Operation early warning monitoring system of power transmission line based on environment monitoring

Abstract

The invention relates to the technical field of intelligent operation, maintenance and early warning of a power grid, and discloses a power transmission line operation early warning monitoring system based on environmental monitoring, which comprises a sag measurement module for measuring sag data of a power transmission line lead in real time, and a power information acquisition module for acquiring power information of a power transmission line in real time, the electromagnetic environment calculation module establishes a sag-electromagnetic coupling model and calculates electromagnetic environment characteristics of the space around the power transmission line, the threshold management module dynamically generates a fault probability distribution matrix and an early warning threshold by fusing multi-dimensional environment factors, and the early warning decision module compares the fault probability with the early warning threshold in real time and automatically triggers an alarm. The linkage response module automatically plans an unmanned aerial vehicle inspection route, issues an inspection task and obtains return data, and the feedback updating module optimizes the model online based on a cross-modal defect recognition result. According to the method, the false alarm rate and the missing report rate of the system can be remarkably reduced, the operation and maintenance period is shortened to the minute level, and the safety and economy of the power transmission line are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and early warning technology for power grids, and in particular to an operation early warning monitoring system for transmission lines based on environmental monitoring. Background Technology

[0002] For a long time, the safe operation of transmission lines has mainly relied on manual inspections and offline simulation calculations. Maintenance personnel regularly climb the towers or use telescopes and infrared instruments for visual inspections, which is time-consuming, labor-intensive, and weather-dependent. Furthermore, it cannot reflect the sag fluctuations of conductors caused by real-time changes in temperature, icing, and wind load, nor can it dynamically assess the probability of faults. This results in delayed discovery of hidden dangers, high false alarm and missed alarm rates, and makes it difficult to meet the lean management needs of ultra-high voltage, long-span, and complex environmental lines.

[0003] In existing online monitoring solutions for transmission lines, most solutions lack a unified coupling model. Sag data and electromagnetic parameters are processed separately, and the monitoring terminal only alarms when the threshold exceeds the limit without providing a quantitative probability of failure. After the alarm is generated, manual re-inspection is still required. The closed-loop cycle is long and historical defects cannot be fed back to the model side, making it difficult to improve the prediction accuracy.

[0004] In recent years, although artificial intelligence and drone technology have been introduced into the field of power transmission inspection, on the one hand, drone flight path planning still relies mainly on human experience and lacks a mechanism to link with real-time early warning information. On the other hand, the defect identification results are only used to dispatch work orders and not to correct the fault probability model in reverse, which leads to the model relying on the initial training set for a long time and having insufficient adaptability. In addition, existing environmental classification methods mostly use a single meteorological factor or simple threshold division, ignoring the comprehensive impact of multi-dimensional factors such as terrain and vegetation on faults, which further amplifies the prediction bias of the model.

[0005] Existing patents and literature have not solved the key problems mentioned above, such as mechanical-electromagnetic coupling modeling, multi-dimensional environment classification, reinforcement learning threshold optimization, and UAV linkage response. These are the technical bottlenecks that this invention aims to overcome. Summary of the Invention

[0006] This invention provides an early warning monitoring system for the operation of power transmission lines based on environmental monitoring, in order to solve existing technical problems.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an operation early warning monitoring system for power transmission lines based on environmental monitoring, comprising: Sag Measurement Module: Used for real-time measurement of sag data of transmission line conductors; Power information acquisition module: used to collect power information of transmission lines in real time; Electromagnetic environment calculation module: used to establish a sag-electromagnetic coupling calculation model based on sag data and power information, and calculate the electromagnetic environment characteristics of the space around the transmission line; Threshold management module: used to establish the fault probability distribution matrix and early warning thresholds for transmission lines under different environmental levels; Early warning decision module: used to generate fault probability based on electromagnetic environment characteristics and fault probability distribution matrix, and generate early warning information based on the comparison result of fault probability and early warning threshold; Linkage Response Module: Used to output early warning information to the operation and maintenance terminal and automatically trigger drone inspection tasks; Feedback Update Module: This module performs secondary analysis on the high-definition images and infrared temperature measurement data transmitted back by the UAV using a defect identification algorithm, and updates the fault probability distribution matrix and early warning threshold based on the analysis results.

[0008] The beneficial effects of the technical solution provided by this invention include at least the following: This invention constructs a closed-loop system of "real-time monitoring - dynamic modeling - intelligent early warning - linkage response - feedback update", which overturns the inefficient traditional "static model + manual inspection" mode. With the sag measurement module and power information acquisition module as front-end perception, the electromagnetic environment calculation module as the core engine, the threshold management module as the decision basis, the linkage response module as the execution means, and the feedback update module as the evolution driver, it realizes full-link automation from data to decision, from decision to action, and from action to model regeneration, which significantly reduces the manpower cost of operation and maintenance and greatly improves the timeliness of hidden danger detection.

