Power transmission line forest fire spreading range prediction method and system based on multi-factor fusion

Abstract

The invention relates to the technical field of disaster prevention and reduction of power systems, in particular to a power transmission line forest fire spreading range prediction method and system based on multi-factor fusion, and the method comprises the steps: collecting multi-source data in real time, and building a fusion database through time-space alignment processing; extracting fire scene features by adopting an image processing technology, and calculating a live line distance and vegetation combustion parameters by combining laser radar data; constructing a prediction model, and dynamically correcting fire spreading parameters; and simulating a spreading path based on the three-dimensional model and topographic data, evaluating the influence degree of the fire behavior on the power transmission line, and generating an early warning strategy. The problems of smog shielding monitoring failure, insufficient night prediction precision and the like caused by dependence of a single data source are effectively solved, the forest fire early warning capability in a complex environment is remarkably improved, and high-precision and real-time technical support is provided for disaster prevention and reduction of a power system.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster prevention and mitigation of power systems, and in particular to a method and system for predicting the spread range of wildfires on power transmission lines based on multi-factor fusion. Background Art

[0002] Currently, the primary method for forest fire risk forecasting in power transmission line wildfire hazard monitoring is a multi-factor, comprehensive indicator method. This method relies on comparative analysis of historical fire data and contemporaneous meteorological data, incorporating fuzzy set theory to develop a mathematical model for fire hazard levels and a fire risk level forecasting system. However, the digital images of wildfire hazards generated by this method suffer from significant noise interference and high error rates, severely impacting accurate assessment of fire spread trends and resulting in unreliable forecast results.

[0003] Publication No. CN106682580A discloses a fire prediction method and system based on transmission line fire images. This technology analyzes grayscale images of fires on transmission lines, extracts and classifies the fire edges, and then combines real-time geographic information to predict the fire's spread.

[0004] Although this method improves prediction accuracy through image noise reduction processing, it still has significant limitations: it relies solely on visible light / infrared image analysis and lacks multi-source data fusion, resulting in insufficient fire monitoring capabilities in complex environments such as smoke obscuration, at night or in bad weather, making it difficult to meet the power system's high-precision and real-time requirements for wildfire warnings. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for predicting the spread of wildfires on transmission lines based on multi-factor fusion, which solves the problem that existing wildfire prediction methods based on transmission line wildfire images have insufficient fire monitoring capabilities in complex environments, such as smoke obscuration, at night or in severe weather, and are difficult to meet the power system's high-precision and real-time requirements for wildfire warnings.

[0006] To achieve the above objectives, the present invention provides a method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion, comprising the following steps:

[0007] Real-time collection of multi-source monitoring data, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data, and lidar point cloud data;

[0008] Perform spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database;

[0009] Using image processing technology, fire scene feature information is extracted from infrared thermal imaging data and visible light image data. The spatial distance between the fire point and the power transmission line is calculated by combining the lidar point cloud data, and the vegetation burning characteristic parameters are obtained.

[0010] Build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation;

[0011] Based on the prediction model calculation results, combined with the three-dimensional model of the transmission line and terrain data, the fire spread path and development trend are simulated to assess the impact of the fire on the transmission line;

[0012] Generate early warning information and disposal suggestions.

[0013] Among them, real-time collection of multi-source monitoring data, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data and lidar point cloud data, including:

[0014] Micrometeorological data includes wind speed, wind direction, temperature and humidity parameters, and lidar point cloud data includes tower location information, conductor sag data and vegetation distribution data in the transmission line corridor.

[0015] The collected multi-source monitoring data are subjected to spatiotemporal alignment to establish a multi-source fusion database. The specific steps include:

[0016] Perform time synchronization processing on infrared thermal imaging data, visible light image data, micrometeorological data, Beidou positioning data, and lidar point cloud data, so that the time stamp deviation of each data collection time does not exceed 10 seconds;

[0017] Differential positioning is used to process Beidou positioning data to obtain fire point location data;

[0018] The LiDAR point cloud data is gridded, where the fire front area uses the first resolution grid and other areas use the second resolution grid;

[0019] The multi-source monitoring data after synchronization and grid processing are stored in a unified format to form a multi-source fusion database.

