Power transmission tower edge multi-device linkage method based on ad hoc network and two-dimensional and three-dimensional fusion

Abstract

The invention discloses a power transmission tower edge multi-device linkage method based on an ad hoc network and two-dimensional and three-dimensional fusion, and relates to the technical field of intelligent power transmission networks, and the method comprises the steps: deploying multiple devices, embedding a passive sensor, and employing an unmanned plane to carry a laser radar to generate a tower three-dimensional space reference model; an unmanned aerial vehicle autonomous inspection route is planned based on the iron tower three-dimensional space reference model, laser point cloud, infrared thermal imaging and backscattering data are synchronously collected, passive sensor deformation parameters are extracted by analyzing channel state information, and a multi-dimensional data set is constructed; fusing the multi-dimensional data set, generating a dynamic risk thermodynamic diagram and labeling risk levels; and when a high risk level is detected, triggering an equipment deployment strategy and recording operation data. According to the invention, by constructing a mechanical-thermal-geometric multi-field coupled power transmission tower dynamic monitoring system, a monitoring normal form with high-precision positioning, real-time response and extremely simple operation and maintenance is provided for an intelligent power grid, and intelligent and unmanned evolution of power transmission facility management is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent power transmission networks, and particularly to a method for multi-device linkage at the edge of transmission towers based on self-organizing networks and two- and three-dimensional fusion. Background Art

[0002] In recent years, the intelligent monitoring technology of transmission towers has gradually developed into a fusion system of multi-source perception and edge computing. Existing technologies mostly combine fixed sensor networks (such as strain gauges and fiber Bragg gratings) with periodic UAV inspections, and abnormal identification is realized through a cloud big data platform. At the data acquisition level, lidar and infrared thermal imaging technologies have been maturely applied to the detection of tower deformations and temperature fields, and self-organizing network technologies based on ZigBee and LoRa provide a feasible solution for device interconnection in remote areas. However, there are significant bottlenecks in existing technologies: firstly, traditional sensors rely on wired power supply or battery maintenance, and are vulnerable to electromagnetic interference and have high deployment costs in the complex metal structure scenario of transmission towers; secondly, multi-dimensional data (mechanics, thermotics, geometry) lack a unified benchmark for spatio-temporal alignment, resulting in limited fault location accuracy and difficulty in triggering precise linkage responses in a timely manner. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides a method for multi-device linkage at the edge of transmission towers based on self-organizing networks and two- and three-dimensional fusion to solve the problems of spatio-temporal misalignment of multi-source heterogeneous data of transmission towers, insufficient fault location accuracy, and high device linkage response delay.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a method for multi-device linkage at the edge of transmission towers based on self-organizing networks and two- and three-dimensional fusion, which includes deploying multiple devices at the edge of the transmission tower, embedding passive sensors based on ambient backscattering at the same time, and using a UAV carrying lidar to generate a three-dimensional spatial reference model of the tower; Based on the three-dimensional spatial reference model of the tower, plan the autonomous inspection route of the UAV, synchronously collect point cloud data, infrared thermal imaging, and backscattering data, extract the deformation parameters of the passive sensors by analyzing the channel state information, and construct a multi-dimensional data set; Fuse the multi-dimensional data set at the edge node, generate a dynamic risk heat map and mark the risk level; When a high risk level is detected, trigger the device deployment strategy, record the operation data, and update the three-dimensional spatial reference model of the tower.

[0006] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the multi-device deployment includes deploying an edge computing node at the bottom of the tower, integrating a ZigBee / Wi-Fi dual-mode ad-hoc network communication tool, and physically connecting it to an embedded airport management system.

[0007] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the three-dimensional spatial reference model of the tower is obtained by flying a drone equipped with a Velodyne VLP-16 lidar along a calibration path to collect the point cloud data of the tower, aligning the point cloud data with the BIM model coordinate system through the ICP algorithm to generate a fusion coordinate system, mapping the installation position coordinates of the passive sensor to the BIM model coordinate system, and obtaining the three-dimensional spatial reference model of the tower.

[0008] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the autonomous inspection route of the planned drone is based on the three-dimensional spatial reference model of the tower, and by setting a horizontal safety distance and a vertical layered scanning path, a drone circular route is dynamically generated, and a real-time obstacle avoidance strategy is integrated.

[0009] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the channel state information includes signal strength, phase offset, propagation time difference, and subcarrier index.

[0010] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the deformation parameters of the passive sensor include cantilever beam displacement and bolt loosening angle.

[0011] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: generating a dynamic risk heat map and marking the risk level is to use an AI model to associate the three-dimensional spatial reference model of the tower with the fused multi-dimensional data set, calculate the risk value, divide the risk level, and generate a dynamic risk heat map.