[0009] This invention proposes a "sag-electromagnetic coupling calculation model," which for the first time maps the mechanical deformation of the conductor into electromagnetic field boundary conditions in real time. This solves the problem that traditional offline simulation cannot reflect the electromagnetic field calculation error caused by the dynamic change of sag. Combined with parallel FEM solving and CIGRE semi-empirical formula, it outputs high-precision electromagnetic environment characteristics within the transmission line range in seconds, providing reliable data support for fault probability prediction and making the early warning results more consistent with the actual operating state of the line.

[0010] This invention introduces a multi-dimensional environmental feature fusion and reinforcement learning online optimization mechanism to overcome the one-sidedness of a single meteorological factor. It fully explores the nonlinear relationship between electromagnetic environment features and faults using a Transformer-based time series model to achieve accurate quantification of fault probability. The PPO reinforcement learning algorithm is used to dynamically adjust the warning threshold, significantly reducing the false alarm rate and missed alarm rate of warnings.

[0011] This invention establishes a closed-loop linkage mechanism that automatically drives UAV inspections based on early warning information and feeds back defect data. Once the probability of a fault exceeds a dynamic threshold, the system immediately calls upon the UAV to collect image, infrared, and laser point cloud data of the suspected faulty transmission line. Subsequently, the ImageBind cross-modal network is used to perform pixel-level identification of the defect and send the results back to the threshold management module to achieve continuous model evolution. This mechanism compresses the traditional "alarm-manual application-dispatch-feedback-manual analysis-experience correction" cycle of several days to a "minute-level" closed loop, greatly improving operation and maintenance efficiency and line reliability. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a system structure diagram provided in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0015] This embodiment provides an operation early warning monitoring system for transmission lines based on environmental monitoring. Please refer to... Figure 1 This is a system structure diagram provided in an embodiment of the present invention.

[0016] An operational early warning monitoring system for transmission lines based on environmental monitoring, as described in this embodiment, includes: 1. Sag Measurement Module: Used for real-time measurement of sag data of transmission line conductors; The sag measurement module includes a data acquisition submodule, a data preprocessing submodule, a sag calculation submodule, and a data transmission submodule. The data acquisition submodule is used to obtain the conductor temperature through a temperature sensor, the local tilt angle of the insulator string and conductor through a tilt sensor, the axial tension of the conductor through a tension sensor, and the vertical distance of the conductor to the ground through a laser rangefinder, thereby obtaining raw data. The data preprocessing submodule is used to clean, standardize, and normalize the raw data to form preprocessed raw data; The sag calculation submodule is used to calculate the sag data of the conductor in real time based on the preprocessed raw data using the mechanics-caten line equation. The data transmission submodule is used to encapsulate the sag data of the conductors via the MQTT protocol and cache it in a local embedded real-time database. The encapsulated sag data is then uploaded to the electromagnetic environment calculation module in real time via the OPGW fiber optic private network.

[0017] It should be noted that the temperature sensor is used to measure the surface temperature of the conductor and calculate the thermal expansion of the conductor. The recommended model is the Pt100 platinum resistance sensor (IEC 60751, Class A, ±0.15℃). It should be used with aluminum-clad strapping or pre-twisted wire suspension clamp to ensure a tight fit with the outer surface of the conductor.

[0018] Tilt sensors are used to measure the tilt angle of insulator strings and the local tilt angle of conductors, and to calculate the change in conductor tension direction. The recommended model is a MEMS dual-axis tilt module (±30° range, 0.001° resolution, -40℃~+85℃), which is fixed to the insulator string hanging point hardware and the conductor connecting plate by bolts.

[0019] Tension sensors are used to directly measure the axial tension of conductors. The recommended model is a column strain gauge tension sensor (rated range 100kN, overall accuracy 0.05%FS, breaking load ≥300%FS), which is connected in series between the tension string or suspension string and the tower. It adopts a double shear beam structure and is equipped with an anti-torsion arm.

[0020] Laser rangefinders are used to measure the vertical distance of a conductor to the ground and to calibrate the absolute height of sag. The recommended model is a pulsed ToF laser rangefinder module (range 0.2-200m, ±2mm@100m, 905nm Class 1 eye-safe). It is fixed to the lower plane of the crossarm of the tower and vertically aligned with the lowest sag point of the conductor. It is equipped with a 2D pan-tilt head with ±30° scanning function.

