An external damage prevention monitoring system and method for live-line work

Abstract

The application discloses a live working anti-external damage monitoring system and method, relates to the technical field of power operation safety protection, and the system comprises a laser radar, an electric field sensor, a main control module and a strong magnetic force holder. The bottom of the strong magnetic force holder is provided with a strong magnetic force adsorption disc, the top is provided with a mounting platform, the inside is provided with a horizontal shaft motor and a vertical shaft motor, and the side is provided with a data interface and a power supply interface. The laser radar is connected with the strong magnetic force holder through the mounting platform and is used for collecting laser point cloud data. The electric field sensor is connected with the strong magnetic force holder through the mounting platform and is used for collecting electric field fluctuation data. The main control module is connected with the strong magnetic force holder through a shielding cable and is used for acquiring operation area data and performing anti-external damage monitoring according to the operation area data. Through the cooperative matching of the laser radar, the electric field sensor, the main control module and the strong magnetic force holder, effective anti-external damage monitoring of the operation area during live working can be realized, and the safety of live working is improved.

Description

Technical Field

[0001] This application relates to the field of power operation safety protection technology, and in particular to a monitoring system and method for preventing external damage during live-line work. Background Technology

[0002] Safe and efficient operation and maintenance in the power industry are crucial to social stability and development. The maintenance, repair, and upgrading of transmission and transformation lines are important factors in ensuring the stable operation of the power grid. Live-line work is highly risky due to its involvement with high-voltage power equipment. Substations, as core nodes for power transmission and distribution, have dense internal equipment and complex working environments. Especially in high-risk work scenarios such as crane lifting and equipment maintenance, workers are prone to collisions due to limited visibility and obstructions. Even slight carelessness can lead to serious safety accidents such as electric shock, falls, or equipment damage. In severe weather conditions such as rain, snow, and fog, visibility and safety of the working environment are greatly reduced, further exacerbating the risks and making it difficult for workers to accurately assess the safety conditions of the work area and avoid potential external damage in a timely manner.

[0003] Currently, the main method for monitoring external damage during live-line work is through manual monitoring by safety officers. However, human energy is limited, making it impossible to comprehensively monitor all aspects of the work area. Workers also find it difficult to intuitively grasp the spatial relationship between themselves, their tools, live equipment, and obstacles, which can easily lead to safety accidents due to misjudgment. Existing technologies also include monitoring external damage during live-line work by installing cameras at fixed locations. However, this method is time-consuming to install, has high environmental requirements, and cannot achieve comprehensive, blind-spot-free monitoring coverage. There are monitoring blind spots, and the cameras are affected by lighting conditions; excessively bright or dark environments can lead to recognition difficulties.

[0004] Therefore, how to effectively monitor the work area to prevent external damage during live-line work and improve the safety of live-line work has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] This application provides a live-line work protection monitoring system and method, which can effectively monitor the work area for external damage during live-line work, thereby improving the safety of live-line work.

[0006] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, this application provides a live-line work protection monitoring system for external damage, comprising: LiDAR, electric field sensor, main control module, and strong magnetic gimbal; The strong magnetic gimbal is equipped with a strong magnetic adsorption plate at the bottom for mounting the strong magnetic gimbal in the work area, an installation platform at the top, two rotating motors (horizontal axis motor and vertical axis motor) inside, and a data interface and a power supply interface on the side. The lidar is connected to the high-magnetic gimbal via the mounting platform and is used to collect laser point cloud data of the work area. The electric field sensor is connected to the strong magnetic gimbal via the mounting platform and is used to collect electric field fluctuation data generated by electrical equipment in the work area. The main control module is connected to the data interface and power supply interface of the strong magnetic gimbal via a shielded cable, and is used to acquire the work area data collected by the lidar and the electric field sensor, and to perform external damage prevention monitoring based on the work area data.

[0007] Optionally, the system further includes: A camera connected to the high-magnetic gimbal via the mounting platform is used to collect image data of the work area; The main control module is also used to acquire image data of the work area collected by the camera, and to perform external damage prevention monitoring based on the image data and the work area data.

[0008] Optionally, the main control module is specifically used for: Acquire laser point cloud data of the work area collected by the lidar and electric field fluctuation data of the work area collected by the electric field sensor; The laser point cloud data and the electric field fluctuation data are fused using spatiotemporal registration and weighted Bayesian decision-making to obtain a first data fusion result. A three-dimensional model is constructed based on the first data fusion result using the voxel method; The three-dimensional model is used for external damage monitoring.

[0009] Optionally, the main control module is specifically used for: Acquire laser point cloud data of the work area collected by the lidar, electric field fluctuation data of the work area collected by the electric field sensor, and image data of the work area collected by the camera; The laser point cloud data and the image data are fused to obtain a fused feature vector; The fusion feature vector and the electric field fluctuation data are fused using a weighted Bayesian decision method to obtain a second data fusion result. A three-dimensional model is constructed based on the second data fusion result using the voxel method; The three-dimensional model is used for external damage monitoring.

[0010] Optionally, the process by which the main control module uses the three-dimensional model to monitor for external damage includes: In the three-dimensional model, different levels of danger zones are divided according to the location and voltage level of charged bodies and / or obstacles; Real-time monitoring of the positions of workers and tools within the 3D model; Calculate the distances between workers and tools and the boundaries of each hazardous area; When the distance is less than a preset threshold, a warning signal of the corresponding level is triggered.

[0011] Optionally, the process by which the main control module fuses the laser point cloud data and the electric field fluctuation data using spatiotemporal registration and weighted Bayesian decision-making to obtain a first data fusion result includes: The electric field fluctuation data is interpolated using a linear interpolation method so that the interpolated electric field fluctuation data corresponds one-to-one with the laser point cloud data in time. Transform the laser point cloud data in the lidar coordinate system to the global coordinate system; The spatiotemporally registered laser point cloud data and electric field fluctuation data are fused using a weighted Bayesian decision method to obtain the object category classification result for each laser point cloud data. The discrimination result is used as the first data fusion result.

[0012] Optionally, the main control module fuses the spatiotemporally registered laser point cloud data and electric field fluctuation data using a weighted Bayesian decision method to obtain the discrimination result of the object category corresponding to each laser point cloud data, including: For each laser point cloud data, local geometric features are extracted to obtain the lidar feature vector; Feature extraction is performed on the electric field fluctuation data to obtain the electric field intensity data; Using a Gaussian mixture model, class-conditional probability modeling is performed based on the lidar feature vector and the electric field intensity data, respectively, to obtain the class-conditional probability of lidar features and the class-conditional probability of electric field features. Based on the class conditional probabilities of the lidar features and the electric field features, calculate the joint posterior probability of each lidar point cloud data belonging to each object category. Based on the joint posterior probability, the object category corresponding to each laser point cloud data is determined according to the maximum posterior probability criterion.

[0013] Optionally, the main control module performs feature fusion on the laser point cloud data and the image data to obtain a fused feature vector, including: Extracting lidar geometric features from lidar point cloud data; Extracting camera visual features from image data; Based on the geometric features of the LiDAR and the visual features of the camera, a multi-scale attention fusion network is adopted. Features of different granularities are extracted through multi-scale convolution, and the contribution weights of the geometric features of the LiDAR and the visual features of the camera are dynamically adjusted through the attention mechanism to obtain the fused feature vector.