[0020] The method uses image processing technology to extract fire scene feature information from infrared thermal imaging data and visible light image data, combines it with lidar point cloud data to calculate the spatial distance between the fire point and the power transmission line, and obtains vegetation combustion characteristic parameters. The specific steps include:

[0021] Perform noise reduction on infrared thermal imaging data and use adaptive median filtering algorithm to eliminate image noise;

[0022] Perform edge detection on the infrared thermal imaging data and visible light image data after noise reduction to extract the contour features of the fire scene;

[0023] The fire scene contour features are spatially registered with the LiDAR point cloud data to calculate the three-dimensional distance between the fire point and the nearest transmission line conductor;

[0024] Based on the vegetation point cloud features in the lidar point cloud data, the vegetation height distribution information and vegetation density information are extracted, and combined with the preset vegetation burning characteristic parameter library, the vegetation burning characteristic parameters of the current fire scene are obtained.

[0025] Among them, a fire spread prediction model is constructed, and the processed multi-source monitoring data is input into the prediction model for calculation. The specific steps include:

[0026] Construct a fire spread prediction model that includes a physical calculation module and a neural network correction module;

[0027] The environmental parameters in the multi-source fusion database are input into the physical calculation module, and the basic fire spread speed is calculated using the Rothermel surface fire spread equation;

[0028] The historical disaster case data and real-time monitoring data in the multi-source fusion database are input into the neural network correction module to dynamically correct the basic fire spread speed and obtain the corrected fire spread parameters.

[0029] Among them, a fire spread prediction model is constructed, and the processed multi-source monitoring data is input into the prediction model for calculation:

[0030] The neural network correction module updates the model parameters every 30 minutes based on the latest collected monitoring data

[0031] The prediction model calculation results are combined with the three-dimensional transmission line model and terrain data to simulate the fire spread path and development trend and assess the impact of the fire on the transmission line. The specific steps include:

[0032] Based on the corrected fire spread parameters, combined with the digital elevation model and the three-dimensional model of the transmission line, the cellular automation algorithm is used to simulate the fire spread path in three-dimensional space.

[0033] Calculate the estimated time for the fire front to reach each tower location based on the fire spread path;

[0034] Assess the extent of the fire's impact on transmission lines;

[0035] Generate impact assessment results including affected tower number, estimated arrival time and risk level.

[0036] A transmission line wildfire spread prediction system based on multi-factor fusion includes an acquisition module, an alignment module, a feature extraction module, a prediction module, a deduction module and a decision module, wherein the alignment module is connected to the acquisition module, the feature extraction module is connected to the alignment module, the prediction module is connected to the feature extraction module, the deduction module is connected to the prediction module, and the decision module is connected to the deduction module;

[0037] The acquisition module is used to collect multi-source monitoring data in real time, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data and lidar point cloud data;

[0038] The alignment module is used to perform spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database;

[0039] The feature extraction module is used to extract fire scene feature information from infrared thermal imaging data and visible light image data based on image processing technology, calculate the spatial distance between the fire point and the transmission line in combination with the lidar point cloud data, and obtain vegetation combustion characteristic parameters;

[0040] The prediction module is used to build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation;

[0041] The deduction module is used to simulate the fire spread path and development trend based on the calculation results of the prediction model, combined with the three-dimensional model of the transmission line and terrain data, and evaluate the impact of the fire on the transmission line;

[0042] The decision-making module is used to generate early warning information and disposal suggestions.