[0021] Data cleaning: Hampel filtering method (window length 15s) was used to identify instantaneous jumps in conductor temperature, tilt angle, tension, and vertical distance to ground, and then the data was cleared. Standardization: Temperature is standardized to the ITS-90 temperature scale, with a unit of 0.01℃; tension is standardized to kN and retained to 3 decimal places; tilt angle is standardized to rad and retained to 6 decimal places; and vertical distance to the ground is standardized to m and retained to 3 decimal places. Normalization: The Min-Max method is used to map each physical quantity to the interval [0,1].

[0022] OPGW fiber optic private network: The network topology is an industrial-grade SDH / PTN ring network, the transmission medium is OPGW-24B1-110 composite overhead ground wire, and the communication equipment includes tower-mounted equipment: industrial-grade protocol converter (RS-485→100Base-FX optical port), and station-end equipment: aggregation switch (supporting ERPS ring network protocol, VLAN isolation service and protection channel), and adopts IEEE 1588v2 synchronization time.

[0023] II. Power Information Acquisition Module: Used to collect power information of transmission lines in real time; The power information acquisition module includes a signal coupling submodule, a signal preprocessing submodule, a power information calculation submodule, a clock synchronization submodule, and a data transmission submodule; The signal coupling submodule is used to convert voltage and current signals on the transmission line into secondary measurable levels through optical voltage transformers and Rogowski coils; The signal preprocessing submodule is used to perform electrical isolation and amplitude adjustment on the secondary measurable level through an isolation amplifier, and to perform synchronous multi-channel A / D conversion on the adjusted secondary measurable level through a synchronous sampling method to obtain the preprocessed electrical signal. The power information calculation submodule is used to extract fundamental / harmonic frequencies, calculate effective values, track frequencies, and calculate power quality parameters from preprocessed electrical signals within the local FPGA / SoC using a joint FFT+sliding DFT algorithm to obtain power information. The clock synchronization submodule is used to provide a time reference for all acquisition nodes via GNSS / BeiDou dual-mode timing. The data transmission submodule is used to encapsulate power information via the MQTT protocol and cache it in a local embedded real-time database. The encapsulated power information is then uploaded to the electromagnetic environment calculation module in real time via a multi-path redundant network.

[0024] It should be noted that an optical voltage transformer (OVT) is a voltage sensing unit based on an electro-optic crystal. It is based on the Pockels electro-optic effect and transmits the voltage signal on the high potential side to the low potential side in the form of light intensity changes through an optical fiber.

[0025] A Rogowski coil is a hollow toroidal current transformer that uses the principle of electromagnetic induction to output a voltage signal that is proportional to the derivative of the high-voltage side current. The voltage signal is then integrated by an integrator to obtain the current signal. It features no magnetic core, no saturation, wide bandwidth, and easy installation under energization.

[0026] The isolation amplifier is located at the front end of the low-voltage side and is used to perform electrical isolation, level shifting and anti-aliasing filtering of analog signals, ensuring safe isolation between the high-voltage side and the low-voltage side.

[0027] Synchronous multi-channel A / D conversion uses a multi-channel, synchronously sampled analog-to-digital converter to instantaneously sample analog data channels such as voltage and current on the same clock edge, ensuring strict time alignment of data from each channel and providing a foundation for subsequent vector calculations.

[0028] The FFT+Sliding DFT joint algorithm first uses Fast Fourier Transform to obtain the fundamental and harmonic spectra in the digital domain, and then uses Sliding Discrete Fourier Transform to continuously track frequency drift and perform smooth output, thereby achieving high-precision real-time calculation of RMS values, power, harmonics and frequencies.

[0029] Local FPGAs / SoCs are heterogeneous chips that integrate programmable logic and embedded processors. The logic side is responsible for high-speed sampling, algorithm acceleration, and data packaging, while the processor side is responsible for managing the communication protocol stack and local cache, together forming a complete edge computing node.

[0030] GNSS / BeiDou dual-mode timing is used to receive second pulses and timing information from the global navigation satellite system, providing a unified and high-precision absolute time reference for all acquisition nodes, thereby ensuring cross-node data synchronization and traceability of event sequence.