[0014] Secondly, this application provides a method for monitoring external damage during live-line working, including: The main control module controls the electric field sensor to continuously collect electric field fluctuation data in the working area; The main control module controls the strong magnetic gimbal to drive the lidar and electric field sensor to rotate at a preset speed. The main control module determines whether there are any anomalies in the work area based on the electric field fluctuation data; When an anomaly occurs in the work area, the main control module controls the lidar to collect laser point cloud data of the work area; The main control module performs external damage prevention monitoring based on the electric field fluctuation data and the laser point cloud data.

[0015] Optionally, if there are no anomalies in the work area and the lidar does not detect any changes in the lidar point cloud data of the work area for a period exceeding a preset duration, the method further includes: The main control module controls the lidar to enter sleep mode; The main control module controls the strong magnetic gimbal to rotate to the position where the electric field sensor can collect the working area over the maximum range, and then stops rotating. The main control module controls itself to enter a low-power operation mode.

[0016] As can be seen from the above technical solution, this application has at least the following beneficial effects: This application provides a live-line work protection and external damage monitoring system composed of a lidar, an electric field sensor, a main control module, and a strong magnetic gimbal. By installing a strong magnetic adsorption plate at the bottom of the strong magnetic gimbal for mounting it to the work area, the system can directly adsorb the strong magnetic gimbal onto the surface of metal equipment such as the vehicle body and substation metal supports in the work area without bolt fixation. This solves the problems of long installation time and high requirements for the installation environment of fixed-position cameras in the prior art, improving the system's installation convenience and environmental adaptability, and enabling rapid deployment in different work scenarios. The strong magnetic gimbal has two rotating motors, a horizontal axis motor and a vertical axis motor, which drive the lidar and electric field sensor to rotate synchronously in the horizontal and vertical directions, achieving omnidirectional coverage of the work area. This solves the problem of fixed cameras in the prior art being unable to achieve omnidirectional, blind-spot-free monitoring coverage and having monitoring blind spots, ensuring that the entire work area is within the monitoring range. By using lidar to collect laser point cloud data of the work area, the lidar's working principle is not limited by light conditions and can work normally in excessively bright or dark environments. It possesses the ability to resist interference from rain, snow, fog, and haze, solving the problems of existing technologies where cameras are limited by lighting conditions and have poor visibility in bad weather, leading to difficulties in identification. In addition, laser point cloud data can provide three-dimensional spatial coordinates and contour information of objects in the work area, enabling workers to intuitively grasp the spatial positional relationship between themselves, their tools, energized equipment, and obstacles, solving the problem of safety accidents caused by workers misjudging spatial positions in existing technologies. By using electric field sensors to collect electric field fluctuation data generated by energized equipment in the work area, it can directly perceive the electric field characteristics of energized equipment and distinguish between energized and non-energized bodies, solving the problem that manual monitoring and cameras cannot effectively identify energized bodies in existing technologies, and reducing the risk of misjudgment caused by the inability to distinguish between energized equipment and non-energized obstacles. Through the main control module, it automatically acquires work area data collected by lidar and electric field sensors and performs external damage prevention monitoring, enabling continuous and uninterrupted comprehensive monitoring of the work area, solving the problem that safety officers have limited energy for manual monitoring and cannot comprehensively monitor all situations in the work area in existing technologies, reducing manual intervention and improving the continuity and comprehensiveness of monitoring. By coordinating lidar, electric field sensors, main control modules, and strong magnetic gimbals, the system can effectively address the problems of manual monitoring and fixed camera monitoring in existing technologies, thereby improving the safety of live-line work by enabling effective monitoring of the work area to prevent external damage.

[0017] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0018] Figure 1 A schematic diagram of a live-line working external damage prevention monitoring system provided in this application embodiment; Figure 2 This is a schematic diagram of another live-line working external damage prevention monitoring system provided in an embodiment of this application; Figure 3 This application provides a schematic flowchart of a live-line working protection monitoring method for external damage. Detailed Implementation The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0019] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0020] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first: Live-line working refers to the inspection, maintenance, modification, and hoisting activities carried out on high-voltage and ultra-high-voltage power equipment (such as transmission lines, transformers, circuit breakers, disconnect switches, etc.) while they are in operation and energized. Because workers need to be in close contact with high-voltage live parts, there are extremely high safety risks such as electric shock, arc burns, and falls from heights.

[0021] External damage refers to damage caused by external forces. In the power industry, it specifically refers to damage to power equipment or power outages caused by factors outside the power system. In this application, "external damage" mainly refers to damage caused by human forces during live-line work, such as work machinery (cranes, aerial work platforms, excavators, etc.) colliding with live equipment or transmission lines; workers or tools accidentally entering the safe distance for live equipment; foreign objects (such as ropes, tools, building materials) falling and touching live conductors during the work; and temporary obstacles (such as trees, temporary buildings) encroaching on the safe work area.

[0022] Current technologies for monitoring the prevention of external damage during live-line work mainly rely on fixed-point cameras or manual on-site monitoring. However, these methods are cumbersome to install, have blind spots, are greatly affected by ambient light, and cannot distinguish between live and non-live conductors. Consequently, inadequate monitoring may lead to safety accidents.

[0023] In view of this, the embodiments of this application provide a live-line work protection and external damage monitoring system and method. To solve the problems of existing technologies where fixed cameras are difficult to achieve all-round, blind-spot-free monitoring coverage, have monitoring blind spots, and are cumbersome to install, this application designs a strong magnetic gimbal with two rotating motors, a horizontal axis motor and a vertical axis motor, inside. To solve the problems of existing technologies being greatly affected by ambient light and unable to distinguish between live and non-live bodies, this application designs a system that can perform live-line work protection and external damage monitoring based on the data fused from the data collected by lidar and electric field sensors. The system includes a lidar, an electric field sensor, a main control module, and a strong magnetic gimbal. The strong magnetic gimbal has a strong magnetic adsorption plate at its bottom for mounting it in the work area, a mounting platform at its top, two rotating motors (horizontal and vertical axes) inside, and data and power interfaces on its sides. The lidar is connected to the strong magnetic gimbal via the mounting platform to collect laser point cloud data of the work area. The electric field sensor is also connected to the strong magnetic gimbal via the mounting platform to collect electric field fluctuation data generated by energized equipment in the work area. The main control module is connected to the data and power interfaces of the strong magnetic gimbal via shielded cables to acquire the work area data collected by the lidar and electric field sensor, and to perform external damage prevention monitoring based on the work area data.

[0024] To make the technical solution of this application clearer and easier to understand, the following description, in conjunction with the accompanying drawings, introduces a live-line working external damage prevention monitoring system provided by an embodiment of this application. For example... Figure 1 As shown in the figure, this is a schematic diagram of a live-line working anti-external damage monitoring system provided in an embodiment of this application. The system includes: a lidar 2, an electric field sensor 3, a main control module 4, and a strong magnetic gimbal 1.

