Power transmission line real-time monitoring system and method

Abstract

The invention discloses a power transmission line real-time monitoring system, and the system is characterized in that the system comprises an infrared laser radar detection module which is used for carrying out the three-dimensional scanning of a power transmission line channel, and generating point cloud data; the image acquisition module is used for acquiring video images of the surrounding environment of the power transmission line in real time; the central processing unit is electrically connected with the infrared laser radar detection module and the image acquisition module, and is used for receiving, fusing and processing the point cloud data and the video image, carrying out target identification, distance calculation and behavior analysis, and generating an early warning signal; and the early warning prompt module is electrically connected with the central processing unit and is used for giving out sound-light alarm and / or voice warning when receiving the early warning signal. The power transmission line real-time monitoring system is not affected by day and night illumination changes through the infrared laser radar, all-weather uninterrupted monitoring can be achieved, and the problem that traditional video monitoring fails at night, in foggy days and in rainy and snowy days is thoroughly solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of line monitoring, in particular to a power transmission line real-time monitoring system and method. BACKGROUND

[0002] With the continuous advancement of urban construction and infrastructure projects, construction activities in the power transmission line channel are increasingly frequent, and large machinery such as cranes and excavators operating under high-voltage lines can easily cause line tripping, equipment damage, and even large-scale power outage accidents.

[0003] Traditional manual patrol methods are inefficient and slow to respond, and cannot achieve all-weather monitoring. Although existing video monitoring systems can achieve image acquisition, they lack intelligent recognition and accurate ranging capabilities, have a high false alarm rate, and have limited early warning effectiveness.

[0004] Therefore, there is an urgent need for an intelligent monitoring system that integrates high-precision ranging, real-time early warning, and remote communication to improve the external damage protection capability of power transmission lines. SUMMARY

[0005] In order to solve the technical problems existing in the prior art, the present application provides a power transmission line real-time monitoring system and method.

[0006] In order to achieve the above purpose, the technical scheme adopted by the present application is: The present application provides a power transmission line real-time monitoring system, characterized by comprising: an infrared laser radar detection module for three-dimensional scanning of the power transmission line channel to generate point cloud data; an image acquisition module for real-time acquisition of video images of the environment surrounding the power transmission line; a central processing unit electrically connected to the infrared laser radar detection module and the image acquisition module, for receiving and fusing processing of the point cloud data and the video images, target recognition, distance calculation and behavior analysis, and generating an early warning signal; an early warning prompt module electrically connected to the central processing unit for issuing an audible and / or visual alarm and / or voice warning when receiving the early warning signal; a communication module electrically connected to the central processing unit for remotely transmitting early warning information and related data to a background monitoring platform and / or a mobile terminal.

[0007] Optionally, the horizontal scanning angle of the infrared laser radar detection module is 180°±10°, and the vertical scanning angle is ±35°.

[0008] Optionally, the central processing unit comprises a point cloud processing subunit for three-dimensional reconstruction and safety distance calculation of the point cloud data. An image recognition subunit is configured to perform feature extraction and target recognition on the video image based on a convolutional neural network model; A data fusion and decision subunit is configured to fuse the point cloud processing result and the image recognition result, judge whether the target behavior constitutes an external force damage risk, and generate a warning signal.

[0009] Optionally, the image recognition subunit is pre-stored with a power transmission line hidden object database, and is configured to recognize at least one large construction machinery in the tower crane, excavator and crane.

[0010] Optionally, the warning prompt module comprises a wireless loudspeaker, which is configured to receive an instruction from the central processing unit to perform remote shouting.

[0011] Optionally, the system further comprises an environment sensor configured to collect wind speed, temperature and humidity data; and the central processing unit is further configured to dynamically adjust the warning distance threshold according to the environment data.

[0012] The application further provides a power transmission line real-time monitoring system method, which comprises the following steps: S1: scanning a power transmission channel by an infrared laser radar detection module to obtain three-dimensional point cloud data; S2: collecting environment video data by an image acquisition module; S3: processing the point cloud data by a central processing unit to calculate the minimum safety distance of the target object from the conductor; S4: identifying the video data by the central processing unit to determine whether there is a construction machinery; S5: fusing the point cloud ranging result and the image recognition result, and if a risk target is recognized and the minimum safety distance of the risk target from the conductor is less than a warning threshold, generating a warning signal; S6: starting an audible and / or visual alarm and / or voice driving reminder by a warning prompt module; S7: sending the alarm information and related data to a monitoring platform and a mobile terminal by a communication module.