[0031] III. Electromagnetic Environment Calculation Module: This module is used to establish a sag-electromagnetic coupling calculation model based on sag data and power information, and to calculate the electromagnetic environment characteristics of the space surrounding the transmission line. The electromagnetic environment calculation module includes a parameter import and verification submodule, a 3D scene construction submodule, an arc-electromagnetic coupling modeling submodule, a spatial mesh discretization submodule, an electromagnetic field solving submodule, and a feature fusion submodule. The parameter import and verification submodule is used to verify the integrity, time scale consistency and unit legality of sag data and power information through a JSON / Protobuf parser and generate model parameters. When the verification result is abnormal, data retransmission or interpolation compensation is triggered. The model parameters include tower coordinates, terrain DEM / OPGW parameters, conductor sag, split spacing and phase sequence arrangement. The 3D scene construction submodule is used to build a 3D geometric model of the transmission line in real time based on the model parameters by combining the GIS engine with the catenary / parabola hybrid equation solver. The sag-electromagnetic coupling modeling submodule is used to establish mechanical-electromagnetic coupling equations using MATLAB, mapping the change in conductor sag to the change in the conductor's height above the ground, and thus correcting the spatial distribution functions of electric and magnetic fields in the three-dimensional geometric model. The spatial grid discretization submodule is used to establish the coordinates of discrete grid nodes in the three-dimensional geometric model with a width of 0~200m and a height of 0~50m on both sides of the transmission line using the tetrahedral partitioning algorithm. The electromagnetic field solution submodule is used to solve the quasi-static field of Maxwell's equations on discrete grid node coordinates using a HYPRE-based parallel FEM solver, outputting the power frequency electric field strength and power frequency magnetic induction intensity, and obtaining radio interference and audible noise by calling the CIGREWG 36-01 semi-empirical formula; The feature fusion submodule is used to normalize the dimensions of power frequency electric field intensity, power frequency magnetic induction intensity, radio interference and audible noise using the Min-Max normalization method to obtain electromagnetic environment characteristics.

[0032] It should be noted that the JSON / Protobuf parser is a lightweight, cross-language structured data encoding and decoding tool that combines the high readability of JSON with the compactness and efficiency of Protobuf. In this invention, its function is to parse binary / text stream data containing sag data and power information into objects with verified key-value pairs, and to ensure that the fields are complete, the time scale is aligned, and the units are correct.

[0033] GIS engine: A software core for storing, indexing, and visualizing geospatial data, with built-in coordinate transformation, projection, and terrain overlay functions. In this invention, its role is to unify the tower coordinates, DEM elevation, and line corridor vectors of transmission lines into the same three-dimensional coordinate system, providing a "skeleton" for geometric modeling of three-dimensional scenes.

[0034] Catenary / parabolic hybrid equation solver: The idea behind the catenary / parabolic hybrid solution is that the solver first determines the threshold of the span and tension of the transmission line conductor and automatically switches the equations. The catenary equation is used to solve the large span or high tension part of the conductor, while the parabolic equation is used to correct the small span part. In this invention, it is used to quickly generate the three-dimensional spatial curve of the transmission line conductor based on the real-time sag value, as the geometric input of the electromagnetic field model.

[0035] Mechanical-electromagnetic coupling equation: a mathematical model that maps mechanical deformation to electromagnetic field boundary conditions. The mapping logic is: change in conductor sag → change in height above ground → change in electric / magnetic field distribution function. In this invention, it is used to realize a closed loop of "micro-deformation of the line → real-time refresh of the electromagnetic environment".

[0036] Tetrahedral meshing algorithm: An algorithm that discretizes a continuous three-dimensional space into a tetrahedral mesh, commonly known as Delaunay or AdvancingFront. In this invention, it is used to generate discrete mesh node coordinates in a space of 0-200m (horizontal) × 0-50m (vertical), providing a mesh skeleton for subsequent FEM solution.

[0037] A HYPRE-based parallel FEM solver: an open-source, high-performance preconditioning library used in this invention to solve the 50Hz quasi-static Maxwell equations on a tetrahedral mesh to obtain the power frequency electric field intensity E and the power frequency magnetic induction intensity B.

[0038] CIGREWG36-01 Semi-Empirical Formula: An empirical formula for radio interference (RI) and audible noise (AN) based on a large number of measured statistics of high-voltage lines. In this invention, it is used to substitute the maximum electric field strength on the surface of the conductor into the formula after solving the electromagnetic field and directly output RI and AN, which are combined with E and B to form the electromagnetic environment characteristics.