[0025] The strong magnetic gimbal 1 serves as the system's mounting carrier and rotating mechanism, supporting various sensors and driving their rotation for comprehensive monitoring of the work area. Its bottom features a strong magnetic adsorption plate for mounting the gimbal 1 to the work area, while the top houses an installation platform. Internally, it contains two rotating motors: a horizontal axis motor and a vertical axis motor. Data and power interfaces are located on the sides. As an example, the strong magnetic gimbal 1 can be cylindrical, with a diameter of 80-100mm and a height of 60-80mm, made of high-strength aluminum alloy, weighing no more than 1.5kg for easy carrying and installation. The strong magnetic adsorption plate uses neodymium iron boron magnets with an adsorption force of no less than 500N, allowing direct adhesion to metal equipment surfaces, such as vehicle bodies and substation metal supports, without bolt fixation. Installation takes less than one minute. The adsorption plate surface is equipped with anti-slip rubber pads to increase friction and prevent scratching the equipment surface. The installation platform is circular with four evenly distributed threaded holes for fixing the lidar 2 and other sensors. Both the horizontal and vertical axis motors are stepper motors, with speeds adjustable between 0-10° / s. The rotation angle range is 0-360° horizontally and -45° to 90° vertically, enabling synchronous rotation of sensors on the mounting platform for comprehensive data acquisition across the work area. The control and power interfaces on the side of the strong magnetic gimbal 1 are connected to the main control module 4 via shielded cables, receiving rotation commands and power from the module. This structural design solves the problems of cumbersome and time-consuming installation of traditional gimbals. The anti-slip rubber pads enhance the stability of the strong magnetic gimbal 1 during operation, preventing sensor accuracy degradation due to vehicle vibration, wind, or other factors. The dual-axis rotation design allows for omnidirectional, blind-spot-free data acquisition coverage, adapting to the characteristics of dense equipment and varied work scenarios in substations. Connecting to the main control module 4 via shielded cables, rather than wirelessly, avoids interference from electromagnetic fields in the power work environment.

[0026] It should be noted that, to address the impact of electromagnetic interference generated by the strong magnetic adsorption plate at the bottom of the strong magnetic gimbal 1 and the internal motor on the sensor data acquisition accuracy, as an example, the following approach can be adopted: For stray magnetic field interference from the strong magnetic adsorption plate, only the bottom adsorption surface of the plate can be exposed, while the remaining sides and top are completely encased in a double-layer permalloy shield, for example, with a thickness of 0.5mm. A 1mm thick silicon steel sheet magnetic permeable layer is attached to the inner wall of the shield to guide the stray magnetic field to close along the inside of the shield. A 6mm thick 6061 aluminum alloy non-magnetic barrier layer is placed between the strong magnetic adsorption plate and the gimbal body to block the upward transmission of the magnetic path to the sensor mounting area. Simultaneously, adjust the installation position of the electric field sensor 3, fixing it at the edge of the mounting platform on top of the gimbal, so that the horizontal and vertical distances between its sensing electrode and the center of the strong magnetic adsorption disk are both large, controlling the offset of the strong magnetic stray magnetic field from the measurement reference of the electric field sensor 3 to a low level, thereby solving the stray magnetic field interference of the strong magnetic adsorption disk; for the electromagnetic interference of the horizontal axis motor and the vertical axis motor inside the gimbal, both stepper motors can be completely encapsulated with a cold-rolled steel shield shell with a thickness of, for example, 1mm. The shield shell is connected to the metal body of the strong magnetic gimbal 1 through a conductive copper strip to achieve equipotential connection, so that the grounding resistance is no more than 1Ω, thereby solving the electromagnetic interference of the horizontal axis motor and the vertical axis motor inside the gimbal.

[0027] The LiDAR 2 is connected to the high-magnetic gimbal 1 via a mounting platform. It is used to collect laser point cloud data of the work area. This laser point cloud data is a discrete set of points containing information such as the spatial coordinates, distance, and contours of objects, used to construct a 3D environment model and calculate distances between objects. As an example, the LiDAR 2 has a rectangular structure, measuring 60mm × 40mm × 30mm and weighing no more than 200g. It uses a 905nm laser wavelength, with a detection range of 0.5m-50m, a detection accuracy of at least ±2cm, and an adjustable frame rate of 10-20Hz. It is resistant to rain, snow, and fog interference, making it suitable for complex outdoor weather environments. The front of the LiDAR 2 has a laser window made of high-transmittance glass with an anti-reflective coating to reduce interference from external light on the detection results. The lidar 2 internally integrates a laser emitter, a laser receiver, and a signal processing module. The laser emitter emits a laser beam, which reflects off an object. The laser receiver receives the reflected laser beam. The signal processing module calculates the time of flight of the laser beam to obtain the distance between the object and the lidar 2, generating corresponding laser point cloud data, which is then converted into an electrical signal and transmitted to the main control module 4. The lidar 2 is fixed to the center of the mounting platform on top of the strong magnetic gimbal 1 using M3 bolts. Vibration-damping pads are installed at the bottom to reduce the impact of vibrations generated during the rotation of the strong magnetic gimbal 1 and during operation on the detection accuracy of the lidar 2. The normal of the laser window of the lidar 2 is parallel to the normal of the lens of the subsequently installed camera 5, with a parallelism error of no more than ±1°, ensuring that the detection range of the lidar 2 matches the acquisition range of the electric field sensor 3. This design enables the lidar 2 to stably acquire spatial data in complex weather conditions, solving the problem of traditional lidar 2 being greatly affected by weather interference. The vibration-damping pads further improve the accuracy of data acquisition.

[0028] The electric field sensor 3 is connected to the strong magnetic gimbal 1 via a mounting platform. It is used to collect electric field fluctuation data generated by energized equipment within the work area. This data represents the change in electric field strength around the energized equipment over time and space, used to identify energized bodies and their operating status. As an example, the electric field sensor 3 has a cylindrical structure with a diameter of 20mm, a height of 30mm, and a weight not exceeding 50g. It employs a capacitive sensing principle, with a measurement range of 0-50kV / m, a measurement accuracy of ±0.1kV / m, a response time of no more than 0.5 seconds, and anti-electromagnetic interference capabilities, making it suitable for the complex electromagnetic environment of substations. The electric field sensor 3 has a sensing electrode at its front end, made of copper with a corrosion-resistant coating, enabling rapid sensing of changes in the surrounding electric field. Internally, the electric field sensor 3 incorporates a detection circuit and a signal amplification module. The sensing electrode converts the changes in the surrounding electric field into a weak voltage signal. The detection circuit filters and denoises the voltage signal, and the signal amplification module amplifies the processed voltage signal to a range recognizable by the main control module 4 before transmitting it to the main control module 4. The electric field sensor 3 is fixed to the side of the lidar 2 with a snap-fit, with the sensing electrode facing outwards to ensure unobstructed operation and rapid sensing of electric field fluctuations generated by surrounding charged objects. The electric field sensor 3, lidar 2, and camera 5 rotate synchronously with the strong magnetic gimbal 1, ensuring the acquisition range covers the entire work area. This design enables the system to identify charged equipment within the work area, solving the problem of traditional sensors being unable to detect charged equipment. The snap-fit ​​connection facilitates the disassembly, calibration, and maintenance of the electric field sensor 3.

