged body device, extracts the point cloud features in the set, and uses the density clustering (DBSCAN) algorithm to cluster the point clouds of adjacent similar points to obtain the complete point cloud data set S of the selected charged body device in three-dimensional space. a .

[0097] The three-dimensional point cloud dataset S of the charged body equipment a Perform noise reduction and filtering to convert the distance data set S a The spatial range of 0.35m from the outermost point cloud is defined as the first-level spatial structure, which is the minimum standard for the safe distance between workers and live objects in the power safety work regulations; the spatial range within 0.7m from the outermost point cloud of the dataset Ta is defined as the second-level spatial structure, which is twice the first-level spatial structure; the spatial range within 1.4m from the outermost point cloud of the dataset Ta is defined as the third-level spatial structure, which is four times the first-level spatial structure.

[0098] The edge computing unit performs noise reduction and hierarchical segmentation on the point cloud data in each spatial structure around the three-dimensional point cloud of the charged body equipment, cuts out the background outside the three-level spatial structure and other point cloud data that are not related to the near-electric alarm, and records the static point cloud features of the three-level spatial structure (mainly including the number, density, distribution, range, etc. of point clouds), and stores them as constants in the edge computing unit.

[0099] The three-level spatial static point cloud features around the charged device are obtained by the LiDAR and transmitted to the edge computing unit for calculation. The calculation steps are as follows:

[0100] (1) First, cut the point cloud outside the field of view.

[0101] (2) Then the point cloud outside the third-level spatial structure is cut again.

[0102] (3) Compare the point cloud quantity, density, distribution and other data within the three-level spatial structure hierarchy with the static point cloud features.

[0103] (4) When the comparison calculation finds that there is dense, continuous, and directional movement within the three-level spatial structure, an identification request is sent to the identification service of the edge computing unit. When no relevant point cloud data changes are found, return to step (2) of this step and continue monitoring and comparison.

[0104] (5) The recognition service calls the public human recognition data set to recognize the surveillance camera images of 10 consecutive frames before and after the current key frame. When more than 50% of the recognition results of 10 consecutive frames are human, the recognition results and the spatial result level where they appear are sent to the mobile receiving device; when the recognition rate of 10 consecutive frames is less than 50%, the recognition results are ignored and the process returns to step (2) of this step to continue monitoring and comparison.

[0105] When the mobile receiving device receives the recognition result sent by the edge computing unit, a corresponding level of alarm screen will be displayed on the screen, and a corresponding sound prompt will be issued to remind the on-site work supervisor to pay attention to the status of people around the charged object to prevent electric shock.

[0106] It can be seen that the present invention discloses a real-time monitoring method and system for foreign object intrusion on railway tracks based on three-dimensional point cloud data. The method includes installing a laser radar near the railway track to perform background detection on the monitoring area, obtain background three-dimensional point cloud data, and transmit it to a computing terminal; at the computing terminal, using the background three-dimensional point cloud data to construct a background three-dimensional point cloud model; the laser radar performs real-time detection on the monitoring area, obtains real-time three-dimensional point cloud data, and transmits it to the computing terminal for real-time processing to obtain a real-time three-dimensional point cloud data set; the real-time three-dimensional point cloud data set is compared and processed with the background three-dimensional point cloud model to obtain foreign object point cloud data, which is dynamically modeled and identified and updated. This method has high recognition accuracy, is easy to implement, and has high value for promotion and application.

[0107] For further explanation of the proximity alarm device, please refer to Figure 3 , Figure 3 3 is a schematic structural diagram of a near-electricity alarm device provided by an embodiment of the present invention, comprising: a calculation module 301 , a construction module 302 and an alarm module 303 .

[0108] The calculation module 301 is used to calculate a complete point cloud data set of the charged body equipment based on the laser radar point cloud data and video image data of the on-site operation area;

[0109] The construction module 302 is used to construct a spatial structure hierarchy of the charged body device based on the point cloud features in the complete point cloud data set of the charged body device;

[0110] The alarm module 303 is used to monitor the point cloud data status within the spatial structure level, and issue a near-electricity alarm when it is determined based on the monitoring result that someone is approaching the charged device.

[0111] In this embodiment, the complete point cloud data set of the charged device is calculated based on the laser radar point cloud data and video image data of the on-site operation area, specifically:

[0112] Acquire a three-dimensional point cloud original data set of the charged body area in the on-site operation area based on the laser radar point cloud data and video image data of the on-site operation area;

[0113] Density clustering is performed on the original three-dimensional point cloud data set of the charged body area to calculate and obtain a complete point cloud data set of the charged body device.