[0013] Optionally, the step of processing the point cloud data comprises: S101: performing denoising and filtering preprocessing on the original point cloud data; S102: registering the preprocessed point cloud with a known three-dimensional model of the power transmission line; S103: segmenting and clustering non-ground point clouds to extract a single target; S104: calculating the minimum spatial distance of each target point cloud cluster from the three-dimensional model of the conductor.

[0014] Optionally, the S101 step specifically comprises: using a statistical outlier removal algorithm to calculate the average distance and standard deviation of each point and its adjacent points, and removing points exceeding a set standard deviation value multiple to filter out discrete noise; and / or using a voxel grid downsampling method to divide the point cloud space into a three-dimensional voxel grid, and using the center of gravity of all points in each voxel to represent the voxel to reduce the data amount while maintaining the shape features of the point cloud.

[0015] Optionally, the S102 step specifically comprises: obtaining the tower coordinates and conductor sag parameters of the power transmission line from the database; Based on the parameters, a high-precision three-dimensional model of the conductor and tower is generated through a catenary equation or a parabolic equation; The real-time collected point cloud data is registered with the high-precision three-dimensional model using an iterative closest point algorithm to achieve accurate alignment of the spatial coordinate system.

[0016] Beneficial effects: The infrared laser radar of the power transmission line real-time monitoring system is not affected by day and night light changes, can realize all-weather uninterrupted monitoring, and completely solves the problem of failure of traditional video monitoring in night, foggy, rainy and snowy weather. Not only can surface data be obtained, but also the three-dimensional shape, accurate position and spatial attitude of the object can be accurately obtained. The system can identify the stretching and lifting of the mechanical boom and other subtle dangerous actions, rather than just finding a stationary object. By fusing the accurate ranging of the point cloud and the accurate identification of the image, the system can greatly exclude environmental interference such as flying birds, floating bags and tree swaying, reduce the false alarm rate to a very low level, and ensure the reliability of the early warning information.

[0017] The present application also provides a power transmission line real-time monitoring method, which makes a decision by fusing the ranging results of the point cloud and the image recognition results, realizes complementary advantages, and provides accurate spatial distance information. The point cloud data is not conducive to judging the object type, and the image data is good at accurately identifying the object type but has poor ranging capability. When both are judged as "dangerous machinery" and "too close", the alarm is triggered. This avoids false alarms caused by flying birds, floating bags and tree swaying, and also avoids missed alarms caused by insufficient light and bad weather leading to image recognition failure, so that the early warning credibility reaches an unprecedented height. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 This is a block diagram of electronic component connections in Embodiment 1 of this application; Figure 2 This is a flowchart illustrating the monitoring method of Embodiment 2 of this application.

[0020] In the diagram: 1-Central Processing Unit; 2-Infrared LiDAR Detection Module; 3-Image Acquisition Module; 4-Early Warning Module; 5-Communication Module; 6-Environmental Sensor. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0023] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0024] In the description of this application, it should be noted that the use of terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" to indicate orientation or positional relationships is based on the orientation or positional relationships shown in the accompanying drawings, or the orientation or positional relationships commonly used when the product is in use. These terms are used solely for the convenience of describing this application and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the use of terms such as "first" and "second" in the description of this application is only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] Furthermore, the use of terms such as "horizontal" and "vertical" in the description of this application does not imply that the component is required to be absolutely horizontal or suspended, but rather that it may be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal relative to "vertical," and does not mean that the structure must be completely horizontal, but rather that it may be slightly tilted.