[0043] The present invention provides a method and system for predicting the spread of wildfires along power transmission lines based on multi-factor fusion. This method collects multi-source data, including infrared thermal imaging, visible light images, micrometeorology, Beidou positioning, and lidar point clouds, in real time, and constructs a fusion database through spatiotemporal alignment. Image processing technology is used to extract fire scene features and, in combination with lidar data, calculates fire line distance and vegetation combustion parameters. A prediction model is constructed that integrates the Rothermel physical model and a neural network to dynamically correct fire spread parameters. The spread path is simulated based on a three-dimensional model and terrain data to assess the impact of the fire on the power transmission line and generate an early warning strategy. This solution effectively addresses existing issues such as smoke obstruction monitoring failure and insufficient nighttime prediction accuracy caused by reliance on a single data source. By integrating multi-source data, the accuracy of fire point positioning is improved, prediction response time is shortened, and dynamic model correction is used to reduce the error in fire front arrival time prediction. This significantly enhances wildfire early warning capabilities in complex environments, providing high-precision, real-time technical support for disaster prevention and mitigation in power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art.

[0045] Figure 1 This is a flowchart of the steps of a method for predicting the spread of wildfires along transmission lines based on multi-factor fusion according to the first embodiment of the present invention.

[0046] Figure 2 This is a principle block diagram of a transmission line wildfire spread range prediction system based on multi-factor fusion according to the second embodiment of the present invention.

[0047] In the figure: 201 - acquisition module, 202 - alignment module, 203 - feature extraction module, 204 - prediction module, 205 - deduction module, 206 - decision module. DETAILED DESCRIPTION

[0048] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0049] The first embodiment of this application is:

[0050] See also Figure 1 ,in, Figure 1 This is a flowchart of the steps of a method for predicting the spread of wildfires along transmission lines based on multi-factor fusion according to the first embodiment of the present invention.

[0051] The present invention provides a method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion, comprising the following steps:

[0052] S101: Real-time collection of multi-source monitoring data, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data, and lidar point cloud data;

[0053] Specifically, it consists of a variety of advanced sensing devices deployed in the transmission line corridors and surrounding areas. High-speed data transmission is achieved between the devices through a 5G private network, and the Precision Time Protocol (PTP) is used to ensure time synchronization accuracy of ±1 millisecond. Specifically, infrared thermal imaging data acquisition uses a dual-spectrum infrared thermal imager, where medium-wave infrared (3-5μm) is used to detect high-temperature open flames, and long-wave infrared (8-14μm) is used to identify low-temperature dark fires. The two bands work together to cover a temperature range of -20°C to 1500°C, with a spatial resolution better than 1 milliradian, and can detect fire points with a diameter of more than 0.5 meters. The visible light monitoring system is equipped with a 20-megapixel high-definition smart camera and a deep learning-based fireworks recognition algorithm. It can maintain an identification accuracy rate of more than 95% even in a low-light environment of 0.01Lux. The micrometeorological monitoring network consists of multiple six-element automatic weather stations, which upload data every 10 seconds. These stations measure key meteorological parameters, including wind speed (accuracy ±0.3 m / s), wind direction (±3°), temperature (±0.5°C), and humidity (±2% RH). The monitoring area covers a 500-meter area on both sides of the line. The positioning system utilizes the BeiDou-3 high-precision positioning terminal, combined with a ground-based augmentation system (GABS), to minimize the location error of fire points to within ±0.5 meters. The LiDAR scanning system, deployed on an autonomous inspection drone platform, uses a 32-line LiDAR at a 100Hz scanning frequency to acquire high-precision 3D point cloud data. This system achieves a positioning error of less than 10 centimeters for towers, measures conductor sag with an accuracy of ±5 centimeters, and identifies vegetation types with over 90% accuracy. These multi-source, heterogeneous data are initially correlated using a unified spatiotemporal reference (CGCS2000 coordinate system), laying the foundation for subsequent fusion processing. This effectively addresses the issue of insufficient severe weather monitoring capabilities.