[0029] The main control module 4 connects to the data and power interfaces of the strong magnetic gimbal 1 via shielded cables. It acquires data from the work area collected by the lidar 2 and electric field sensor 3, and performs external damage prevention monitoring based on this data. As an example, the main control module 4 uses an ARM Cortex-A9 embedded processor with a 1.5GHz clock speed, sufficient for real-time data processing. It has 64GB of built-in storage, capable of storing collected laser point cloud data, electric field fluctuation data, and processing results in real time, facilitating data traceability and analysis after operation. The module also includes a 4G / 5G communication module, enabling remote data transmission and control, sending monitoring data and early warning information to a remote monitoring center. A rechargeable lithium battery with a capacity of at least 5000mAh provides stable power to all components, supporting continuous operation for at least 8 hours, meeting the requirements for prolonged live-line work. Finally, a power management module enables intelligent energy consumption control of each component, reducing overall system power consumption. The main control module 4 features a touchscreen display on its front, allowing operators to adjust parameters and view the operating status and monitoring data of various system components. Data and power interfaces are located on the side of the main control module 4. The main control module 4 can be fixed to the side of the strong magnetic gimbal 1, or placed on the control panel of the work vehicle or in the operator's toolbox. During installation, the main control module 4 must be kept away from high-voltage electrical equipment to avoid electromagnetic interference affecting its operational stability. This design enables the main control module 4 to coordinate the collaborative work of all components, realizing the various functions of the system. The large-capacity lithium battery and power management module extend the system's battery life, meeting the needs of long-term operation.

[0030] It should be noted that the main control module 4 can be directly connected to the lidar 2 and the electric field sensor 3 via shielded cables. Specifically, the data interface of the main control module 4 can be connected to the data interfaces of the lidar 2 and the electric field sensor 3 respectively via shielded cables to achieve data transmission. The power supply interface of the main control module 4 can be connected to the power supply interfaces of the lidar 2 and the electric field sensor 3 respectively via shielded cables to supply power to the lidar 2 and the electric field sensor 3. At this time, the power supply interface and data interface of the lidar 2 are located on the side of the lidar 2, and the power supply interface and data interface of the electric field sensor 3 are located on the side of the electric field sensor 3. The main control module 4 can also be connected directly to the lidar 2 and the electric field sensor 3 via shielded cables. The shielded cable is connected to the data interface and power supply interface of the strong magnetic gimbal 1. The strong magnetic gimbal 1 acts as an intermediary to provide power and data transmission to the lidar 2 and the electric field sensor 3. Specifically, one or more sets of power supply interfaces and data interfaces can be set on the mounting platform at the top of the strong magnetic platform. When the lidar 2 and the electric field sensor 3 are connected to the mounting platform, the power supply interface and data interface of the mounting platform are connected to the power supply interface and data interface of the lidar 2 and the electric field sensor 3 respectively. At this time, the power supply interface and data interface of the lidar 2 are located at the bottom of the lidar 2, and the power supply interface and data interface of the electric field sensor 3 are also located at the bottom of the electric field sensor 3.

[0031] In the embodiments provided in this application, the main control module 4 can perform external damage monitoring through the following steps S101-S104: S101, acquire laser point cloud data of the work area collected by the lidar 2 and electric field fluctuation data of the work area collected by the electric field sensor 3.

[0032] Specifically, lidar 2 acquires laser point cloud data of the work area, and electric field fluctuation data of the work area is acquired through electric field sensor 3. Lidar 2 uses a frequency... Laser point cloud data is collected, each laser point cloud data includes three-dimensional coordinates (x, y, z) and reflection intensity r; electric field sensor 3 uses frequency... Collect electric field intensity data. Each electric field fluctuation data includes the electric field intensity value E and the acquisition time t.

[0033] S102, the laser point cloud data and the electric field fluctuation data are fused by spatiotemporal registration and weighted Bayesian decision-making to obtain the first data fusion result.

[0034] In the embodiments provided in this application, linear interpolation can be used to interpolate the electric field fluctuation data so that the interpolated electric field fluctuation data and the laser point cloud data correspond one-to-one in time; the laser point cloud data in the lidar 2 coordinate system is transformed to the global coordinate system; the spatiotemporally registered laser point cloud data and electric field fluctuation data are fused using a weighted Bayesian decision method to obtain a discrimination result for the object category corresponding to each laser point cloud data; the discrimination result is used as the first data fusion result.

[0035] Specifically, time registration is first performed to unify sensor data from different sampling frequencies to the same time base. Linear interpolation is then used to interpolate the data from electric field sensor 3, ensuring a one-to-one temporal correspondence between the interpolated electric field fluctuation data and the laser point cloud data. The interpolation formula is as follows:

[0036] in, for The electric field strength value after interpolation at time 10:00. and They are respectively and The electric field strength values ​​collected at different times, The acquisition time of the laser point cloud data, and meets the following requirements. ≤ ≤ .

[0037] Then, spatial registration is performed to transform the laser point cloud data from the LiDAR 2 coordinate system to the global coordinate system. The transformation formula is as follows:

[0038] in, Point cloud coordinate vector in global coordinate system R is a 3×3 rotation matrix, and P is the point cloud coordinate vector in the lidar 2-coordinate system. T is a 3×1 translation vector.

[0039] For each laser point cloud dataset, local geometric features are extracted to obtain the lidar feature vector. Extract its local geometric features to construct the lidar feature vector. :

[0040] in: , , The three-dimensional coordinates of the point cloud in the global coordinate system. The intensity of laser reflection from the cloud. , , The normal vector components of the local plane containing the point cloud are obtained by fitting the plane of the neighboring point cloud. Let be the curvature of the local region containing the point cloud, calculated as the ratio of the smallest eigenvalue of the covariance matrix of the neighboring point cloud to the sum of all eigenvalues. The point cloud density is the density of the local area where the point cloud is located. It is calculated as the number of point clouds contained in a sphere centered on the point and with a radius equal to the radius of a preset area.

[0041] It should be noted that the feature vector of LiDAR As an example, the vector is a standardized vector for each laser point cloud data. After extracting its local geometric features to obtain the original lidar feature vector, the original lidar feature vector is then subjected to Z-score normalization to obtain the lidar feature vector. Specifically, it can be calculated using the following formula:

[0042] in, The j-th eigenvalue is the standardized eigenvalue. The j-th eigenvalue is the eigenvalue of the original lidar feature vector; Let be the mean of the j-th feature; Let be the standard deviation of the j-th feature; it should be noted that in the case of extreme features... When that happens, simply set the value of that eigenvalue to 0.

[0043] Feature extraction is performed on the electric field fluctuation data to obtain the electric field intensity data. .