[0039] IV. Threshold Management Module: Used to establish the fault probability distribution matrix and early warning thresholds for transmission lines under different environmental levels; The threshold management module includes an environmental level classification submodule, a fault sample library management submodule, a fault probability modeling submodule, and a threshold calculation submodule; The environmental level classification submodule is used to predefine the multidimensional environmental characteristics of the environment where the transmission line is located. It generates the environmental level through a dual-mode classification algorithm of rule engine + K-means clustering algorithm. The multidimensional environmental characteristics include meteorological characteristics, terrain characteristics, building characteristics, vegetation characteristics and human activity characteristics. The fault sample library management submodule is used to generate a structured fault sample library based on the historical fault records of transmission lines and the corresponding historical electromagnetic environment characteristics, through the joint coding method of fault scenario-electromagnetic feature. The fault probability modeling submodule is used to construct a fault probability mapping model based on the environmental level and a structured fault sample library using a Transformer-based time series algorithm, and outputs a fault probability distribution matrix. The threshold calculation submodule is used to solve for the warning threshold through the reinforcement learning PPO online optimization algorithm, using the following formula:

[0040] In the formula, i represents the environmental level. λ is the warning threshold for environmental level i, λ is the penalty coefficient for missed reports, α is the preset false alarm rate, and β is the preset missed report rate.

[0041] It should be noted that the rule engine uses the open-source lightweight rule engine Drools 7.x, which describes expert rules in the form of DRL files. The following is an example of a predefined multi-dimensional environmental characteristic of the transmission line's environment:

[0042]

[0043]

[0044]

[0045]

[0046] The rule engine + K-means clustering dual-mode fusion strategy is as follows: First, the rule engine is executed to define the environment level for each dimension of environmental features. For environmental features that are not fully covered by a single rule, K-means is used for unsupervised clustering to obtain supplementary environment levels. Finally, a unified environment level i∈{1,2,3,4,5} is output and written to a MySQL table.

[0047] Historical fault records: usually from the PMS system, including the time of the fault, tower number, fault type and cause of the fault; Historical electromagnetic environment characteristics: corresponding to the fault time in the historical fault record, the four-dimensional time series data of E, B, RI and AN fed back by the electromagnetic environment calculation module; The method flow of joint encoding of fault scenarios and electromagnetic features: Scene slicing: Centered on "fault time T0", take a time series of 40 minutes in the interval [T0-30min, T0+10min] and extract the historical electromagnetic environment features within this time series.

[0048] The mapping model for failure probability and the process for generating the failure probability matrix: Model inputs: real-time electromagnetic environment feature vector x_real=[E,B,RI,AN] and environment level i.

[0049] Probability mapping: Perform K-NN (k=5) retrieval on x_real and all samples belonging to the same environmental level i in the structured fault sample library, using radial basis kernel weight w_j=exp(-d_j / σ), where j∈{1,5} and j is an integer, σ=0.05; Then, the fault probability p_fault=Σ(w_j·y_j) / Σ(wj) is calculated, where w_j is the radial basis kernel weight and y_j∈{0,1}. The final output is the fault probability distribution matrix P_fault(i,x,y) for each environmental level, where x and y are the horizontal and vertical coordinates of the line corridor plane, respectively. This matrix is ​​written to memory in real time in the form of an HDF5 file.

[0050] The reinforcement learning PPO online optimization algorithm runs as an independent Python service and communicates with the threshold management module via ZeroMQ. It performs a threshold update every 10 minutes to complete the issuance of the new value of Ti and the response feedback.

[0051] V. Early Warning Decision Module: This module generates fault probabilities based on electromagnetic environment characteristics and fault probability distribution matrix, and generates early warning information based on the comparison between fault probabilities and early warning thresholds. The early warning decision module includes a feature receiving and caching submodule, a probability mapping submodule, an early warning information generation submodule, and a decision log recording submodule; The feature receiving and caching submodule is used to receive electromagnetic environment features, fault probability distribution matrix and early warning threshold in real time and perform local caching and data verification. It extracts the coordinates of each monitoring point and its real-time feature vector x from the electromagnetic environment features through the feature slicing extraction method based on spatial index. The probability mapping submodule is used to call the lightweight inference engine ONNX Runtime to interpolate the real-time feature vector x with the fault probability distribution matrix point by point, so as to obtain the environmental level i and fault probability of each monitoring point. , ; The early warning information generation submodule is used to select the corresponding early warning threshold based on the environmental level i of each monitoring point. ,when ≥ An alarm is triggered when the alarm is triggered, and the monitoring point information that triggered the alarm is encapsulated into structured early warning information, which includes early warning ID, coordinates, timestamp, environmental level and fault probability; The decision log recording submodule is used to record the early warning decision-making process in real time, and calculate the false alarm rate and false alarm rate based on the total number of alarms, false alarms, and false alarms during the early warning decision-making process, and then send the data back to the threshold management module.