[0012] As a preferred solution of the method for multi-device linkage at the edge of a transmission tower based on ad-hoc network and two- and three-dimensional fusion according to the present invention, wherein: the device deployment strategy includes deploying a drone to approach the target position to collect high-definition images, scheduling an inspection robot to perform on-site maintenance, and simultaneously correcting the three-dimensional spatial reference model of the tower.

[0013] In a second aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for multi-device linkage at the edge of a transmission tower based on ad hoc network and two- and three-dimensional fusion as described in the first aspect of the present invention is implemented.

[0014] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for multi-device linkage at the edge of a transmission tower based on ad hoc network and two- and three-dimensional fusion as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: By constructing a dynamic monitoring system for transmission towers that couples mechanics, thermodynamics, and geometry, first, the passive sensor based on ambient backscattering analyzes the displacement of the cantilever beam through phase modulation, and combines the coordinate mapping of the three-dimensional space reference model to reduce the detection error of bolt loosening angle and accurately lock the microscopic defects in space positioning. Secondly, relying on edge computing and ad hoc network communication, the AI model fuses multi-dimensional data to generate a dynamic risk heat map, and triggers the drone to approach for shooting and the robot to perform precise maintenance, forming a perception-decision-execution closed loop to improve the response efficiency. In addition, the passive sensor eliminates the need for battery maintenance through radio frequency power supply; the ad hoc network dynamic routing protocol maintains a high connectivity rate in case of single-node failure, which is significantly better than the fixed topology scheme. It provides a monitoring paradigm for the smart grid with high-precision positioning, real-time response, and extremely simple operation and maintenance, and promotes the evolution of transmission facility management towards intelligence and unmanned operation. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts.

[0017] Figure 1 It is a flowchart of the method for multi-device linkage at the edge of a transmission tower based on ad hoc network and two- and three-dimensional fusion in Embodiment 1.

[0018] Figure 2 It is a flowchart for generating a three-dimensional space reference model of the transmission tower in Embodiment 1.

[0019] Figure 3 It is a logic diagram for generating a risk heat map in Embodiment 1.

[0020] Figure 4 It is a data fusion processing diagram in Embodiment 1. Detailed Embodiments

[0021] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification.

[0022] In the following description, numerous specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0023] Secondly, as used herein, an "embodiment" or "embodiments" refer to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that mutually excludes other embodiments.

[0024] Embodiment 1, referring to Figures 1 to 4 , this embodiment provides a method for multi-device linkage at the edge of transmission towers based on ad-hoc networks and two- and three-dimensional fusion, including the following steps: S1. Deploy multiple devices at the edge of the transmission tower, while embedding passive sensors based on ambient backscattering, and use a drone equipped with a lidar to generate a three-dimensional spatial reference model of the tower.

[0025] S1.1. The multi-device deployment includes deploying an edge computing node at the bottom of the tower, integrating a ZigBee / Wi-Fi dual-mode ad-hoc network communication tool (transmission power 20 dBm, coverage radius 500 m), and physically connecting it to an embedded airport management system (with an in-built drone takeoff and landing platform and a wireless charging pile). It should be noted that the passive sensor captures energy and modulates backscattered data by coupling an external radio frequency signal. At key positions such as bolt connections and wire joints of the transmission tower, passive sensors based on ambient backscattering are embedded at a spacing of 3 mm. The surface of the passive sensor is coated with a metal layer of the same material as the tower to couple Wi-Fi signals. Each passive sensor is equipped with a micro dipole antenna, and the antenna length is adjusted to the resonant wavelength (31.25 mm) of the 2.4 GHz frequency band. The phase offset of the modulated data is changed by changing the displacement of the cantilever beam (accuracy ±0.1 μm), and the phase displacement is linearly related to the bolt loosening angle.