[0026] In the description of the present application, it also needs to be explained that, unless otherwise explicitly specified and limited, if the terms "arrange", "install", "connect", "connect" appear, they should be understood in a broad sense, for example, they can be fixedly connected, or can be detachably connected, or integrally connected; can be mechanically connected, or can be electrically connected; can be directly connected, or can be indirectly connected through an intermediate medium, or can be the communication inside two elements. For those skilled in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0027] Embodiment 1 Please refer to Figure 1 The embodiment provides a power transmission line real-time monitoring system, which is characterized by comprising: an infrared laser radar detection module 2, which is used for three-dimensional scanning of a power transmission line channel and generating point cloud data; the infrared laser radar detection module 2 emits a laser beam and receives light waves reflected from a target object, calculates the time difference and phase difference of the light beam flight, and accurately measures the distance and angle information of each point from the target. A number of points converge together to form three-dimensional point cloud data of the surrounding environment. It can accurately depict the contours and spatial positions of the boom, trees and buildings. It can provide centimeter-level ranging accuracy and accurately calculate the real spatial distance between the target and the conductor. The laser is actively emitted and does not depend on the ambient light, and can work stably in poor light conditions such as night, dusk and fog. Avoid the problem that the image recognition algorithm fails due to strong light, shadow and backlight.

[0028] An image acquisition module 3 is used for real-time acquisition of video images of the surrounding environment of the power transmission line; a high-definition camera is used to continuously shoot a visible light video stream of the power transmission line channel. Rich texture, color and appearance information is obtained, and the scene is recorded in a way that can be directly understood by the human eye. High-resolution images are provided, so that the system can accurately identify the category of the target using a convolutional neural network. The recorded video and pictures provide the most intuitive visual evidence for early warning events, facilitate remote confirmation of the on-site situation by the operation and maintenance personnel, and conduct post-tracing and responsibility identification. Help understand the complex context information such as the posture and operation type of the target.

[0029] The central processing unit 1 is electrically connected with the infrared laser radar detection module 2 and the image acquisition module 3, is used for receiving and fusing point cloud data and video images, target recognition, distance calculation and behavior analysis, and generating a warning signal; the central processing unit 1 is used for running core algorithm software, receiving point cloud data from the laser radar and video stream from the camera at the same time, and performing time synchronization and space calibration. The point cloud data is denoised, segmented, clustered, and independent objects are extracted, and the minimum spatial distance of each object from the guide line three-dimensional model is calculated. The video stream is analyzed in real time, and a deep learning model is used to identify whether there is a construction machine and its type in the picture. The accurate distance calculated by the point cloud is fused with the target type identified by the image to make a judgment. Only when the target is identified as "dangerous machinery" and the distance is less than the safety threshold, the warning signal is finally generated. Through multi-sensor information fusion and comprehensive judgment, false alarms caused by only one sensor are greatly eliminated. Precise warning is achieved, which is no longer a simple "intrusion alarm", but a precise warning based on real three-dimensional distance and target danger, and the reliability is greatly improved.

[0030] The warning prompt module 4 is electrically connected with the central processing unit 1, and is used for issuing sound and light alarm and / or voice warning when receiving the warning signal; The warning prompt module 4 drives the high-brightness LED lamp and the stroboscopic lamp to issue strong visible light warning, and drives the high-pitched loudspeaker to issue a piercing alarm sound and a pre-recorded voice warning prompt after receiving the warning signal from the central processing unit 1. The warning prompt module 4 warns the risky operation personnel at the first time and the first scene before the accident occurs, forces them to stop dangerous behavior and evacuate, so as to realize "stop in the process", which is the core value of the system. Without manual remote shouting, the system is automatically triggered, and the response delay is extremely low.

[0031] The communication module 5 is electrically connected with the central processing unit 1, and is used for remotely transmitting the warning information and related data to the background monitoring platform and / or mobile terminal. The communication module 5 is built-in 4G / 5G wireless communication module, packs the warning information generated by the central processing unit 1, and transmits it to the remote background monitoring platform and the mobile APP of the operation and maintenance personnel through the wireless network. The operation and maintenance personnel can master the line channel safety state at any time and anywhere without going to the scene, realize the intelligent operation and maintenance mode of unattended and remote monitoring. The alarm information reaches the person in charge, which is convenient for quickly starting the emergency response process and improving the management efficiency. All alarm records form a historical database, which is used for analyzing external breaking high-risk points and optimizing the line patrol strategy.