[0044] A Gaussian Mixture Model (GMM) is used to model class-conditional probabilities based on the lidar feature vectors and the electric field intensity data, respectively, to obtain the class-conditional probabilities of the lidar features and the electric field features. Specifically, a Gaussian Mixture Model (GMM) can be used to model the class-conditional probabilities of the lidar features and the electric field features, respectively. The model parameters are trained using an offline labeled dataset. For any preset object category... (k=1,2,...,K, where K is the preset total number of object categories, which can be expanded according to the operational scenario), class-conditional probabilities of LiDAR features. Where M is the number of components in the Gaussian mixture model, set according to the class complexity. Let m be the weight of the m-th Gaussian component of the k-th class, satisfying , The mean is The covariance matrix is The 9-dimensional Gaussian probability density function, Let be the mean vector of the m-th Gaussian component of the k-th class, with dimensions . Consistent, The 9×9 covariance matrix of the m-th Gaussian component in the k-th class; the class conditional probability of the electric field characteristics. Where N is the number of components in the Gaussian mixture model of the electric field characteristics. The weights of the nth Gaussian component of the kth class satisfy the following condition: , The mean is The variance is The one-dimensional Gaussian probability density function, Let be the mean of the nth Gaussian component of the kth class. Let be the variance of the nth Gaussian component of the kth class.

[0045] Introducing a sensor-category reliability matrix , This represents the reliability weight of sensor s for class k, where s=1 corresponds to lidar, s=2 corresponds to electric field sensor, and satisfies... For any k, the reliability matrix can be obtained through offline cross-validation, reflecting the performance differences of different sensors in recognizing different categories. Reliability Matrix The specific acquisition method is as follows: Use the labeled training dataset to train a classification model based solely on LiDAR and a classification model based solely on electric field sensors, and calculate the classification for each category. The classification accuracy of the next two models and , , .

[0046] It should be noted that the characteristics of different sensors are conditionally independent for a given category, as determined by the reliability matrix. Based on the class conditional probability of the lidar features Class conditional probabilities of electric field characteristics Calculate the joint posterior probability of each laser point cloud data belonging to each object category. ,in, Category obtained solely based on lidar features The posterior probability is calculated using the following formula: , For category The prior probability, Categories obtained solely based on electric field characteristics The posterior probability is calculated using the following formula: .

[0047] Based on the joint posterior probability, the object category corresponding to each laser point cloud data is determined using the following formula based on the maximum posterior probability criterion. :

[0048] This formula indicates that the category with the highest posterior probability is selected as the classification result of the point cloud, thus obtaining the first data fusion result.

[0049] S103, a three-dimensional model is constructed using the voxel method based on the first data fusion result.

[0050] In the embodiments provided in this application, the fused laser point cloud data can be used, i.e., the object category corresponding to each laser point cloud data. A 3D environment model of the work area can be constructed in real time. Specifically, a voxelization method can be used to construct the model based on the object category corresponding to each laser point cloud data. The process involves dividing the three-dimensional space into voxels of equal size, with each voxel storing the category information and statistical characteristics of the laser point cloud data within that region.

[0051] S104, Use the three-dimensional model to monitor for external damage.

[0052] In the embodiments provided in this application, different levels of hazardous areas can be divided in a three-dimensional model according to the location and voltage level of charged bodies and / or obstacles; the positions of workers and tools in the three-dimensional model can be monitored in real time; the distance between workers and tools and the boundaries of each hazardous area can be calculated; and when the distance is less than a preset threshold, a warning signal of the corresponding level can be triggered.

[0053] Specifically, in the 3D model, different levels of danger zones can be divided according to the location and voltage level of energized bodies and / or obstacles. For example, for 110kV energized equipment, a high-risk zone is defined as the area within 1m of the equipment, a medium-risk zone as the area within 1m to 2m, and a low-risk zone as the area within 2m to 3m; for 220kV energized equipment, a high-risk zone is defined as the area within 1.5m of the equipment, a medium-risk zone as the area within 1.5m, and a low-risk zone as the area within 3m to 4m; for 500kV energized equipment, a high-risk zone is defined as the area within 2m of the equipment, a medium-risk zone as the area within 2m to 4m, and a low-risk zone as the area within 4m to 5m. The main control module 4 monitors the positions of personnel and tools in the 3D model in real time. The positions of personnel and tools can be identified and tracked using laser point cloud data collected by the lidar 2. The main control module 4 calculates the distances between personnel and tools and the boundaries of each danger zone. When the distance is less than a preset threshold, a warning signal of the corresponding level is triggered. For example, for 110kV live equipment, when the distance between the worker and the boundary of the high-risk area is less than 1m, i.e., the worker has entered the high-risk area, a high-risk warning signal is triggered; when the distance between the worker and the boundary of the medium-risk area is greater than 1m but less than 2m, a medium-risk warning signal is triggered; and when the distance between the worker and the boundary of the low-risk area is greater than 2m but less than 3m, a low-risk warning signal is triggered. During a low-risk warning, the main control module 4 displays a normal status reminder on the touch screen; during a medium-risk warning, the main control module 4 displays medium-risk warning information on the touch screen and simultaneously emits an intermittent alarm sound with a volume not less than 80dB and adjustable; during a high-risk warning, the main control module 4 displays high-risk warning information on the touch screen, emits a continuous alarm sound, and simultaneously pushes the warning information remotely to the monitoring center via the built-in 4G / 5G communication module. This tiered warning system can issue corresponding reminders to workers according to different risk levels, enabling workers to take timely measures to avoid risks and reduce the incidence of safety accidents.

[0054] In the embodiments provided in this application, as a feasible implementation method, the live-line working anti-external damage monitoring system may further include a camera 5 connected to the strong magnetic pan-tilt unit 1 via a mounting platform, for collecting image data of the working area. For details, please refer to... Figure 2 This figure is a schematic diagram of another live-line working anti-external damage monitoring system provided in an embodiment of this application, and is consistent with... Figure 1 Compared to the previous version, a new camera 5 has been added. The system includes: LiDAR 2, electric field sensor 3, main control module 4, strong magnetic gimbal 1, and camera 5.

[0055] The camera 5 is connected to the strong magnetic gimbal 1 via the mounting platform and is used to collect image data of the work area. As an example, the camera 5 can be a wide-angle high-definition camera with a resolution of at least 1080P, a lens angle of at least 120°, low-light imaging capability, and a low-light sensitivity of 0.01Lux, enabling clear image acquisition in low-light environments. It operates at a frame rate of 30fps, allowing real-time acquisition of dynamic images and avoiding image stuttering. The lens of the camera 5 features an anti-fog and waterproof design, adapting to complex outdoor weather conditions. The camera 5 is fixed to the mounting platform on top of the strong magnetic gimbal 1 via a snap-fit ​​connection, located to one side of the lidar 2, adjacent to it, with its lens facing outwards and parallel to the normal of the lidar 2's laser window, ensuring that the acquisition range matches the detection range of the lidar 2. The camera 5, lidar 2, and electric field sensor 3 rotate synchronously with the strong magnetic gimbal 1, achieving omnidirectional image acquisition of the work area. The camera 5 is controlled by the main control module 4 to perform startup, parameter adjustment, and image transmission. Parameter adjustment includes functions such as exposure and frame rate. The main control module 4 is also used to acquire image data of the work area collected by the camera 5, and to perform external damage prevention monitoring based on the image data and work area data. The camera 5 can help identify charged bodies and obstacles, making up for the lack of image color and detail information that the lidar 2 cannot obtain, and at the same time providing texture information for 3D environment modeling, improving the realism of the 3D model.