[0114] In this embodiment, the density clustering process is performed on the original three-dimensional point cloud data set of the charged body area to calculate and obtain the complete point cloud data set of the charged body device, specifically:

[0115] Obtaining point cloud features of marked positions based on the three-dimensional point cloud original data set of the charged body area; wherein the marked positions are used to represent the positions of the charged body devices;

[0116] After feature extraction is performed based on the point cloud features of the marked positions, density clustering is performed on the point cloud data within a preset neighborhood based on the extraction results to obtain a complete point cloud data set for the charged body device.

[0117] In a specific embodiment, the spatial structure hierarchy of the charged body device is constructed based on the point cloud features in the complete point cloud data set of the charged body device, specifically:

[0118] Calculating corresponding normal vectors according to point cloud features in a complete point cloud data set of the charged body device;

[0119] According to the normal vector and the preset safety distance, a data set corresponding to the preset safety distance is calculated;

[0120] According to the data set corresponding to the preset safety distance, the spatial structure hierarchy of the charged body equipment is constructed.

[0121] In a specific embodiment, the point cloud data status within the spatial structure level is monitored, and when it is determined based on the monitoring results that someone is approaching the charged device, a near-electricity alarm is issued, specifically:

[0122] monitoring the state of the point cloud data within the spatial structure hierarchy, and determining that an intruding object is approaching the charged device when the state of the point cloud data within the spatial structure hierarchy fluctuates and there is intensive and continuous directional movement;

[0123] The real-time video image data is identified by a convolutional neural network algorithm. When the intruding object is determined to be a human being based on the identification result, it is determined that someone is approaching the charged device and a proximity alarm is issued.

[0124] In this embodiment, the laser radar point cloud data of the on-site operation area is collected by a laser radar; and the video image data of the on-site operation area is collected by a surveillance camera.

[0125] In this embodiment, it also includes:

[0126] Performing noise reduction processing on the spatial three-dimensional point cloud data corresponding to different spatial structure levels to generate a first processing result;

[0127] Performing hierarchical segmentation and cropping processing on the first processing result according to a preset safety distance, filtering out irrelevant point cloud data outside the spatial structure hierarchy, and generating a second processing result;

[0128] Record the static point cloud features of the second processing result.

[0129] In a specific embodiment, it is determined that the state of the point cloud data within the spatial structure level fluctuates and there is continuous directional movement of high-density point clouds, specifically:

[0130] The point cloud features within the spatial structure level are compared with the static point cloud features, and based on the comparison results, it is determined whether the state of the point cloud data within the spatial structure level fluctuates and whether there is continuous directional movement of high-density point clouds.

[0131] In a specific embodiment, the point cloud features include: point cloud quantity data, point cloud density data, point cloud distribution data and point cloud range data.

[0132] In an embodiment of the present invention, a calculation module is used to calculate a complete point cloud data set of a charged body device based on the lidar point cloud data and video image data of the on-site operation area; a construction module is used to construct a spatial structure hierarchy of the charged body device based on the point cloud features in the complete point cloud data set of the charged body device; an alarm module is used to monitor the status of the point cloud data in the spatial structure hierarchy, and when it is determined that someone is approaching the charged body device based on the monitoring results, a proximity alarm is issued.

[0133] After calculating the complete point cloud data set of the charged body equipment based on the laser radar point cloud data and video image data of the on-site operation area, the embodiment of the present invention constructs the spatial structure hierarchy of the charged body equipment based on the point cloud features in the complete point cloud data set of the charged body equipment, and finally monitors the point cloud data status within the spatial structure hierarchy. When it is determined that someone is approaching the charged body equipment based on the monitoring results, a near-electricity alarm is issued. The embodiment of the present invention performs a near-electricity alarm based on laser point cloud status monitoring, and only monitors the point cloud data status within the spatial structure hierarchy around the charged body. After identification and detection, a near-electricity alarm is issued, avoiding the problem of low alarm efficiency caused by the real-time tracking and real-time measurement of the operating personnel in the existing technology. The embodiment of the present invention can save the computing pressure of the edge computing unit to the greatest extent, without the need for real-time calculation, and improves the detection accuracy in the process of detecting the position of the operating personnel and the charged body, thereby improving the efficiency of the near-electricity alarm. At the same time, it can save energy and reduce consumption, and effectively extend the monitoring time.

[0134] Furthermore, the embodiments of the present invention are not limited to power operation environments. The principles of this method can be applied to mobile deployments within local area networks. Unlike other fixed-mount monitoring and measurement technologies, which require multi-layered analysis of complex surrounding environments, this method is highly portable and enables full, transparent, and visualized near-power warning monitoring.