[0032] Further, the above laser radar and camera simultaneously collect data on the same scene, and obtain three-dimensional spatial information and two-dimensional texture information, respectively. The two types of heterogeneous data are synchronously sent to the central processing unit 1 and aligned in space-time through the pre-calibrated parameters, ensuring that the obtained target data is the same target. For example, the laser radar outputs data to the central processing unit 1, at a southwest direction of 45°, a distance of 5.2 meters, a height of 15 meters, and is moving, and the image data of the camera is a yellow tower crane. The central data processing unit combines the two pieces of information to make a final judgment as high risk, and then generates a warning signal. The warning signal triggers two actions at the same time, including local linkage and remote linkage. The local linkage starts the warning prompt module 4 to perform on-site sound and light alarm. The remote linkage reports the complete warning information to the monitoring center through the communication module 5. Through information redundancy and cross verification, the false alarm or temporary failure of a single sensor will not cause the overall system to misjudge or malfunction, and the system has very high reliability. The combination of laser radar and camera produces new functions that cannot be achieved by a single sensor, that is, high-precision distance monitoring and early warning of specific types of targets in complex environments.

[0033] Optionally, the horizontal scanning angle of the infrared laser radar detection module 2 is 180°±10°, and the vertical scanning angle is ±35°.

[0034] The above laser emits laser pulses, and through a specific deflection angle of the scanning mirror, a fan-shaped area in front is scanned. The large-range horizontal scanning of 180° generates a high-density point cloud, and voxel downsampling reduces the calculation amount by 40%-70% through regularization dimension reduction, while retaining key features. The vertical field of view covers the ground to about 15m in height, and the voxel resolution can be adaptively adjusted to balance accuracy and speed. The horizontal 180° ensures that the front half-circular area is covered without dead angles, and the vertical ±35° covers the key target height of pedestrians, vehicles, low-altitude obstacles, etc., reducing invalid data.

[0035] Optionally, the central processing unit 1 includes a point cloud processing subunit for three-dimensional reconstruction and safety distance calculation of point cloud data; based on the point cloud data of the laser radar, a high-precision three-dimensional model is generated through voxel downsampling, normal vector estimation, and surface reconstruction algorithm. Real-time point cloud registration is achieved by using ICP algorithm or NDT, and the minimum distance of the target and the key facility is calculated in combination with the dynamic threshold. Three-dimensional reconstruction can accurately restore the geometric structure of the scene, supporting modeling of complex environments. The safety distance calculation frequency reaches 50Hz, and the threshold is dynamically adjusted. Voxel downsampling can filter noise points, retain key features, simplify data, and reduce noise.

[0036] An image recognition subunit extracts features and identifies targets based on a convolutional neural network model. A lightweight CNN model is used to achieve 10ms-level inference on edge devices, extracting target texture, color, and shape features. Based on pre-trained model transfer learning, multi-class detection is supported, such as personnel, vehicles, and tools, with an average precision mAP of over 85%. The CNN can distinguish subtle features and reduce misjudgments, especially in complex backgrounds such as fog, haze, and shadows. Model quantization combined with hardware acceleration improves inference speed by 3 times and reduces power consumption by 50%, making it suitable for embedded device deployment. Online incremental learning updates the model to adapt to new targets without retraining all data.

[0037] A data fusion and decision subunit is used to fuse point cloud processing results and image recognition results to determine whether the target behavior constitutes a force destruction risk and generate a warning signal. Kalman filtering or Bayesian network is used to fuse point clouds and images to generate a unified scene representation. Based on a rule engine and a lightweight ML model, it is determined whether the target behavior constitutes a destruction risk.

[0038] Optionally, the image recognition subunit is preloaded with a power transmission line hidden object database to identify at least one large construction machinery in the tower crane, excavator, and crane.

[0039] In this embodiment, the image recognition subunit integrates multi-source remote sensing data, historical investigation data, and real-time monitoring data to build a comprehensive database covering tower cranes, excavators, cranes, and other construction machinery. According to deformation rate, cumulative settlement, maintenance status, and other indicators, hidden dangers are divided into four levels: extremely low, low, medium, and high, and corresponding scores are assigned to achieve quantitative management. Through multi-source data integration, the database covers the entire life cycle state of construction machinery, reducing the identification blind area. For example, combining the deformation rate interpreted by radar data with the direct interpretation of optical images, construction machinery and ordinary vehicles can be accurately distinguished.

[0040] According to the type of construction machinery and the voltage level of the power transmission line, the safety distance threshold is dynamically adjusted. Combined with the hidden danger level and distance threshold, yellow, orange, and red three-level warnings are triggered, and sound and light alarms and remote push are linked.

[0041] Optionally, the warning prompt module 4 includes a wireless loudspeaker for receiving instructions from the central processing unit 1 for remote shouting.