[0054] S102: Performing spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database;

[0055] Specifically, the data streams from different monitoring devices are first time-synchronized and calibrated. A hybrid synchronization mechanism based on the Network Time Protocol (NTP) and the Precision Time Protocol (PTP) is used to ensure that the timestamp deviation of heterogeneous data such as infrared thermal imaging, visible light imagery, micrometeorological data, Beidou positioning, and lidar point clouds is strictly controlled within 10 seconds, and key data is synchronized at the sub-second level. In terms of spatial alignment, a unified spatial reference framework is established. Using the centimeter-level positioning data provided by the Beidou Ground-Based Augmentation System as a benchmark, the lidar point cloud and visible light imagery are registered using a feature point matching algorithm, with the registration error controlled within 0.3 meters. In response to the characteristics of data at different resolutions, an adaptive grid processing strategy is adopted: a high-precision grid of 0.5m×0.5m is used in the fire front area to accurately depict the details of the fire spread; a standard grid of 5m×5m is used in other areas to balance processing efficiency and data volume. A multi-level data fusion architecture was also designed. The bottom layer is the raw data layer, storing the raw data from each sensor; the middle layer is the feature extraction layer, storing the spatiotemporally aligned feature data; and the top layer is the fusion application layer, generating structured data that can be directly used in predictive models. All data is stored in a distributed database using a unified spatiotemporal encoding format (timestamp + spatial coordinates + data type + data value), supporting millisecond-level query response. This efficient data fusion process effectively solves the problems of high data noise and error rates, and is adaptable to complex and changing real-world application environments.

[0056] S103: Extracting fire scene feature information from infrared thermal imaging data and visible light image data using image processing technology, and calculating the spatial distance between the fire point and the power transmission line in combination with the lidar point cloud data, and obtaining vegetation burning characteristic parameters;

[0057] Specifically, the collected infrared thermal imaging data is first preprocessed using a noise reduction algorithm based on an adaptive windowed median filter. This algorithm dynamically adjusts the filter window size (3×3 to 7×7 pixels) based on the local noise characteristics of the image, effectively eliminating thermal noise interference while preserving the fire edge features. For visible light images, a Canny edge detection algorithm is used, combined with morphological closing operations to eliminate small interferences and extract a continuous and complete fire outline. During the feature fusion stage, a cross-modal registration algorithm based on SIFT feature points accurately matches the fire outline extracted from the infrared and visible light images with the lidar point cloud. The spatial distance calculation module uses the Delaunay triangulation algorithm to construct a three-dimensional model of the transmission line. A nearest neighbor search algorithm is used to calculate the minimum spatial distance between the fire point and the conductors and towers, achieving centimeter-level accuracy. The vegetation analysis subsystem uses point cloud classification technology to identify different vegetation types, such as trees and shrubs. Combined with a preset vegetation parameter database (containing 128 parameters such as moisture content and calorific value for 32 common tree species), it estimates vegetation burning characteristics in real time. A dynamic combustion model was introduced to predict burn rates based on vegetation height distribution (extracted by LiDAR) and meteorological data (wind speed and humidity). The model parameters were updated every five minutes. This multimodal feature fusion method effectively overcomes the limitations of single-image analysis, such as smoke obscuration and insufficient nighttime monitoring capabilities, enabling accurate fire analysis in all weather and terrain.

[0058] S104: Build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation;

[0059] Specifically, a "dual-engine" computing architecture was constructed: the physical calculation engine is based on the Rothermel surface fire spread equation, whose core parameters include wind speed and direction provided by micrometeorological data, vegetation moisture content extracted by lidar, and slope data generated by a digital elevation model. The neural network correction engine utilizes a three-layer LSTM architecture. The input layer receives the past two hours of historical fire spread data, meteorological trends, and transmission line status parameters. The hidden layer contains 128 neurons, dynamically weighting the importance of each feature through an attention mechanism. The model performs online learning every 30 minutes, using an incremental training algorithm to update network weights. The computational process begins with the physical engine generating a basic spread velocity field. The neural network then dynamically corrects this field based on real-time fire scene image features (a 1024-dimensional feature vector extracted using ResNet50) and conductor temperature monitoring data (sampling rate 1Hz). The final output fire spread prediction includes spread velocity, direction, and intensity classification. This fusion modeling approach effectively addresses the limited adaptability of a single model, enabling accurate prediction of fire evolution in complex environments.