[0026] S1.2. The three-dimensional spatial reference model of the tower is obtained by flying a drone equipped with a Velodyne VLP-16 lidar along a calibrated path to collect point cloud data of the tower, aligning the point cloud data with the coordinate system of the BIM model (Building Information Model) (prior art) through the ICP algorithm to generate a fused coordinate system, and mapping the installation position coordinates of the passive sensors to the BIM model coordinate system to obtain the three-dimensional spatial reference model of the tower; It should be noted that the DJI Matrice 300 RTK drone is equipped with a Velodyne VLP-16 lidar (16-beam, vertical field of view ±15°, horizontal accuracy ±3 cm). The scanning frequency is set to 10 Hz, and the point cloud density is 300,000 points per second. The tower design drawings (format: IFC) are imported into the BIM model coordinate system. Redundant components are cleared, and the geometric and attribute data of the main materials (angle steel, bolts) are retained. The coordinate system is converted to WGS84 UTM. The outer contour line of the tower is extracted in the BIM model coordinate system, and a safety flight boundary is generated by expanding it outward by 20 m. The flight path is planned using the equal-height layer cutting method. The equal-height layer cutting method for flight path planning takes an interval of every 5 m of the tower height (such as 0 m, 5 m, 10 m...). A closed polygon path (radius = tower width + 10 m) around the tower is generated for each layer. The flight height error is controlled within ±0.5 m. The flight speed of the drone is set to 3 m / s, and the tilt angle of the lidar is adjusted to -10°. The drone flies along the calibrated path, and the lidar outputs point cloud data at a frequency of 10 Hz, records the timestamp in real time, enables the intensity correction function, and filters out noise points according to the material reflectivity. It should be noted that the DJI Matrice 300 RTK drone integrates high-precision RTK (Real-Time Kinematic) technology and is designed specifically for complex mission scenarios such as inspection, surveying, and security. The principle of RTK technology is to eliminate satellite signal errors through real-time differential correction between the ground reference station and the drone, significantly improving the positioning accuracy. RTK + GNSS dual redundant system: supports multiple satellite systems such as GPS, GLONASS, Galileo, and Beidou, with a positioning accuracy of centimeter level (horizontal ±1 cm, vertical ±1.5 cm). The Velodyne VLP-16 lidar uses 16 vertically distributed laser emitters, with a vertical field of view of ±15° (a total of 30°) and a horizontal field of view of 360°. It outputs 300,000 point cloud data per second (300,000 points / sec), and the horizontal angular resolution is adjustable from 0.1° to 0.4°.

[0027] The drone outputs positioning data, which is aligned with the lidar timestamp through the NMEA-0183 protocol. Five control points (the four corners and the center of the tower foundation) are set on the ground, and reflective markers are pasted. The precise coordinates (error ±2 mm) are measured using a total station (such as Leica TS16) as the point cloud registration reference. Voxel filtering (voxel size 0.05 m) is used to downsample the point cloud registration reference, retaining the main structure points of the iron tower (removing interfering points such as vegetation and birds). The statistical outlier removal algorithm is applied to calculate the average distance of 50 points within the neighborhood (radius 0.1 m) of each main structure point of the iron tower, and the points exceeding the mean + 3σ (σ is the standard deviation of the distance) are removed. The seven-parameter transformation method (3 translations + 3 rotations + 1 scaling) is used to calculate the transformation matrix through the control point coordinates, and the transformation matrix is applied to map all point cloud data to the BIM model coordinate system; The iterative closest point (ICP) algorithm is used for registration in two stages, including coarse registration and fine registration. For coarse registration, the end points of the angle steel of the BIM model are selected as feature points and matched with the point cloud data by RANSAC (maximum number of iterations 1000). For fine registration, the result of coarse registration is used as the initial value, and the full point cloud is iterated (maximum number of iterations 50, convergence condition: translation < 0.01 m, rotation < 0.1°); It should be noted that RANSAC matching means randomly selecting 3 pairs of BIM angle steel end points and suspected angle steel end points in the point cloud, calculating the rigid body transformation (rotation + translation), verifying the number of inliers (inlier threshold set to 5 cm), reaching 1000 iterations or finding a transformation matrix with an inlier rate ≥ 90%, and outputting the optimal coarse registration result; full point cloud iteration is based on the result of coarse registration, constructing a point cloud KD-Tree index, quickly matching the corresponding points between the BIM model and the point cloud, assigning higher weights to the high-confidence regions (such as the surface of the angle steel) in the point cloud, reducing the influence of noise points, and terminating when the translation change < 0.01 m and the rotation change < 0.1° in 10 consecutive iterations, or reaching the upper limit of 50 iterations.

[0028] Semantically associate the registered point cloud data with BIM model components (such as bolt holes and cross arms), assign point cloud attribute labels (such as tower leg angle steel L3, insulator support), and generate a fusion coordinate system; It should be noted that based on geometric features (such as curvature, normal vector, density), the registered point cloud is segmented to extract local point cloud clusters corresponding to BIM components (such as cylindrical point cloud of bolt holes, planar point cloud of cross arms); the target components (such as the center point of the cross arm, axis of the bolt hole) are located according to the coordinates in the BIM model, and the bounding box, geometric parameters and attribute information (such as component ID, material specification) are obtained; the segmented local point cloud clusters are spatially aligned with the BIM components (such as fine-tuning the position through the ICP algorithm), and the geometric similarity (such as diameter, angle) is verified; the attributes of the BIM components (such as "tower leg angle steel L3", "insulator support - model X") are associated with the corresponding local point cloud clusters to generate point cloud attribute tags (including component name, function classification, maintenance record); based on the BIM model coordinate system, the point cloud data is transformed to the same coordinate system by applying the point cloud registration reference (rotation matrix, translation vector) to generate a fused coordinate system; an association index between the point cloud data and the BIM model is established to support the automatic synchronization of tags and spatial relationships when new point cloud data or BIM model modifications are made subsequently.