[0042] In this embodiment, low-power, long-range wireless communication technologies such as LoRa, 4G / 5G or Zigbee are used to ensure real-time data transmission with the central processing unit 1. Equipped with a 10W or more high-power speaker, supporting 120dB or more sound pressure level output, covering a radius of 100m or more. Integrated noise detection sensor, supports automatic volume adjustment. Wireless transmission delay <50ms, ensures that the warning instruction is broadcast within 1 second after the risk occurs, which is 90% faster than traditional manual shouting. 120dB sound pressure level can penetrate the noise of the construction site and complex terrain, ensuring that the on-site personnel can clearly receive the instructions.

[0043] In addition, through the data fusion result of the central processing unit 1, the shouting content is automatically generated and the loudspeaker is triggered. Different voice templates and volumes are called according to the risk level.

[0044] For example, a red warning triggers "Emergency evacuation! High-voltage line fracture risk!" accompanied by a 110dB high-frequency alarm.

[0045] Optionally, it also includes an environmental sensor 6 for collecting wind speed, temperature, humidity data; the central processing unit 1 is also used to dynamically adjust the warning distance threshold according to the environmental data.

[0046] The above-mentioned environmental sensor 6 integrates an anemometer, a temperature sensor and a humidity sensor, with a sampling frequency of 1Hz. Through the central processing unit 1, the environmental data is associated with the point cloud and image data to establish an environmental and risk model.

[0047] Specifically, when the wind speed is greater than 10m / s, the system automatically increases the safety distance threshold of tall machinery such as tower cranes and cranes to avoid swinging collisions caused by wind load.

[0048] When the temperature is below-20℃, the brittleness of the metal structure of the device increases, and the warning threshold is adjusted from 3m to 4m to prevent accidental contact caused by material shrinkage.

[0049] When the humidity is greater than 80%, the noise of the laser radar point cloud increases, and the system dynamically reduces the point cloud processing weight, increases the image recognition proportion, and ensures the reliability of the warning.

[0050] Embodiment 2 Please refer to Figure 2 The application also provides a power transmission line real-time monitoring system method, characterized by comprising the following steps: S1: scanning the power transmission channel by the infrared laser radar detection module 2 to obtain three-dimensional point cloud data; Using 180° horizontal scanning and ±35° vertical scanning, covering the semicircular area in front of the power transmission line, generating high-density point cloud.

[0051] Divide the three-dimensional space into 0.1m³ voxels, and take the center point to reduce the data volume.

[0052] Based on K=10 neighborhood density analysis, remove noise points outside the 1.5σ threshold.

[0053] S2: Collect environmental video data through the image acquisition module 3; Adopt 4K resolution camera, support 30fps video acquisition, equipped with wide dynamic range and infrared fill light, high resolution video provides texture, color characteristics of construction machinery, support CNN model accurate classification.

[0054] Through automatic exposure and automatic white balance adjustment, adapt to strong light, shadow, haze and other complex environment.

[0055] S3: Process point cloud data through central processing unit 1, calculate the minimum safety distance between target object and conductor; Based on the point cloud after voxel downsampling, generate three-dimensional model of power transmission line and equipment through Poisson surface reconstruction algorithm. Align real-time point cloud with conductor model, calculate minimum distance. Adjust threshold combined with environmental sensor 6 data.

[0056] S4: Identify video data through central processing unit 1, judge whether there is construction machinery; Adopt MobileNetV3 architecture, combined with power transmission line hidden danger object database, improve small target recognition rate through transfer learning. Combined with the spatial information of point cloud and the visual features of image.

[0057] S5: Fuse point cloud ranging result and image recognition result, if the risk target is recognized and its minimum safety distance from the conductor is less than the warning threshold, generate warning signal; Combine point cloud and image to generate unified scene representation. Based on rule engine and lightweight ML model, judge whether the target behavior constitutes risk.

[0058] S6: Start sound and light alarm and / or voice driving reminder through warning prompt module 4; Adopt LoRa / 4G communication, support 120dB sound pressure level output, coverage radius greater than 100m, integrated noise detection sensor automatically adjusts volume. Link with LED stroboscopic lamp and laser warning lamp to form sound and visual double warning.

[0059] S7: Send alarm information and related data to monitoring platform and mobile terminal through communication module 5.