[0052] It should be noted that the feature slicing extraction method based on spatial index aims to locate the real-time feature vector x of each monitoring point in the three-dimensional electromagnetic environment features of 0~200 m (horizontal) × 0~50 m (vertical) × continuous time flow in milliseconds, thereby avoiding full data copying and reducing CPU usage and latency.

[0053] The index structure is a two-level index consisting of R*-Tree space and time sharding, including: First-level index: R*-Tree node key=( , , 3D bounding box; Secondary index: Each R*-Tree node is further divided into time-series blocks (timestamps) by 1 minute. The time-series blocks are stored in columnar format, with the field order as follows: E, B, RI, AN, timestamp.

[0054] VI. Linkage Response Module: Used to output early warning information to the operation and maintenance terminal and automatically trigger drone inspection tasks; The linkage response module includes a warning information parsing and verification submodule, an operation and maintenance terminal push submodule, an inspection task generation submodule, and an inspection task distribution submodule; The early warning information parsing and verification submodule is used to read and unpack the early warning information from the early warning decision module, as well as read the three-dimensional geometric model of the transmission line from the electromagnetic environment calculation module. The operation and maintenance terminal push submodule is used to push warning information to operation and maintenance management personnel through SMS push, APP push and internal system push methods; The inspection task generation submodule is used to automatically retrieve UAV airports based on early warning information and the three-dimensional geometric model of the power transmission line, and plan the inspection route through the NSGA-III-Topo route planning algorithm, and output the inspection task plan, which includes airport ID, UAV ID, route ID, inspection time and inspection path. The inspection task distribution submodule is used to distribute the inspection task plan to the drone airport via 5G network or local Wi-Fi Mesh, based on the inspection task plan. The drone airport then dispatches drones to perform the inspection task and transmits the inspection data back in real time. The inspection data includes image data, infrared temperature measurement data, and laser point cloud data.

[0055] It should be noted that the NSGA-III-Topo route planning algorithm is an improved multi-objective optimization algorithm based on the classic NSGA-III framework, which introduces a topology repair operator (Topo-operator). It uses the reference point mechanism to maintain diversity in the high-dimensional target space, and automatically eliminates self-intersections and no-fly zone conflicts through the Topo-operator after intersection and mutation, ensuring the continuity and feasibility of the path. The algorithm seeks optimization in parallel with three objectives: "shortest flight distance, least time, and maximum risk coverage", and gives the optimal set of routes for UAV inspection of power transmission lines.

[0056] 5G network: It adopts 5G NR standalone networking, with 64T64R base stations and drones with built-in 5G modules. It uses URLLC slicing and dedicated DNN to ensure high-priority and high-reliability transmission of inspection task messages, and supports certificate encryption and dual-transmission deduplication redundancy.

[0057] Local Wi-Fi Mesh: Based on 802.11ax, tower-mounted industrial APs and nested Mesh nodes are mixed in a network. When the 5G signal is weak or there is packet loss, it automatically switches to the Wi-Fi Mesh link to ensure that task distribution and image transmission are uninterrupted.

[0058] VII. Feedback Update Module: This module is used to perform secondary analysis on the high-definition images and infrared temperature measurement data transmitted back by the UAV using a defect identification algorithm, and to update the fault probability distribution matrix and early warning threshold based on the analysis results.

[0059] The feedback update module includes a data processing submodule, a defect identification submodule, an operation and maintenance terminal push submodule, and a matrix update submodule; The data processing submodule is used to receive the inspection data transmitted back by the UAV and perform integrity verification, data cleaning, enhancement and normalization to obtain preprocessed inspection data. The defect identification submodule is used to calculate and output defect identification results based on preprocessed inspection data through a downstream task network with ImageBind as the backbone. The defect identification results include defect line coordinates, defect type, defect confidence, pixel-level defect location mask, and hot spot temperature at the defect. The operation and maintenance terminal push submodule is used to push defect identification results to operation and maintenance management personnel through SMS push, APP push and internal system push methods; The matrix update submodule is used to send the defect identification results back to the threshold management module, which then uses the electromagnetic environment features to optimize and update the fault probability distribution model.

[0060] It should be noted that ImageBind is a cross-modal unified encoder open-sourced by Meta. Based on the ViT-Huge backbone, it utilizes contrastive learning and mask modeling for joint training. Downstream, only a lightweight head needs to be added to complete detection, segmentation, or retrieval tasks. A single training iteration can map image, infrared, and point cloud data to the same high-dimensional semantic space. It also supports exporting PyTorch code and ONNX scripts. Downstream lightweight heads can include: DETR inspection head: Used to output the coordinates of the defective line, the defect type, and the defect confidence level; The Mask2Former segmentation head is used to output pixel-level defect localization masks; The 1×1 convolutional temperature regression head is used to output the hot spot temperature at the defect location.