[0029] When embedding passive sensors, use an RTK rover (such as Trimble R12, Trimble R12 Real - Time Kinematic Positioning Measurement System) to measure the three - dimensional coordinates of the installation location, record the passive sensor ID, map the measured three - dimensional coordinates of the installation location to the BIM model coordinate system through a transformation matrix, and update the fused coordinate system based on the passive sensor coordinates through automatic matching and manual correction to generate a three - dimensional spatial reference model of the iron tower; It should be noted that automatic matching is to search for the nearest BIM component (such as a bolt hole) within a 5 - cm radius in the BIM model coordinate system and bind the passive sensor to this component; manual correction is to manually mark the coordinates and associate the component in the BIM model coordinate system when automatic matching fails (such as when the passive sensor is located on a non - standard part).

[0030] It should be noted that passive sensors based on ambient backscattering are embedded at key positions such as bolts and wire joints, powered by Wi-Fi signals and modulating the phase offset to achieve micron-level deformation monitoring (the displacement accuracy of the cantilever beam is ±0.1μm); the drone is equipped with a lidar and flies along the calibrated path. The point cloud is aligned with the BIM model through the ICP algorithm, and the sensor coordinates are mapped by combining RTK positioning to form a fusion coordinate system; the passive sensor eliminates the battery dependence through radio frequency power supply, and the surface is plated with a metal layer of the same material as the iron tower to improve the signal coupling efficiency, enabling maintenance-free operation for more than 10 years in harsh environments (traditional battery sensors need to be replaced every 1-2 years); through the semantic registration of the point cloud data and the BIM model (error <5cm), a unified coordinate system is provided for subsequent data fusion, solving the problem of spatial misalignment of traditional multi-source data (the error of traditional methods >1m); the combination of automatic matching (searching for BIM components within a 5cm radius) and manual correction ensures the accurate mapping of the coordinates of passive sensors for non-standard components (such as reinforcement brackets), and the adaptation rate is increased from 70% to 98%.

[0031] S2. Based on the three-dimensional spatial reference model of the iron tower, plan the autonomous inspection route of the drone, synchronously collect lidar point cloud, infrared thermal imaging and backscattering data, extract the deformation parameters of the passive sensor by analyzing the channel state information, and construct a multi-dimensional data set.

[0032] S2.1. Planning the autonomous inspection route of the drone is based on the three-dimensional spatial reference model of the iron tower. By setting the horizontal safety distance and the vertical layer scanning path, the drone's surrounding route is dynamically generated, and a real-time obstacle avoidance strategy is integrated.

[0033] It should be noted that according to the height H of the iron tower (unit: meter), the horizontal safety distance is set to 1.2H. For example, when the iron tower is 50 meters high, the flight radius of the drone remains 60 meters. The vertical direction is divided into scanning layers every 5 meters. The drone flies at a constant speed of 3m / s. The flight path of each layer is a closed spiral around the iron tower (pitch 5 meters). Based on the known obstacles (such as adjacent iron towers and trees) in the three-dimensional spatial reference model of the iron tower, the AI algorithm is used to plan the local obstacle avoidance path in real time, and the minimum obstacle avoidance distance is 2 meters.