[0060] Adopt 4G / 5G or Zigbee, support data compression and encryption, ensure real-time and security. Push alarm information to monitoring platform and mobile terminal through API interface.

[0061] Optionally, the step of processing point cloud data comprises: S101: Denoising and filtering preprocessing of original point cloud data; Outliers are removed by calculating the average distance of each point to its neighborhood. Assuming that the local region of a point cloud should have consistency and smoothness, statistical filtering calculates the average and standard deviation of the distance between each point and its near neighbors, identifying and removing points that exceed a threshold.

[0062] The point cloud space is divided into cubes of equal size, and the average or centroid of all points in each voxel is used to replace these points. This method can compress the amount of point cloud data, smooth the point cloud, while preserving the overall geometric structure.

[0063] Using a weighted average method, the point cloud is smoothed using Gaussian distribution weights to reduce high-frequency noise.

[0064] S102: Register the preprocessed point cloud with the known power line three-dimensional model; The best transformation parameters are found by iteratively minimizing the distance error between point clouds to achieve point cloud alignment. The algorithm steps include initial alignment, nearest point matching, transformation matrix calculation, transformation application and error calculation, and convergence judgment. Ensure that the point cloud data and the three-dimensional model are accurately aligned in space, improving the accuracy of subsequent processing. Provide accurate spatial reference for subsequent segmentation, clustering and distance calculation steps.

[0065] S103: Segment and cluster non-ground point clouds to extract individual targets; The point cloud is divided into different clusters according to the distance between points. A seed point is selected as the starting point of the current cluster, and all unclassified points are traversed to calculate their distance from the seed point. Points with a distance less than a set threshold are classified into the same cluster.

[0066] Based on similarity criteria, points with high similarity are clustered together to form continuous regions. Starting from a seed point, recursively add neighbor points that meet the similarity condition to the current region.

[0067] S104: Calculate the minimum spatial distance of each target point cloud cluster from the conductor three-dimensional model.

[0068] The minimum distance is found by calculating the Euclidean distance between each target point cloud cluster and the conductor three-dimensional model. All pairs of points are traversed to find the nearest pair and calculate their distance.

[0069] Optionally, the S101 step specifically includes: using a statistical outlier removal algorithm to calculate the average distance and standard deviation of each point from its neighbors, and removing points that exceed a set multiple of the standard deviation to filter out discrete noise; and / or, using a voxel grid downsampling method, dividing the point cloud space into a three-dimensional voxel grid, and using the center of gravity of all points in each voxel to represent the voxel, to reduce the data volume while preserving the shape features of the point cloud.

[0070] Statistical outlier removal algorithm, for each point in the point cloud, selects its K nearest neighbors (KNN) algorithm. Calculate the average distance (μ) and standard deviation (σ) of all neighboring points. Set a threshold, if the average distance of a point is more than μ + 2.5σ, it is determined as an outlier and removed. Repeat the above process, gradually filter out multi-layer noise.

[0071] Alternatively, the voxel grid downsampling method divides the point cloud space into a three-dimensional voxel grid.

[0072] Centroid calculation: for each non-empty voxel, calculate the geometric center of mass or centroid of all points inside. Replace all original points in the voxel with the centroid point to form the downsampled point cloud. By changing the voxel edge length, balance the data volume and detail retention.

[0073] Optionally, the S102 step specifically comprises: obtaining the tower coordinates of the power transmission line and the conductor sag parameters from the database; Based on the parameters, generate high-precision three-dimensional models of the conductor and the tower through catenary equation or parabolic equation; Using the iterative closest point algorithm, the real-time collected point cloud data is registered with the high-precision three-dimensional model to realize accurate alignment of the spatial coordinate system.

[0074] Extract the pre-stored power transmission line parameters from the database of the power company, including: tower coordinates, conductor sag parameters and data verification. Specifically, the tower coordinates are obtained by GPS or total station measurement. The conductor sag parameters include conductor type, span, tension, temperature, etc., which are used to calculate the natural sag curve of the conductor under the action of gravity. Data verification is a complete integrity check and consistency check of the parameters to ensure data reliability. Directly use the real measurement data to avoid repeated field collection and improve the model generation efficiency. Support line model generation of different voltage levels, adapt to complex terrain.