[0056] It should be noted that the main control module 4 can be directly connected to the camera 5 via a shielded cable. Specifically, the data interface of the main control module 4 can be connected to the data interface of the camera 5 via a shielded cable to achieve data transmission. The power supply interface of the main control module 4 can be connected to the power supply interface of the camera 5 via a shielded cable to supply power to the camera 5. In this case, the power supply interface and data interface of the camera 5 are located on the side of the camera 5. Alternatively, the main control module 4 can be connected to the data interface and power supply interface of the strong magnetic gimbal 1 via only a shielded cable. The strong magnetic gimbal 1 acts as an intermediary for supplying power and transmitting data to the camera 5. Specifically, one or more sets of power supply interfaces and data interfaces can be set on the mounting platform at the top of the strong magnetic platform. When the camera 5 is connected to the mounting platform, the power supply interface and data interface of the mounting platform are simultaneously connected to the power supply interface and data interface of the camera 5. In this case, the power supply interface and data interface of the camera 5 are located at the bottom of the camera 5.

[0057] In the embodiments provided in this application, the main control module 4 can perform external damage monitoring through the following steps S201-S205: S201, acquire laser point cloud data of the work area collected by the lidar 2, electric field fluctuation data of the work area collected by the electric field sensor 3, and image data of the work area collected by the camera 5.

[0058] Specifically, the lidar 2 uses frequency Laser point cloud data is collected, each laser point cloud data includes three-dimensional coordinates (x, y, z) and reflection intensity r; electric field sensor 3 uses frequency... The camera collects electric field strength data, with each electric field fluctuation data point containing the electric field strength value E and the acquisition time t. The camera 5 operates at a frequency... Collect image data.

[0059] S202, the laser point cloud data and the image data are fused to obtain a fused feature vector.

[0060] In the embodiments provided in this application, lidar geometric features can be extracted from lidar point cloud data, and camera visual features can be extracted from image data. Then, based on the lidar geometric features and camera visual features, a multi-scale attention fusion network is used to extract features of different granularities through multi-scale convolution. The contribution weights of the lidar geometric features and the camera visual features are then dynamically adjusted through an attention mechanism to obtain a fused feature vector.

[0061] Specifically, spatiotemporal registration is performed first, and the method of spatiotemporal registration is the same as... Figure 1 Similar to the illustrated embodiment, for time registration, linear interpolation can be used to interpolate the electric field fluctuation data and image data, ensuring a one-to-one temporal correspondence between the interpolated electric field fluctuation data and image data and the laser point cloud data. For spatial registration, the image data in the camera coordinate system can be transformed to the global coordinate system using the pinhole camera model, according to the following formula:

[0062] in,( , ) represents the image pixel coordinates, ( , , () represents the three-dimensional coordinates in the global coordinate system. and These are the focal lengths of the camera along the x-axis and y-axis, respectively. , ( ) represents the coordinates of the camera's principal point.

[0063] Then, the camera's visual features are extracted for each laser point cloud data. Based on the spatiotemporal registration results, find the corresponding pixel position in the image. , ), with that pixel as the center and a size of Image patches are used as regions to extract visual feature vectors. This includes RGB color histogram features, histogram of oriented gradients (HOG) features, and local binary pattern (LBP) features; it should be noted that the extracted visual feature vectors This is a standardized vector. In the embodiments provided in this application, as an example, the min-max normalization method can be used to map the original visual feature vector to the [0,1] interval. The specific calculation formula is as follows:

[0064] in, Let j be the value of the standardized visual feature in the j-th dimension. The j-th dimension value of the original visual features. and These are the minimum and maximum values ​​of the j-th dimension of the visual feature, respectively; it should be noted that in extreme special cases... At that time, we directly set the standardized value of this dimension to 0.5.

[0065] Simultaneously extract the feature vector of the lidar The method of extracting geometric features of lidar and Figure 1 The embodiments shown are similar and will not be described again here. It should be noted that the lidar feature vector This is the vector after standardization.

[0066] Then, a multi-scale attention fusion network (MSAF-Net) is used to process the lidar feature vectors. and visual feature vectors The fusion is performed to obtain the fused feature vector. The mathematical expression of the fusion process is as follows:

[0067] Where L is the total number of scales in the multi-scale convolution. This is the convolutional layer for the lidar features at the l-th scale, with an output dimension of... ; For the visual feature convolutional layer at the l-th scale, the output dimension is ; For the attention fusion module at the l-th scale, its output is: ,in, Let L be the convolutional feature of the LiDAR at the l-th scale, with dimension L. , Let L be the visual convolutional feature at the l-th scale, with dimension L. , For element-wise multiplication, It is the Sigmoid activation function. For a trainable weight matrix, It is a trainable bias vector; For feature concatenation operation, the output dimension is ; As a fully connected layer, it maps the concatenated features to a dimension of . fused feature vector .

[0068] S203, the fusion feature vector and the electric field fluctuation data are fused using a weighted Bayesian decision method to obtain a second data fusion result.

[0069] Specifically, a Gaussian mixture model can be used to model class-conditional probabilities based on the fused feature vector and electric field intensity data, respectively, to obtain the class-conditional probabilities of the fused features and the electric field features. For any predefined object category... (k=1,2,...,K), class conditional probabilities of fused features Where M is the number of components in the Gaussian mixture model. Let m be the weight of the m-th Gaussian component of the k-th class, satisfying , The mean is The covariance matrix is of Gaussian probability density function Let be the mean vector of the m-th Gaussian component of the k-th class, with dimensions . Consistent, For the m-th Gaussian component of the k-th class Covariance matrix; Class-conditional probability of electric field characteristics and Figure 1 The acquisition method is the same in the embodiments shown.

[0070] Then, the sensor-category reliability matrix is ​​introduced. , This represents the reliability weight of sensor s for class k, where s=1 corresponds to the fused feature, s=2 corresponds to the electric field sensor, and satisfies... The reliability matrix holds true for any k. The specific acquisition method is as follows: using the labeled training dataset, train a classification model based solely on fused features and a classification model based solely on electric field sensors, respectively, and calculate the classification for each category. The classification accuracy of the next two models and , , .

[0071] Through the reliability matrix Based on the class conditional probability of fusion features Class conditional probabilities of electric field characteristics Calculate the joint posterior probability of each laser point cloud data belonging to each object category. ,in, For categories obtained solely based on fusion features The posterior probability is calculated using the following formula: .

[0072] Based on the joint posterior probability, the object category corresponding to each laser point cloud data is determined using the following formula, based on the maximum posterior probability criterion. :

[0073] This formula indicates that the category with the highest posterior probability is selected as the classification result of the point cloud, thus obtaining the second data fusion result.

[0074] S204, Based on the second data fusion result, a three-dimensional model is constructed using the voxel method.

[0075] It should be noted that the specific implementation method of this step is different from... Figure 1 Step S103 in the illustrated embodiment is similar and will not be described again here.

[0076] S205, using the aforementioned three-dimensional model for external damage monitoring.