[0034] S2.2. The channel state information is obtained by analyzing the physical layer characteristics of the wireless communication signal, including signal strength, phase offset, propagation time difference and subcarrier index; The deformation parameters of the passive sensor include the displacement of the cantilever beam and the loosening angle of the bolt; Multi-sensor collaborative acquisition of point cloud data, infrared thermal imaging and backscattering data; Specifically, the Velodyne VLP-16 radar operates in a 10-line scanning mode, with a vertical angular resolution of 2°, a horizontal angular resolution of 0.1°, a point cloud density of ≥300 points per square meter, a data output frequency of 10 Hz. The FLIR T1020 camera is set with a temperature range of -20°C to 150°C, a thermal sensitivity of 0.05°C, a frame rate of 30 fps, and is triggered synchronously with the lidar hardware. In night mode, an 850 nm near-infrared LED array (power 10 W, illumination distance 30 m) is enabled, the camera exposure time is 1 / 60 s, and the ISO is 1600; The edge node sends 802.11n beacon frames (center frequency 2.432 GHz, bandwidth 20 MHz) at a 10 ms period, with a transmit power of 20 dBm and a coverage radius of 50 m. The passive sensor changes the antenna impedance through cantilever beam displacement (range ±0.5 mm, resolution 0.01 mm), encodes the displacement data as the phase offset of the reflected signal, and the modulation depth is linearly related to the displacement; The edge node uses an Intel 5300 network card to capture channel state information, extracts the phase values of 30 subcarriers, and calculates the phase difference between adjacent subcarriers; Based on the phase difference between adjacent subcarriers, calculate the cantilever beam displacement, expressed as, ; where, represents the cantilever beam displacement, represents the phase difference between adjacent subcarriers, represents the Wi-Fi wavelength, = 12.5 cm; Using the cantilever beam displacement, calculate the bolt loosening angle, expressed as, ; where, represents the bolt loosening angle, represents the effective length of the bolt (for example when = 10 cm, = 0.2 mm corresponds to ≈1.15°); Convert the point cloud data into a voxel grid (voxel size 0.1 m × 0.1 m × 0.1 m), each voxel stores the point cloud density, the mean reflection intensity, and the maximum height difference. Crop the infrared thermal image into a 512×512 pixel area, normalize the temperature value to [-1,1], extract the local temperature gradient (such as calculating the temperature difference with a 3×3 sliding window), group the backscatter data by passive sensor ID, and calculate the sliding window mean (window length 1 s, step size 0.1 s) of the bolt loosening angle ; Combine the laser point cloud voxels, infrared pixels, and passive sensors The values are interpolated and aligned according to the timestamp with a time window of 10 ms to ensure the spatial consistency of data at the same moment (for example, the temperature and deformation data corresponding to the bolt coordinates (10.2, 5.3, 15.0)), and are mapped to the 3D spatial reference model of the iron tower to form a multi-dimensional data set.

[0035] It should be noted that based on the 3D spatial reference model of the iron tower, a horizontal safety distance (1.2H) and vertical stratified scanning (at 5 m intervals) are set, and AI performs real-time obstacle avoidance; the lidar (10 Hz point cloud), infrared camera (0.05 °C thermal sensitivity), and backscatter data (analyzed at a 10 ms cycle) are hardware-level synchronized and triggered; the 360° surface of the iron tower is covered by a closed spiral flight path (pitch 5 m), and the point cloud density is ≥ 300 points / m², reducing the omission rate by 80% compared with traditional single-view scanning; based on the phase difference calculation of the channel state information (CSI) (Δφ = 2πd / λ), the parsing delay of the cantilever beam displacement is compressed to 10 ms; voxel filtering (0.05 m³) and statistical outlier removal (3σ threshold) are used to filter vegetation and bird interference, and the proportion of valid data is increased from 60% to 95%.

[0036] S3. Integrate the multi-dimensional data set at the edge node, generate a dynamic risk heat map, and mark the risk level.

[0037] Use the AI model to associate the 3D spatial reference model of the iron tower with the integrated multi-dimensional data set, calculate the risk value, divide the risk level, and generate a dynamic risk heat map; Use a 3D convolutional layer (kernel size 3×3×3), input the voxel grid, and extract spatial structure features; It should be noted that the first-layer 3D convolution kernel (3×3×3) scans the voxel grid, and through weight learning, it identifies local geometric mutations, including hole responses and connector identifications. The hole response is that the voxel density drops suddenly in the area with high activation values (such as missing bolt holes and structural damage); the connector identification is to capture the voxel patterns of small-scale regular geometries such as bolt heads and angle steel end faces, and the ReLU activation function and batch normalization (BatchNorm) are superimposed to suppress the interference of noise voxels. By stacking multiple layers of 3D convolutions (such as 4 - 6 layers), the physical range associated with a single voxel is gradually expanded. Level 1: Identify the orientation of a single angle steel (such as horizontal / vertical), Level 2: Locate the intersection of multiple angle steels (such as the tower leg gusset plate connection area), Level 3: Map the topological structure of the overall framework (such as the quadrilateral base of the transmission tower and the symmetry of the cross-arm distribution); in the deep layer, strided convolution (such as stride = 2) is used to compress the spatial dimension and retain high-order semantics (such as the uniformity of bolt distribution and key stress nodes), and the high-resolution detail features in the shallow layer (such as the bolt hole position) and the global structure features in the deep layer (such as the tower inclination) are concatenated along the channel dimension to form spatial structure features.

[0038] Use a 2D convolutional layer (kernel size 5×5), input the temperature gradient map, and extract thermodynamic features; It should be noted that the 5×5 convolutional kernel learns the local temperature gradient, strengthens the temperature mutation boundary (such as the transition zone between the bolt loosening area and the normal area), filters the environmental thermal noise through batch normalization (BatchNorm) and activation functions (such as ReLU), outputs the hot spots on the device surface corresponding to the high-activation area (such as the bolt connection with local high temperature), marks the suspected abnormal positions, and uses the features such as bolt positions and weld morphologies extracted by the spatial structure branch (3D convolution) as the attention mask (AttentionMask) to guide the thermal convolution to focus on the relevant areas, and establishes long-distance thermal correlations (such as the abnormal temperature symmetry at both ends of the cross arm indicating structural stress imbalance) through deep convolutional stacking (such as ResNet blocks), thus forming thermodynamic features.