[0075] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A real-time monitoring system for power transmission lines, characterized in that, The application relates to an infrared laser radar detection module (2) for three-dimensional scanning of a power transmission line channel to generate point cloud data, an image acquisition module (3) for real-time acquisition of video images of the environment around the power transmission line, a central processing unit (1) electrically connected to the infrared laser radar detection module (2) and the image acquisition module (3) for receiving and fusing point cloud data and video images, target recognition, distance calculation and behavior analysis, and generation of a warning signal, a warning prompt module (4) electrically connected to the central processing unit (1) for issuing an audible and / or visual alarm and / or voice warning when receiving the warning signal, and a communication module (5) electrically connected to the central processing unit (1) for remote transmission of warning information and related data to a background monitoring platform and / or a mobile terminal. The horizontal scanning angle of the infrared laser radar detection module (2) is 180 DEG+ / -10 DEG, and the vertical scanning angle is + / -35 DEG. The central processing unit (1) comprises a point cloud processing subunit for three-dimensional reconstruction and safety distance calculation of point cloud data, an image recognition subunit for feature extraction and target recognition of video images based on a convolutional neural network model, and a data fusion and decision subunit for fusing point cloud processing results and image recognition results to determine whether the target behavior constitutes an external force damage risk and generate a warning signal. The image recognition subunit is preloaded with a power transmission line hidden object database for identifying at least one large construction machine among tower cranes, excavators and cranes. The warning prompt module (4) comprises a wireless loudspeaker for receiving instructions from the central processing unit (1) to perform remote shouting. The application further comprises an environment sensor (6) for collecting wind speed, temperature and humidity data, and the central processing unit (1) is further configured to dynamically adjust a warning distance threshold value according to the environment data.

2. A real-time monitoring system for power transmission lines according to claim 1, characterized in that, The application comprises the following steps:

3. A real-time monitoring system for power transmission lines according to claim 1, characterized in that, S1: scanning a power transmission channel by the infrared laser radar detection module (2) to obtain three-dimensional point cloud data; S2: collecting environment video data by the image acquisition module (3); S3: processing the point cloud data by the central processing unit (1) to calculate the minimum safety distance of a target object from a conductor; S4: identifying whether there is a construction machine by the central processing unit (1) by processing video data; 4. A real-time monitoring system for power transmission lines according to claim 3, characterized in that, S5: fusing point cloud distance calculation results and image recognition results, and generating a warning signal if a risk target is identified and the minimum safety distance of the target from the conductor is less than a warning threshold value; 5. A real-time monitoring system for power transmission lines according to claim 1, characterized in that, S6: starting audible and / or visual alarm and / or voice driving warning by the warning prompt module (4); 6. A real-time monitoring system for power transmission lines according to claim 5, characterized in that, S7: sending alarm information and related data to a monitoring platform and a mobile terminal by the communication module (5). The processing of the point cloud data comprises the following steps:

7. A method of a power line monitoring system based on the power line monitoring system according to any one of claims 1 to 6, characterized by, S101: denoising and filtering preprocessing of original point cloud data; S102: registration of the preprocessed point cloud and a known three-dimensional model of a power transmission line; S103: segmentation and clustering of non-ground point clouds to extract individual targets; S104: calculation of the minimum spatial distance of each target point cloud cluster from a three-dimensional model of a conductor. ​ ​ ​ ​ 8. A method of a real-time monitoring system of a power transmission line according to claim 7, characterized in that, ​ ​ ​ ​ ​ 9. A method of a real-time monitoring system of a power transmission line according to claim 8, characterized in that, The S101 step specifically comprises: using a statistical outlier removal algorithm to calculate the average distance and standard deviation of each point and its adjacent points, and removing points exceeding a set standard deviation value multiple to filter out discrete noise; and / or using a voxel grid downsampling method to divide the point cloud space into a three-dimensional voxel grid, and using the center of gravity of all points in each voxel to represent the voxel, so as to reduce the data amount while maintaining the shape characteristics of the point cloud.

10. The method of claim 8, wherein, The S102 step specifically comprises: obtaining the tower coordinates and conductor sag parameters of the power transmission line from a database; Based on the parameters, a high-precision three-dimensional model of the conductor and the tower is generated through a catenary equation or a parabolic equation; The real-time collected point cloud data is registered with the high-precision three-dimensional model by using an iterative closest point algorithm, so as to realize accurate alignment of the spatial coordinate system.

Images