[0135] Finally, the embodiment of the present invention obtains complete three-dimensional point cloud data of the charged body equipment by manually marking the charged body equipment, automatically divides the spatial hierarchical structure around the charged body for inductive alarm, and the overall operation is simple and easy to use, which is convenient for training and promotion.

[0136] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for providing a near-electricity warning, characterized in that: include: Based on the LiDAR point cloud data and video image data of the on-site operation area, a complete point cloud dataset of the charged equipment is calculated; Calculating corresponding normal vectors according to point cloud features in a complete point cloud data set of the charged body device; According to the normal vector and the preset safety distance, a data set corresponding to the preset safety distance is calculated; wherein the preset safety distance includes: a first safety distance d, a second safety distance 2d, and a third safety distance 4d, where d is the minimum standard safety distance between the worker and the charged device; Constructing a spatial structure hierarchy of the charged body equipment according to the data set corresponding to the preset safety distance; The point cloud data status within the spatial structure level is monitored, and when it is determined based on the monitoring results that someone is approaching the charged device, a near-electricity alarm is issued.

2. The method for near-electricity warning according to claim 1, characterized in that: The complete point cloud data set of the charged equipment is calculated based on the LiDAR point cloud data and video image data of the on-site operation area, specifically: Acquire a three-dimensional point cloud original data set of the charged body area in the on-site operation area based on the laser radar point cloud data and video image data of the on-site operation area; Density clustering is performed on the original three-dimensional point cloud data set of the charged body area to calculate and obtain a complete point cloud data set of the charged body device.

3. The method for near-electricity alarm according to claim 2, characterized in that: The density clustering process is performed on the original three-dimensional point cloud data set of the charged body area to calculate the complete point cloud data set of the charged body device, specifically: Obtaining point cloud features of marked positions based on the three-dimensional point cloud original data set of the charged body area; wherein the marked positions are used to represent the positions of the charged body devices; After feature extraction is performed based on the point cloud features of the marked positions, density clustering is performed on the point cloud data within a preset neighborhood based on the extraction results to obtain a complete point cloud data set for the charged body device.

4. The method for near-electricity warning according to claim 3, characterized in that: The point cloud data status within the spatial structure level is monitored, and when it is determined based on the monitoring results that someone is approaching the charged device, a near-electricity alarm is issued, specifically: monitoring the state of the point cloud data within the spatial structure hierarchy, and determining that an intruding object is approaching the charged device when the state of the point cloud data within the spatial structure hierarchy fluctuates and there is intensive and continuous directional movement; Real-time video image data is recognized through a convolutional neural network algorithm. When the intruding object is determined to be a human according to the recognition result, it is determined that someone is approaching the charged device and a proximity alarm is issued.

5. The method for near-electricity warning according to claim 4, characterized in that: The laser radar point cloud data of the on-site operation area is collected by the laser radar; the video image data of the on-site operation area is collected by the monitoring camera.

6. The method for near-electricity warning according to claim 5, characterized in that: Also includes: Performing noise reduction processing on the spatial three-dimensional point cloud data corresponding to different spatial structure levels to generate a first processing result; Performing hierarchical segmentation and cropping processing on the first processing result according to a preset safety distance, filtering out irrelevant point cloud data outside the spatial structure hierarchy, and generating a second processing result; Record the static point cloud features of the second processing result.

7. The method for near-electricity warning according to claim 6, characterized in that: Determining whether the state of the point cloud data within the spatial structure level fluctuates and whether there is continuous directional movement of high-density point clouds is specifically: The point cloud features within the spatial structure level are compared with the static point cloud features, and based on the comparison results, it is determined whether the state of the point cloud data within the spatial structure level fluctuates and whether there is continuous directional movement of high-density point clouds.

8. The method for near-electricity warning according to claim 7, characterized in that: The point cloud features include: point cloud quantity data, point cloud density data, point cloud distribution data and point cloud range data.

9. A near-electric alarm device, characterized in that: include: Calculation module, construction module and alarm module; The calculation module is used to calculate the complete point cloud data set of the charged body equipment based on the laser radar point cloud data and video image data of the on-site operation area; The construction module is used to calculate the corresponding normal vector based on the point cloud features in the complete point cloud data set of the charged body equipment; based on the normal vector and in combination with the preset safety distance, calculate the data set corresponding to the preset safety distance; wherein the preset safety distance includes: a first safety distance d, a second safety distance 2d and a third safety distance 4d, d being the minimum standard safety distance between the worker and the charged body equipment; and construct the spatial structure hierarchy of the charged body equipment based on the data set corresponding to the preset safety distance; The alarm module is used to monitor the point cloud data status within the spatial structure level, and when it is determined based on the monitoring results that someone is approaching the charged device, an electric proximity alarm is issued.

Images