[0060] S105: Based on the prediction model calculation results, combined with the three-dimensional model of the transmission line and terrain data, simulate the fire spread path and development trend, and evaluate the impact of the fire on the transmission line;

[0061] Specifically, the fire parameters (including spread speed, direction, and intensity) output by the prediction model are first integrated with a high-precision 3D transmission line model (built from a LiDAR point cloud with a tower positioning error of <10 cm) and 1-meter resolution digital terrain data. A cellular automaton algorithm is then used to perform a 3D spatial prediction. This algorithm incorporates the influence of power facilities and divides the transmission corridor into 0.5m×0.5m cells. The state transition probability of each cell is dynamically calculated based on the following factors: a base spread probability (based on the output of a Rothermel-ANN hybrid model), a terrain correction factor (taking into account slope and aspect), a vegetation type weight (derived from LiDAR classification results), and conductor thermal radiation feedback (calculated using a Flame-EM coupled model). During the prediction process, the spatial distance between the fire front and each tower conductor is calculated in real time. When the distance falls below a set threshold (dynamically adjusted based on voltage level, such as 15m for a 500kV line), a refined assessment module is activated. This module uses finite element analysis to calculate the impact of flame thermal radiation on conductors. It considers parameters such as conductor material (such as steel-core aluminum stranded conductor), operating temperature (obtained through online monitoring), and mechanical load to predict the risk of conductor melting or increased sag. It also combines insulator contamination levels and flame ion concentration data to assess the probability of insulator flashover and output three levels of risk warning: Level 1 (Emergency, expected impact time <30 minutes), Level 2 (Severe, 30-60 minutes), and Level 3 (General, >60 minutes). This three-dimensional dynamic assessment method effectively addresses the issue of inaccurate disaster impact assessments and provides a scientific basis for power grid emergency decision-making.

[0062] S106: Generate early warning information and disposal suggestions.

[0063] Specifically, based on the three-dimensional fire simulation results and risk assessment data from the previous stage, a multi-level early warning release mechanism was established: first, according to the arrival time of the fire front and the intensity of thermal radiation, the affected towers were divided into three warning levels: red (immediate disposal), orange (emergency preparation) and yellow (continuous monitoring), and early warning information was pushed to the relevant responsible units through the power dispatching network with a millisecond delay. Each warning contains structured data such as the precise tower number, GPS coordinates, expected impact time and risk type. In terms of generating disposal recommendations, the system integrates three databases: power grid topology analysis, emergency resource distribution, and fire extinguishing solution library. A multi-objective optimization algorithm is used to generate the optimal disposal strategy: For red warning towers, the system automatically generates the optimal power outage plan in combination with power grid flow calculations to ensure that the fault section is isolated while maintaining the stable operation of the power grid to the greatest extent possible. It also comprehensively considers surrounding fire water sources (the locations of fire hydrants and natural water sources within a 3-kilometer radius are obtained through the GIS database), road conditions (based on real-time traffic data), and the status of drone clusters to recommend the fastest accessible fire extinguishing resource deployment path. For orange warning areas, preventive measures are initiated, including adjusting the load of adjacent lines and preparing mobile fire extinguishing equipment. All warning information is visualized in three dimensions through the digital twin platform, supporting multi-perspective viewing of the spatial relationship between the fire and transmission equipment, and providing historical records of similar case disposals for reference. This intelligent warning and disposal system effectively solves the problems of untimely warnings and insufficient decision-making basis, achieving closed-loop management from risk warnings to disposal recommendations.

[0064] It effectively solves the problems of smoke obstruction monitoring failure and insufficient nighttime prediction accuracy caused by reliance on a single data source in existing technologies. It improves the fire point positioning accuracy and shortens the prediction response time through multi-source data fusion, and reduces the prediction error of the fire front arrival time by using dynamic model correction. It significantly improves the wildfire early warning capability in complex environments and provides high-precision and real-time technical support for disaster prevention and mitigation of power systems.