[0077] It should be noted that the specific implementation method of this step is different from... Figure 1 Step S104 in the illustrated embodiment is similar and will not be repeated here.

[0078] Based on the above description, this application has the following beneficial effects: This application provides a live-line work protection monitoring system composed of a lidar, an electric field sensor, a main control module, and a strong magnetic gimbal (a camera can also be added). By fusing the spatial information from the lidar with the charge information from the electric field sensor, it can simultaneously determine the position and charge state of an object, solving the problem that existing technologies cannot simultaneously achieve charge body identification and obstacle spatial positioning. The lidar provides the object's three-dimensional coordinates and contour information, solving the problem that the electric field sensor cannot locate it. The electric field sensor provides the object's charge information, solving the problem that the lidar cannot distinguish between charged and uncharged objects. The combination of these two methods can effectively reduce the false alarm rate and false negative rate of the monitoring results. Employing a two-level fusion method of spatiotemporal registration-weighted Bayesian decision (without a camera) or a three-level fusion method of spatiotemporal registration-feature-decision-control level fusion (with a camera), it can eliminate... The temporal and spatial differences in data from different sensors address the lack of effective data fusion mechanisms in existing technologies. The weighted Bayesian decision model dynamically adjusts weights based on the reliability of each sensor, further improving classification accuracy. When the system includes cameras, a convolutional neural network fuses the geometric features of the LiDAR with the visual features of the cameras, extracting richer object feature information and further improving object classification accuracy. Simultaneously, image data provides texture information for the 3D model, enhancing its visualization and allowing workers to intuitively understand the work environment. By constructing a 3D environment model and defining hazardous areas within it, the system can monitor the distance between workers and tools and these areas in real time, triggering different levels of warning signals based on the distance. This provides clear safety guidance for workers, helping them avoid risks in a timely manner.

[0079] The above text combined Figure 1 and Figure 2 This application provides a detailed description of a live-line work protection monitoring system for preventing external damage, and the following section will describe a live-line work protection monitoring method for preventing external damage, in conjunction with the accompanying drawings.

[0080] like Figure 3 As shown in the figure, this is a schematic flowchart of a live-line work protection monitoring method according to an embodiment of this application. The method includes: S301, the main control module controls the electric field sensor to continuously collect electric field fluctuation data in the working area.

[0081] It should be noted that when monitoring for external damage during live-line work, the following preparations are required: The operator uses the strong magnetic gimbal's magnetic adsorption plate to attach the gimbal to a designated location in the work area, such as the top of the work vehicle or a metal support frame in a substation, at a height of 2-3 meters above the ground, ensuring unobstructed sensor coverage. Connect all component cables and press the power button on the main control module to start the system. The main control module automatically sends self-test commands to the lidar, electric field sensor, strong magnetic gimbal, and camera (if the system includes a camera) to check the working status of each component. If a component malfunctions, the main control module immediately displays a fault message on the touchscreen and issues an alarm sound to alert the operator to troubleshoot the problem. After the self-test is complete, the operator sets relevant system parameters through the main control module's touchscreen, including the electric field sensor's measurement range, warning threshold, strong magnetic gimbal rotation speed, lidar detection range, and camera resolution, frame rate, and exposure time parameters (if the system includes a camera).

[0082] The main control module controls the electric field sensor to continuously collect electric field fluctuation data in the working area. After the electric field sensor is powered on, it is in working condition throughout the entire process and transmits the collected electric field fluctuation data to the main control module in real time.

[0083] S302, the main control module controls the strong magnetic gimbal to drive the lidar and electric field sensor to rotate according to the preset speed.

[0084] The main control module controls the electric field sensor to continuously collect electric field fluctuation data in the working area. After the electric field sensor is powered on, it is in working condition throughout the entire process and transmits the collected electric field fluctuation data to the main control module in real time.

[0085] S303, the main control module determines whether there are any abnormalities in the working area based on the electric field fluctuation data.

[0086] The main control module controls the strong magnetic gimbal to rotate the lidar and electric field sensor at a preset rotation speed. When the system includes a camera, it also rotates the camera, enabling omnidirectional monitoring of the work area. The preset rotation speed refers to the angular velocity at which the strong magnetic gimbal rotates the lidar and electric field sensor. It can be set to 5° / s by default and can be adjusted according to the size of the work area and monitoring requirements.

[0087] S304: When an anomaly occurs in the work area, the main control module controls the lidar to collect laser point cloud data of the work area.

[0088] Specifically, the main control module determines whether there are any anomalies in the work area based on electric field fluctuation data. The main control module analyzes the electric field fluctuation data in real time, calculating the change in electric field intensity per unit time. When the change in electric field intensity exceeds a preset low-risk threshold, an anomaly is determined to exist in the work area. When the system includes a camera, background subtraction processing can be performed on continuous image frames captured by the camera simultaneously to calculate the pixel difference values ​​between frames, assisting in verifying whether there is object movement in the work area and improving the reliability of anomaly detection.

[0089] S305, the main control module performs external damage prevention monitoring based on the electric field fluctuation data and the laser point cloud data.

[0090] Specifically, the main control module determines whether there are any anomalies in the work area based on electric field fluctuation data. The main control module analyzes the electric field fluctuation data in real time, calculating the change in electric field intensity per unit time. When the change in electric field intensity exceeds a preset low-risk threshold, an anomaly is determined to exist in the work area. When the system includes a camera, background subtraction processing can be performed on continuous image frames captured by the camera simultaneously to calculate the pixel difference values ​​between frames, assisting in verifying whether there is object movement in the work area and improving the reliability of anomaly detection.

[0091] When an anomaly occurs in the work area, the main control module controls the lidar to collect the lidar point cloud data of the work area. When the system includes a camera, it also controls the camera to collect the image data of the work area.

[0092] It should be noted that the LiDAR and camera are in sleep standby mode when no anomalies are detected, maintaining only minimal standby power consumption. Upon receiving a start command from the main control module, the LiDAR and camera quickly start up, with a startup time of no more than one second, and begin collecting corresponding data in real time and transmitting it to the main control module. This control method can reduce system power consumption and extend system battery life.

[0093] The main control module performs external damage prevention monitoring based on electric field fluctuation data and laser point cloud data. When the system includes a camera, it performs external damage prevention monitoring based on electric field fluctuation data, laser point cloud data, and image data. When the system does not include a camera, the monitoring process involves fusing laser point cloud data and electric field fluctuation data through spatiotemporal registration and weighted Bayesian decision-making to obtain a first data fusion result. Based on the first data fusion result, a three-dimensional model is constructed using the voxel method. In the three-dimensional model, different levels of danger zones are divided according to the location and voltage level of charged bodies and obstacles. The positions of workers and tools in the three-dimensional model are monitored in real time, and the distances between workers and tools and the boundaries of each danger zone are calculated. When the distance is less than a preset threshold, the corresponding level of warning signal is triggered. When the system includes a camera, the monitoring process involves fusing laser point cloud data and image data to obtain a fused feature vector. The fused feature vector and electric field fluctuation data are then fused using spatiotemporal registration and weighted Bayesian decision-making to obtain a second data fusion result. Based on the second data fusion result, a 3D model with texture information is constructed using the voxel method. The subsequent processes of danger zone delineation, location monitoring, distance calculation, and early warning triggering are consistent with those when the system does not include a camera.