[0061] Furthermore, it should be noted that the present invention can be provided as a method, apparatus, or computer program product. Therefore, embodiments of the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0062] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing terminal equipment to cause a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0064] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0065] Finally, it should be noted that the above description represents a preferred embodiment of the present invention. It should be pointed out that although preferred embodiments have been described, those skilled in the art, once they understand the basic inventive concept of the present invention, can make various improvements and modifications without departing from the principles described herein. These improvements and modifications should also be considered within the scope of protection of the present invention. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

Claims

1. A transmission line operation early warning monitoring system based on environmental monitoring, characterized in that, include: Sag Measurement Module: Used for real-time measurement of sag data of transmission line conductors; Power information acquisition module: used to collect power information of transmission lines in real time; Electromagnetic environment calculation module: used to establish a sag-electromagnetic coupling calculation model based on sag data and power information, and calculate the electromagnetic environment characteristics of the space around the transmission line; Threshold management module: used to establish the fault probability distribution matrix and early warning thresholds for transmission lines under different environmental levels; Early warning decision module: used to generate fault probability based on electromagnetic environment characteristics and fault probability distribution matrix, and generate early warning information based on the comparison result of fault probability and early warning threshold; Linkage Response Module: Used to output early warning information to the operation and maintenance terminal and automatically trigger drone inspection tasks; Feedback Update Module: This module performs secondary analysis on the high-definition images and infrared temperature measurement data transmitted back by the UAV using a defect identification algorithm, and updates the fault probability distribution matrix and early warning threshold based on the analysis results.

2. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The sag measurement module includes a data acquisition submodule, a data preprocessing submodule, a sag calculation submodule, and a data transmission submodule. The data acquisition submodule is used to acquire conductor temperature, local tilt angle of insulator string and conductor, conductor axial tension and vertical distance to ground to form raw data; The data preprocessing submodule cleans, standardizes, and normalizes the raw data. The sag calculation submodule calculates the sag in real time based on preprocessed data using the mechanics-caten equation. The data transmission submodule encapsulates the sag data via the MQTT protocol and caches it in a local embedded database, then uploads it in real time to the electromagnetic environment calculation module via the OPGW fiber optic private network.

3. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The power information acquisition module includes a signal coupling submodule, a signal preprocessing submodule, a power information calculation submodule, a clock synchronization submodule, and a data transmission submodule; The signal coupling submodule converts the voltage and current signals of the transmission line into secondary measurable levels through an optical voltage transformer and a Rogowski coil; The signal preprocessing submodule uses an isolation amplifier to achieve electrical isolation and amplitude adjustment, and adopts a synchronous sampling method to perform multi-channel A / D conversion to obtain a preprocessed electrical signal; The power information calculation submodule performs fundamental and harmonic extraction based on a joint FFT and sliding DFT algorithm within the local FPGA / SoC, completes effective value calculation, frequency tracking and power quality parameter analysis, and generates power information. The clock synchronization submodule provides a unified time reference for each acquisition node through GNSS / BeiDou dual-mode time synchronization; The data transmission submodule uses the MQTT protocol to encapsulate power information and caches it in a local embedded database, then uploads it to the electromagnetic environment calculation module in real time via a multi-path redundant network.

4. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The electromagnetic environment calculation module includes a parameter import and verification submodule, a 3D scene construction submodule, an arc-electromagnetic coupling modeling submodule, a spatial grid discretization submodule, an electromagnetic field solving submodule, and a feature fusion submodule. The parameter import and verification submodule uses a JSON / Protobuf parser to verify the integrity, time scale consistency and unit legality of sag data and power information, and generates model parameters. The model parameters include tower coordinates, terrain DEM and OPGW parameters, conductor sag, split spacing and phase sequence arrangement. When the verification is abnormal, data retransmission or interpolation compensation is triggered. The 3D scene construction submodule establishes a 3D geometric model of the transmission line in real time based on the GIS engine and the catenary and parabola hybrid equation solver. The sag-electromagnetic coupling modeling submodule uses MATLAB to establish mechanical and electromagnetic coupling equations, maps sag changes to changes in the conductor's height above the ground, and corrects the electromagnetic field distribution in the three-dimensional model. The spatial grid discretization submodule uses a tetrahedral partitioning algorithm to generate discrete grid nodes in the three-dimensional model with a width of 0 to 200 meters and a height of 0 to 50 meters on both sides of the transmission line. The electromagnetic field solving submodule is based on the HYPRE parallel finite element solver. It solves the quasi-static field of Maxwell's equations on the grid nodes, outputs the power frequency electric field strength and power frequency magnetic induction intensity, and calls the CIGRE WG 36-01 semi-empirical formula to obtain radio interference and audible noise. The feature fusion submodule performs dimensional normalization processing on power frequency electric field strength, magnetic induction intensity, radio interference and audible noise through a normalization mapping function to obtain electromagnetic environment characteristics.

5. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The threshold management module includes an environmental level classification submodule, a fault sample library management submodule, a fault probability modeling submodule, and a threshold calculation submodule. The environmental classification submodule predefines the meteorological characteristics, topographic features, architectural features, vegetation features, and human activity features of the transmission line, and generates environmental levels through a rule engine and a K-means clustering dual-mode algorithm. The fault sample library management submodule constructs a structured fault sample library based on historical fault records and corresponding electromagnetic environment characteristics, using a joint encoding method of fault scenarios and electromagnetic features. The fault probability modeling submodule is based on environmental level and structured fault sample library, and uses Transformer time series algorithm to establish fault probability mapping model and output fault probability distribution matrix. The threshold calculation submodule is used to solve the warning threshold using the reinforcement learning PPO online optimization algorithm, and the following formula is used: In the formula, i represents the environmental level. λ is the warning threshold for environmental level i, λ is the penalty coefficient for missed reports, α is the preset false alarm rate, and β is the preset missed report rate.

6. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The early warning decision module includes a feature receiving and caching submodule, a probability mapping submodule, an early warning information generation submodule, and a decision log recording submodule. The feature receiving and caching submodule receives electromagnetic environment features, fault probability distribution matrix and warning threshold in real time, and performs local caching and data verification. It extracts the coordinates of each monitoring point and the real-time feature vector x through the feature slicing method of spatial indexing. The probability mapping submodule is used to call the lightweight inference engine ONNX Runtime to perform point-by-point interpolation between the real-time feature vector x and the fault probability distribution matrix to obtain the environmental level i and fault probability of each monitoring point. , ; The early warning information generation submodule is used to select the corresponding early warning threshold based on the environmental level i of each monitoring point. ,when ≥ An alarm is triggered when the alarm is triggered, and the monitoring point information that triggered the alarm is encapsulated into structured early warning information, which includes early warning ID, coordinates, timestamp, environmental level and fault probability; The decision log recording submodule records the early warning decision-making process in real time, calculates the false alarm rate and false alarm rate based on the total number of alarms, the number of false alarms, and the number of false alarms, and sends the data back to the threshold management module.

7. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The linkage response module includes a warning information parsing and verification submodule, an operation and maintenance terminal push submodule, an inspection task generation submodule, and an inspection task distribution submodule. The early warning information parsing and verification submodule reads the early warning information from the early warning decision module, unpacks it, and obtains the three-dimensional geometric model of the transmission line provided by the electromagnetic environment calculation module. The operation and maintenance terminal push submodule pushes early warning information to operation and maintenance management personnel via SMS, APP, and internal system. The inspection task generation submodule automatically retrieves UAV airports based on early warning information and the three-dimensional geometric model of the transmission line, and generates an inspection task plan using the NSGA-III-Topo route planning algorithm. The inspection task plan includes airport ID, UAV ID, route ID, inspection time, and inspection path. The inspection task distribution submodule distributes the inspection task plan to the drone airport via 5G network or local Wi-Fi Mesh, and the drone performs the inspection and transmits images, infrared temperature measurement and laser point cloud data back in real time.

8. The transmission line operation early warning monitoring system based on environmental monitoring according to claim 1, characterized in that: The feedback update module includes a data processing submodule, a defect identification submodule, an operation and maintenance terminal push submodule, and a matrix update submodule. The data processing submodule receives the inspection data transmitted back by the UAV and performs integrity verification, cleaning, enhancement and normalization to obtain preprocessed inspection data. The defect identification submodule calculates the preprocessed inspection data based on the ImageBind downstream task network and outputs the defect identification results, including the defect line coordinates, defect type, defect confidence, pixel-level defect location mask, and hot spot temperature at the defect. The operation and maintenance terminal push submodule pushes the defect identification results to operation and maintenance management personnel via SMS, APP and internal system; The matrix update submodule sends the defect identification results back to the threshold management module for optimizing and updating the fault probability distribution model.

Images