[0065] The second embodiment of this application is:

[0066] Based on the first embodiment, please refer to Figure 2 ,in, Figure 2 This is a principle block diagram of a transmission line wildfire spread range prediction system based on multi-factor fusion according to the second embodiment of the present invention.

[0067] The transmission line wildfire spread range prediction system based on multi-factor fusion in this embodiment includes an acquisition module 201 , an alignment module 202 , a feature extraction module 203 , a prediction module 204 , a deduction module 205 and a decision module 206 .

[0068] In this specific embodiment, the alignment module 202 is connected to the acquisition module 201, the feature extraction module 203 is connected to the alignment module 202, the prediction module 204 is connected to the feature extraction module 203, the deduction module 205 is connected to the prediction module 204, and the decision module 206 is connected to the deduction module 205;

[0069] The acquisition module 201 is used to collect multi-source monitoring data in real time, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data and lidar point cloud data;

[0070] The alignment module 202 is used to perform spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database;

[0071] The feature extraction module 203 is used to extract fire scene feature information from infrared thermal imaging data and visible light image data based on image processing technology, calculate the spatial distance between the fire point and the power transmission line in combination with the lidar point cloud data, and obtain vegetation combustion characteristic parameters;

[0072] The prediction module 204 is used to build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation;

[0073] The deduction module 205 is used to simulate the fire spread path and development trend based on the calculation results of the prediction model, combined with the three-dimensional model of the transmission line and terrain data, and evaluate the impact of the fire on the transmission line;

[0074] The decision module 206 is used to generate warning information and handling suggestions.

[0075] The multi-factor fusion-based transmission line wildfire spread prediction system of this embodiment, through the collaborative work of the acquisition module 201, the alignment module 202, the feature extraction module 203, the prediction module 204, the deduction module 205, and the decision module 206, establishes a complete technical chain from data acquisition to early warning decision-making. The system first collects infrared thermal imaging, visible light imagery, micrometeorological data, Beidou positioning, and lidar point cloud data in real time through a multi-source sensor network. It then performs high-precision spatiotemporal alignment to establish a fusion database. It extracts fire scene features based on image processing algorithms and calculates fire line distance and vegetation parameters based on three-dimensional point cloud data. It predicts fire spread using a hybrid model combining physical mechanisms and neural networks. It implements three-dimensional dynamic deduction and risk assessment using a cellular automation algorithm. Finally, it generates a graded early warning and optimized disposal plan. This provides efficient and accurate technical support for power grid wildfire prevention and control.

[0076] The above disclosure is merely one or more preferred embodiments of the present application and is not intended to limit the scope of the present application. A person skilled in the art will understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present application are still within the scope of the present application.

Claims

1. A method for predicting the spread of wildfires along power transmission lines based on multi-factor fusion, characterized in that: The following steps are involved: Real-time collection of multi-source monitoring data, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data, and lidar point cloud data; Perform spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database; Using image processing technology, fire scene feature information is extracted from infrared thermal imaging data and visible light image data. The spatial distance between the fire point and the power transmission line is calculated by combining the lidar point cloud data, and the vegetation burning characteristic parameters are obtained. Build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation; Based on the prediction model calculation results, combined with the three-dimensional model of the transmission line and terrain data, the fire spread path and development trend are simulated to assess the impact of the fire on the transmission line; Generate early warning information and disposal suggestions.

2. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 1, characterized in that: Real-time collection of multi-source monitoring data, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data and lidar point cloud data, including: Micrometeorological data includes wind speed, wind direction, temperature and humidity parameters, and lidar point cloud data includes tower location information, conductor sag data and vegetation distribution data in the transmission line corridor.

3. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 2, characterized in that: Perform spatiotemporal alignment on the collected multi-source monitoring data to establish a multi-source fusion database. The specific steps include: Perform time synchronization processing on infrared thermal imaging data, visible light image data, micrometeorological data, Beidou positioning data, and lidar point cloud data, so that the time stamp deviation of each data collection time does not exceed 10 seconds; Differential positioning is used to process Beidou positioning data to obtain fire point location data; The LiDAR point cloud data is gridded, where the fire front area uses the first resolution grid and other areas use the second resolution grid; The multi-source monitoring data after synchronization and grid processing are stored in a unified format to form a multi-source fusion database.

4. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 3, characterized in that: Using image processing technology, we extract fire scene feature information from infrared thermal imaging data and visible light image data. We then combine this with LiDAR point cloud data to calculate the spatial distance between the fire point and the power transmission line, and obtain vegetation combustion characteristic parameters. The specific steps include: Perform noise reduction on infrared thermal imaging data and use adaptive median filtering algorithm to eliminate image noise; Perform edge detection on the infrared thermal imaging data and visible light image data after noise reduction to extract the contour features of the fire scene; The fire scene contour features are spatially registered with the LiDAR point cloud data to calculate the three-dimensional distance between the fire point and the nearest transmission line conductor; Based on the vegetation point cloud features in the lidar point cloud data, the vegetation height distribution information and vegetation density information are extracted, and combined with the preset vegetation burning characteristic parameter library, the vegetation burning characteristic parameters of the current fire scene are obtained.

5. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 4, characterized in that: Construct a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation. The specific steps include: Construct a fire spread prediction model that includes a physical calculation module and a neural network correction module; The environmental parameters in the multi-source fusion database are input into the physical calculation module, and the basic fire spread speed is calculated using the Rothermel surface fire spread equation; The historical disaster case data and real-time monitoring data in the multi-source fusion database are input into the neural network correction module to dynamically correct the basic fire spread speed and obtain the corrected fire spread parameters.

6. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 5, characterized in that: Construct a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation: The neural network correction module updates the model parameters every 30 minutes based on the latest collected monitoring data.

7. The method for predicting the spread of wildfires on power transmission lines based on multi-factor fusion according to claim 5, characterized in that: Based on the prediction model calculation results, combined with the three-dimensional model of the transmission line and terrain data, the fire spread path and development trend are simulated to assess the impact of the fire on the transmission line. The specific steps include: Based on the corrected fire spread parameters, combined with the digital elevation model and the three-dimensional model of the transmission line, the cellular automation algorithm is used to simulate the fire spread path in three-dimensional space. Calculate the estimated time for the fire front to reach each tower location based on the fire spread path; Assess the extent of the fire's impact on power transmission lines; Generate impact assessment results including affected tower number, estimated arrival time and risk level.

8. A transmission line wildfire spread prediction system based on multi-factor fusion, applicable to the transmission line wildfire spread prediction method based on multi-factor fusion as claimed in claim 1, characterized in that: It includes an acquisition module, an alignment module, a feature extraction module, a prediction module, a deduction module and a decision module, wherein the alignment module is connected to the acquisition module, the feature extraction module is connected to the alignment module, the prediction module is connected to the feature extraction module, the deduction module is connected to the prediction module, and the decision module is connected to the deduction module; The acquisition module is used to collect multi-source monitoring data in real time, including infrared thermal imaging data, visible light image data, micro-meteorological data, Beidou positioning data and lidar point cloud data; The alignment module is used to perform spatiotemporal alignment processing on the collected multi-source monitoring data to establish a multi-source fusion database; The feature extraction module is used to extract fire scene feature information from infrared thermal imaging data and visible light image data based on image processing technology, calculate the spatial distance between the fire point and the transmission line in combination with the lidar point cloud data, and obtain vegetation combustion characteristic parameters; The prediction module is used to build a fire spread prediction model and input the processed multi-source monitoring data into the prediction model for calculation; The deduction module is used to simulate the fire spread path and development trend based on the calculation results of the prediction model, combined with the three-dimensional model of the transmission line and terrain data, and evaluate the impact of the fire on the transmission line; The decision-making module is used to generate early warning information and disposal suggestions.

Images