[0094] If there are no anomalies in the work area and the LiDAR does not detect any changes in the laser point cloud data for a duration exceeding a preset time (which can be set to 5 seconds), the main control module will control the LiDAR to enter sleep mode and stop collecting laser point cloud data. If the system includes a camera, the camera will also enter sleep mode and stop collecting image data. The main control module will control the strong magnetic gimbal to rotate to a position where the electric field sensor can collect the maximum range of the work area, and then stop rotating to reduce the energy consumption of the strong magnetic gimbal. The main control module will then enter a low-power operation mode, shutting down unnecessary power supply modules, such as communication modules and GPU computing modules. When the system includes a camera, the camera's power supply module will also be shut down, retaining only the basic operating functions of the electric field sensor and the main control module. This low-power control method can further reduce system energy consumption and meet the needs of long-term live-line operations.

[0095] In the embodiments provided in this application, when the main control module controls the strong magnetic gimbal to rotate to a position where the electric field sensor can collect the maximum range of the working area, the calculation method for the determined maximum range can be: pre-calibrating the effective detection angle of the electric field sensor, taking the gimbal installation position as the origin, and traversing all possible gimbal rotation angles. ,in, For horizontal corners, For the vertical rotation angle, calculate the overlap area between the effective detection range of the electric field sensor and the preset working area at this angle. Choose the angle that maximizes the overlapping area. As an electric field sensor, it can collect the angle of the working area over the largest range.

[0096] After the live-line work is completed, the operator presses the power-off button on the main control module, shutting down the system and stopping all components. The main control module automatically saves all monitoring data from this operation, including electric field fluctuation data, laser point cloud data, image data, and early warning records, facilitating subsequent traceability and data analysis. The operator removes the high-powered magnetic gimbal from its installation position, cleans the dust and dirt from the surfaces of all components, and charges the main control module for future use.

[0097] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A live-line working protection monitoring system for external damage, characterized in that, The system includes: LiDAR, electric field sensor, main control module, and strong magnetic gimbal; The strong magnetic gimbal is equipped with a strong magnetic adsorption plate at the bottom for mounting the strong magnetic gimbal in the work area, an installation platform at the top, two rotating motors (horizontal axis motor and vertical axis motor) inside, and a data interface and a power supply interface on the side. The lidar is connected to the high-magnetic gimbal via the mounting platform and is used to collect laser point cloud data of the work area. The electric field sensor is connected to the strong magnetic gimbal via the mounting platform and is used to collect electric field fluctuation data generated by electrical equipment in the work area. The main control module is connected to the data interface and power supply interface of the strong magnetic gimbal via a shielded cable, and is used to acquire the work area data collected by the lidar and the electric field sensor, and to perform external damage prevention monitoring based on the work area data.

2. The system according to claim 1, characterized in that, The system also includes: A camera connected to the high-magnetic gimbal via the mounting platform is used to collect image data of the work area; The main control module is also used to acquire image data of the work area collected by the camera, and to perform external damage prevention monitoring based on the image data and the work area data.

3. The system according to claim 1, characterized in that, The main control module is specifically used for: Acquire laser point cloud data of the work area collected by the lidar and electric field fluctuation data of the work area collected by the electric field sensor; The laser point cloud data and the electric field fluctuation data are fused using spatiotemporal registration and weighted Bayesian decision-making to obtain a first data fusion result. A three-dimensional model is constructed based on the first data fusion result using the voxel method; The three-dimensional model is used for external damage monitoring.

4. The system according to claim 2, characterized in that, The main control module is specifically used for: Acquire laser point cloud data of the work area collected by the lidar, electric field fluctuation data of the work area collected by the electric field sensor, and image data of the work area collected by the camera; The laser point cloud data and the image data are fused to obtain a fused feature vector; The fusion feature vector and the electric field fluctuation data are fused using a weighted Bayesian decision method to obtain a second data fusion result. A three-dimensional model is constructed based on the second data fusion result using the voxel method; The three-dimensional model is used for external damage monitoring.

5. The system according to claim 3 or 4, characterized in that, The main control module is specifically used for: In the three-dimensional model, different levels of danger zones are divided according to the location and voltage level of charged bodies and / or obstacles; Real-time monitoring of the positions of workers and tools within the 3D model; Calculate the distances between workers and tools and the boundaries of each hazardous area; When the distance is less than a preset threshold, a warning signal of the corresponding level is triggered.

6. The system according to claim 3, characterized in that, The main control module is specifically used for: The electric field fluctuation data is interpolated using a linear interpolation method so that the interpolated electric field fluctuation data corresponds one-to-one with the laser point cloud data in time. Transform the laser point cloud data in the lidar coordinate system to the global coordinate system; The spatiotemporally registered laser point cloud data and electric field fluctuation data are fused using a weighted Bayesian decision method to obtain the object category classification result for each laser point cloud data. The discrimination result is used as the first data fusion result.

7. The system according to claim 6, characterized in that, The main control module is specifically used for For each laser point cloud data, local geometric features are extracted to obtain the lidar feature vector; Feature extraction is performed on the electric field fluctuation data to obtain the electric field intensity data; Using a Gaussian mixture model, class-conditional probability modeling is performed based on the lidar feature vector and the electric field intensity data, respectively, to obtain the class-conditional probability of lidar features and the class-conditional probability of electric field features. Based on the class conditional probabilities of the lidar features and the electric field features, calculate the joint posterior probability of each lidar point cloud data belonging to each object category. Based on the joint posterior probability, the object category corresponding to each laser point cloud data is determined according to the maximum posterior probability criterion.

8. The system according to claim 4, characterized in that, The main control module is specifically used for: Extracting lidar geometric features from lidar point cloud data; Extracting camera visual features from image data; Based on the geometric features of the LiDAR and the visual features of the camera, a multi-scale attention fusion network is adopted. Features of different granularities are extracted through multi-scale convolution, and the contribution weights of the geometric features of the LiDAR and the visual features of the camera are dynamically adjusted through the attention mechanism to obtain the fused feature vector.

9. A method for monitoring external damage during live-line working, characterized in that, The method includes: The main control module controls the electric field sensor to continuously collect electric field fluctuation data in the working area; The main control module controls the strong magnetic gimbal to drive the lidar and electric field sensor to rotate at a preset speed. The main control module determines whether there are any anomalies in the work area based on the electric field fluctuation data; When an anomaly occurs in the work area, the main control module controls the lidar to collect laser point cloud data of the work area; The main control module performs external damage prevention monitoring based on the electric field fluctuation data and the laser point cloud data.

10. The method according to claim 9, characterized in that, When there are no anomalies in the work area and the lidar does not detect any changes in the lidar point cloud data of the work area for a period exceeding a preset duration, the method further includes: The main control module controls the lidar to enter sleep mode; The main control module controls the strong magnetic gimbal to rotate to the position where the electric field sensor can collect the working area over the maximum range, and then stops rotating. The main control module controls itself to enter a low-power operation mode.

Images