[0039] Use the LSTM layer, input the value sequence, and extract the time-series deformation features; It should be noted that the LSTM unit receives displacement / strain values at each time step, regulates the information flow through the forget gate, input gate, and output gate, retains the long-term trend (such as the quarterly bridge deflection change), and at the same time inputs multi-channel data such as displacement, strain, and temperature (environmental interference term), learns the cross-influence (such as the increase in temperature causing the expansion deformation of the steel beam), the LSTM captures the pattern of the strain value continuously rising and the rate increasing, and combines the historical value sequence to predict the critical fracture time, and identify the periodic oscillation of the bridge deflection caused by the day-night temperature difference (daytime expansion, nighttime contraction), thus forming the time-series deformation features.

[0040] Concatenate the output feature vectors of the three branches into a fusion feature (dimension: 1024); It should be noted that the spatial-attribute attention mechanism is used to assign weights. For example, assign a weight of 0.7 to the passive sensor value and a weight of 0.3 to the infrared temperature for the bolt area, and assign a weight of 0.6 to the infrared temperature and a weight of 0.4 to the point cloud density for the wire area; Collect the historical laser point cloud (500GB), historical infrared images (200,000), and historical passive sensor time-series data ( and infrared temperature ΔT records) of the transmission tower, covering three types of scenarios: normal, low-risk, and high-risk. Input the historical laser point cloud voxels (spatial density, reflection intensity), historical infrared temperature gradient map, and historical value sequence, and train the AI model using the weighted cross-entropy loss; Input the fusion feature into the fully connected layer (512 nodes) of the trained AI model, the output layer outputs the risk value, and according to the mechanical parameter threshold and thermodynamic parameter threshold, divide the level into three categories, including normal, low-risk, and high-risk; Specifically, high risk: > 1.2° or infrared temperature ΔT > 10°C; Low risk: > 0.8° or infrared temperature ΔT > 5°C; Normal: ≤ 0.8° and infrared temperature ΔT ≤ 5°C; It should be noted that based on the design specification of the pre - tightening force of transmission tower bolts (such as DL / T 646 - 2012), when the loosening angle exceeds 30% of the material yield strength, the structural stability decreases significantly. It shows that when θ = 1.2°, the loss of bolt pre - tightening force reaches 40%, reaching the critical failure threshold. Therefore, the high - risk threshold of mechanical parameters is set at 1.2°. According to historical operation and maintenance records, In the range of 0.8° - 1.2°, the failure rate shows an exponential growth. It needs attention but is not a high risk. Therefore, the low - risk threshold of mechanical parameters is set at 0.8°; based on the expansion coefficient of wire joint materials (such as aluminum) being , a temperature rise of 10°C will cause the contact resistance to increase by 15%, triggering the risk of local overheating and melting. Therefore, the high - risk threshold of thermodynamic parameters is set at 10°C. Based on the national standard GB / T2314, a wire temperature rise of 5°C is the starting point of abnormal temperature rise and needs continuous monitoring. Therefore, the low - risk threshold of thermodynamic parameters is set at 5°C.

[0041] Map the classification results to the three - dimensional space reference model of the tower according to three - dimensional coordinates, and mark the risk level for each voxel (RGB color coding: red - high risk, yellow - low risk, green - normal); Based on the finite - element stress distribution model (existing technology), for the influence range extended outward from the high - risk area (such as the area where the stress of adjacent components increases due to bolt loosening), with the high - risk voxel as the center, calculate the stress attenuation radius according to the material properties (such as the elastic modulus of steel is 210 GPa) (for example, the stress drops by 50% within a radius of 2 m). Within the attenuation radius, linearly reduce the risk level according to the distance (for example, mark as high risk at a distance of 0.5 m and drop to low risk at 1 m). For the voxels affected by stress diffusion, superimpose the original risk level and the diffusion weight (such as the risk level is increased by one level within 1 m around the high - risk voxel). If the diffusion causes the new area to reach the mechanical parameter threshold and the thermodynamic parameter threshold (such as the original normal area becomes low risk due to stress propagation), trigger recalculation.

[0042] It should be noted that the three - branch fusion of 3D convolution (spatial structure), 2D convolution (thermodynamics) and LSTM (temporal deformation) dynamically allocates weights through the spatial - attribute attention mechanism; based on the finite - element stress propagation model to expand the risk area, calculate the stress attenuation radius according to the material properties (such as the elastic modulus of steel is 210 GPa); the bolt area is given Weight 0.7, temperature weight 0.3, wire area temperature weight 0.6, false alarm rate reduced by 40% compared with traditional single-threshold determination; centered on high-risk voxels, dynamically label the associated area according to the stress attenuation radius (such as the risk level linearly attenuates within a radius of 2m), and the early warning coverage rate is increased by 50%.

[0043] S4. When a high-risk level is detected, trigger the device deployment strategy, record the operation data, and update the three-dimensional spatial reference model of the iron tower.

[0044] The edge node receives the high-risk alarm output by the AI model (such as the bolt loosening angle =1.3°, temperature ΔT = 12°C), automatically retrieves the historical data of the coordinates (the change trend in the past 24 hours and temperature records), and verifies whether it is a persistent anomaly (such as continuous growth > 0.1° / minute); If it is confirmed to be valid, start the priority sorting, and assign task weights according to the criticality of the iron tower structure (such as main materials > auxiliary materials) and the risk level (high risk > low risk > normal) (such as the weight of main material bolt loosening = 10, auxiliary material = 3); Broadcast the event coordinates (such as (8.7, 3.5, 12.0)) and task instructions through the self-organizing network. After receiving the instructions, the drone and the inspection robot feedback their current status (battery power, position, task queue). The edge node selects the optimal device to execute the task, including the drone selection criteria and the robot selection criteria. The drone selection criteria are the closest to the target point (<200m), battery power > 40%, and no other high-priority tasks. The robot selection criteria are that the path to the target point is obstacle-free and the robotic arm tool is matched (such as a torque wrench suitable for M20 bolts); The drone plans the approach path based on the three-dimensional spatial reference model of the iron tower, including the global path and the end path. The global path is a B-spline curve from the current hovering point to the target point, with a horizontal obstacle avoidance distance ≥ 2m. The end path switches to the ultra-low altitude mode (height 1.5m) when it is 5m away from the target point, and turns on the lidar obstacle avoidance (detecting the obstacle distance accuracy ±5cm); Use a 30x optical zoom lens (resolution 4096×2160) to take multi-angle pictures of the bolt details (front view, side view, top view), simultaneously turn on the ring fill light (illuminance 5000lux) to ensure night clarity, and transmit the images back to the edge node in real time. Detect detailed defects such as bolt corrosion and cracks through the lightweight YOLOv5s model, and generate an image diagnosis report (such as "the rust area on the bolt surface accounts for 15%"); If the image diagnosis report confirms that the risk is valid, trigger the robot to go out; If it is a false alarm (such as blocked by a bird), terminate the task and record the reason for the false alarm; The robot plans the path based on the three-dimensional space reference model of the iron tower, uses multi-sensor fusion positioning (laser SLAM + UWB base station), and avoids ground obstacles (such as gravel and water accumulation) in real time. After reaching the target point, the robotic arm aligns with the bolt through visual servo (using AR markers for assistance). The robotic arm is equipped with a digital torque wrench (range 0 - 200 Nm), and calculates the required torque according to value (such as = 1.3° corresponding to torque T = 120 Nm), performs clockwise rotation for tightening, and monitors the torque value in real time until it meets the standard (such as stopping when T = 125 Nm). After the operation is completed, the passive sensor is triggered to immediately feedback the updated value (such as = 0.2°). If it does not meet the standard, secondary tightening is initiated. The robot controls the micro-endoscope to extend into the bolt connection gap, takes pictures of the internal state (such as whether the gasket is deformed), transmits the images back to the edge node for secondary confirmation, and records the torque value, correction value and endoscope images to form operation data; According to the recorded operation data (such as = 0.2°), update the mechanical parameters (such as pre-tightening force and stress distribution) of the corresponding bolt in the three-dimensional space reference model of the iron tower. Based on the lightweight model of finite element analysis, recalculate the stress influence range of adjacent components (such as after bolt A is repaired, the stress of adjacent bolt B drops by 20%). Input the corrected parameters into the AI model to regenerate the dynamic risk heat map. The original high-risk area is downgraded to normal (green), and the risk level of the associated area is updated synchronously (such as adjacent bolts change from yellow low risk to green normal).

[0045] It should be noted that the coordinates of the self-organizing network broadcast event are obtained, the UAV approaches for shooting according to the B-spline curve (30 times zoom), and the robot performs torque control based on visual servo; update the mechanical parameters of the three-dimensional space reference model of the iron tower according to the value after maintenance, and recalculate the stress influence range by finite element analysis; after the robot tightens the bolt, it takes pictures of the gasket state through the endoscope, and combines the real-time feedback of the passive sensor value. The repair verification time is compressed from 2 hours to 5 minutes; after the bolt is repaired, the risk level of the area where the stress of adjacent components drops by 20% is automatically downgraded, reducing 30% redundant inspections.

[0046] This embodiment also provides a computer device, which is applicable to the situation of the edge multi-device linkage method of transmission towers based on self-organizing network and two- and three-dimensional fusion, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the edge multi-device linkage method of transmission towers based on self-organizing network and two- and three-dimensional fusion as proposed in the above embodiment.

[0047] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0048] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for realizing multi-device linkage at the edge of a transmission tower based on ad hoc network and two-dimensional and three-dimensional fusion as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read Only Memory (EPROM for short), Programmable Red-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0049] In summary, the present invention: by constructing a dynamic monitoring system for transmission towers that couples mechanics, thermodynamics, and geometry, first, the passive sensor based on ambient backscattering analyzes the displacement of the cantilever beam through phase modulation, and combines the coordinate mapping of the three-dimensional space reference model to reduce the detection error of bolt loosening angle and accurately lock the microscopic defects in space positioning. Second, relying on edge computing and self-organizing network communication, the AI model fuses multi-dimensional data to generate a dynamic risk thermal map, and triggers the drone to approach for shooting and the robot for precise maintenance, forming a perception-decision-execution closed loop to improve the response efficiency. In addition, the passive sensor eliminates the need for battery maintenance through radio frequency power supply; the self-organizing network dynamic routing protocol maintains a high connectivity rate in case of single-node failure, which is significantly better than the fixed topology scheme. It provides a monitoring paradigm for the smart grid with high-precision positioning, real-time response, and extremely simple operation and maintenance, promoting the evolution of transmission facility management towards intelligence and unmanned operation.

[0050] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A multi-device linkage method at the edge of a transmission tower based on self-organizing network and two-dimensional and three-dimensional fusion, characterized in that: include, Deploy multiple devices at the edge of the transmission tower, embed passive sensors based on environmental backscatter, and use drones equipped with lidar to generate a three-dimensional spatial benchmark model of the tower; Based on the three-dimensional spatial benchmark model of the tower, the autonomous inspection route of the UAV is planned, point cloud data, infrared thermal imaging and backscattering data are collected simultaneously, deformation parameters of passive sensors are extracted by analyzing channel state information, and a multidimensional data set is constructed; The multidimensional data sets are integrated at the edge node, and a dynamic risk heat map is generated and the risk levels are marked; When a high risk level is detected, the equipment deployment strategy is triggered and the operation data is recorded, and the three-dimensional spatial benchmark model of the tower is updated.

2. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 1, characterized in that: The multi-device deployment includes deploying edge computing nodes at the bottom of the tower, integrating ZigBee / Wi-Fi dual-mode ad hoc network communication tools, and physically connecting with the embedded airport management system.

3. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 1, characterized in that: The three-dimensional spatial reference model of the tower is obtained by flying a drone equipped with a Velodyne VLP-16 laser radar along a calibration path to collect point cloud data of the tower, aligning the point cloud data with the BIM model coordinate system through the ICP algorithm to generate a fusion coordinate system, and mapping the installation position coordinates of the passive sensor to the BIM model coordinate system to obtain the three-dimensional spatial reference model of the tower.

4. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 1, characterized in that: The planned UAV autonomous inspection route is based on the three-dimensional space benchmark model of the tower. By setting the horizontal safety distance and the vertical layered scanning path, the UAV surrounding route is dynamically generated and a real-time obstacle avoidance strategy is integrated.

5. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 1, characterized in that: The channel state information includes signal strength, phase offset, propagation time difference and subcarrier index.

6. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 1, characterized in that: The passive sensor deformation parameters include cantilever beam displacement and bolt loosening angle.

7. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 6, characterized in that: The generation of a dynamic risk heat map and the marking of risk levels utilize an AI model to associate the three-dimensional spatial benchmark model of the tower with the fused multidimensional data set, calculate the risk value, divide the risk level, and generate a dynamic risk heat map.

8. The method for multi-device linkage at the edge of a transmission tower based on ad hoc networking and two-dimensional and three-dimensional fusion as claimed in claim 7, characterized in that: The equipment deployment strategy includes deploying drones to approach the target location to collect high-definition images, dispatching inspection robots to perform on-site maintenance, and correcting the three-dimensional spatial reference model of the tower.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for linking multiple devices at the edge of a transmission tower based on self-organizing network and two-dimensional and three-dimensional fusion as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for linking multiple devices at the edge of a transmission tower based on self-organizing network and two-dimensional and three-dimensional fusion as described in any one of claims 1 to 8 are implemented.

Images