A method for planning a task path for power engineering surveying and mapping based on a UAV swarm

Abstract

The application discloses a power engineering surveying and mapping task path planning method based on a UAV swarm, and comprises the following steps: S1, multi-source data acquisition and three-dimensional modeling.The application solves many defects of existing UAV power engineering surveying and mapping in all directions through three-dimensional ground simulation path planning, UAV swarm cooperative operation, dynamic self-adaptive path correction, unified swarm scheduling and multi-constraint intelligent fusion planning, can not only realize high-precision surveying and mapping in accordance with terrain undulations, eliminate the risk of missing surveying and aircraft collision, and improve operation safety, but also can greatly improve surveying and mapping efficiency by relying on multi-machine cooperation, can respond to various sudden conditions in the field in real time, ensure continuous operation without interruption, realize orderly operation of the swarm, no airspace conflict, no repeated collection, can realize global path optimization by scientific weight distribution, reduce rework rate and surveying and mapping cost, and perfectly adapt to the power engineering surveying and mapping demand of various complex terrains such as mountainous areas and dense forests.

Description

Technical Field

[0001] This invention relates to the field of power engineering technology, and in particular to a method for planning the path of power engineering surveying tasks based on UAV swarms. Background Technology

[0002] Power engineering surveying is a core part of power line site selection, tower positioning, engineering design, and construction acceptance. The surveying accuracy and work efficiency directly determine the construction cycle and construction safety of power projects. Traditional power engineering surveying mostly adopts manual reconnaissance and single UAV surveying. As power projects extend to remote mountainous areas and complex terrain areas, conventional surveying path planning methods can no longer meet the operational needs. Therefore, UAV power surveying path planning technology has emerged on the market. Existing UAV power line mapping path planning technologies generally suffer from the following drawbacks: 1. Flight paths are mostly planned in two-dimensional planes, lacking the ability to simulate three-dimensional terrain undulations, making them prone to elevation deviations, uneven point cloud data density, missed measurements in local areas, and safety accidents such as UAV collisions with mountains and trees, which are particularly evident in mountainous and dense forest scenarios; 2. The operating range of a single UAV is limited, and its short endurance is a significant drawback, making large-scale, long-distance power line mapping extremely time-consuming and inefficient; 3. Path planning is mostly statically preset, unable to cope with unexpected situations in the field, such as GPS signal obstruction, strong winds, and other unforeseen circumstances. 4. Due to obstacles and equipment failures, the lack of a real-time replanning mechanism can easily lead to surveying interruptions and data invalidation; 5. The lack of unified scheduling in multi-drone collaborative operations results in chaotic task allocation and overlapping routes among drones, which can easily lead to problems such as duplicate data collection, airspace conflicts, and large positioning errors; 6. The low integration of multiple constraints such as terrain, obstacles, weather, and endurance, along with subjective path weight settings, makes it impossible to achieve globally optimal planning, resulting in high rework rates and high surveying costs. In summary, this application proposes a path planning method for power engineering surveying tasks based on drone swarms. Summary of the Invention

[0003] Based on the technical problems existing in the background technology, this invention proposes a path planning method for power engineering surveying tasks based on UAV swarms.

[0004] This invention proposes a path planning method for power engineering surveying tasks based on UAV swarms, comprising the following steps: S1: Multi-source data acquisition and 3D modeling: Acquire multi-source geographic data of the target mapping area, process the data to construct a refined 3D terrain model, and mark no-fly zones, obstacles and terrain parameters; S2: Initialization and task allocation of UAV swarm formation: Construct a UAV swarm formation in master-slave mode, use adaptive grid partitioning method to divide the survey area, and complete the balanced allocation of tasks for each UAV; S3: Multi-constraint fusion 3D global path planning: Construct a multi-constraint objective function, use the entropy weight method to determine the constraint weights, and combine the improved algorithm and the bee colony algorithm to plan a 3D global terrain-following path; S4: Swarm Cooperative Dynamic Path Correction and Obstacle Avoidance: During swarm operations, the host monitors the status of each machine and the surrounding environment in real time, and initiates dynamic replanning in response to emergencies to achieve cooperative obstacle avoidance and flight path correction; S5: Real-time verification of surveying quality and fine-tuning of the route: Real-time verification of surveying data quality, local supplementary surveying for unqualified areas, and fine-tuning of the route; S6: Work Completion and Bee Swarm Recovery: After the survey is completed, guide the bee swarm back in an orderly manner and summarize the survey results.

[0005] Preferably, in step S1, the resolution of the three-dimensional terrain model is not less than 0.5 meters, and the point cloud modeling is encrypted for complex terrain to eliminate terrain distortion and data gaps; The specific logical steps of S1 are as follows: S101: Select the target survey area and collect full-coverage, multi-dimensional geographic data. The collected data includes 0.5 and higher resolution digital elevation model data, aerial satellite remote sensing imagery, ground laser point cloud data, existing power facility vector data, surface obstacle data, medium- and long-term meteorological data, airspace control information, and topographic and geological survey data. All data are uniformly collected using the National Geodetic Coordinate System 2000 to ensure coordinate consistency in subsequent data fusion and modeling, and to prevent coordinate offset and misalignment errors. S102: Performs noise reduction, filtering, and normalization on the raw acquired data, removing outliers, redundant data, and noise points, repairing data holes, and improving data quality. It uses a Gaussian filtering algorithm to remove point cloud noise, with the following formula: ,in Here, x and y are the Gaussian filter weights, and x and y are the plane coordinate offsets of the point cloud data. This is the filtering smoothing coefficient, with a value range of 0.5-1.0; When performing data normalization, various data types are mapped to a unified numerical range. The formula used is as follows: ,in The data is normalized and ranges from [0,1]. x represents the original collected data. The minimum value among the collected data. The maximum value in the collected data; S103: Based on preprocessed high-quality DEM and point cloud data, an irregular triangular mesh 3D terrain model is constructed to restore the real surface undulations, gullies, steep slopes, and vegetation cover morphology, ensuring complete terrain details without distortion or voids. It employs a triangular mesh growing algorithm for modeling, and the plane elevation fitting formula is: ,in Let (x, y) be the ground elevation value at coordinates (x, y). , , These are the plane fitting coefficients. Ground plane coordinates; S104: In the 3D terrain model, no-fly zones, obstacles, and hazardous flight areas are marked, including steep slopes, dense forests, rivers, buildings, existing high-voltage lines, and signal blind spots. Core terrain parameters are extracted for subsequent path planning constraint calculations, with a focus on extracting terrain slope parameters. Surface roughness, visibility conditions, and safe altitude threshold parameters are also extracted simultaneously. These parameters are compiled into a constraint dataset and stored in the path planning database to provide data support for subsequent swarm flight path planning. The path planning constraint calculation formula is as follows: ,in This represents the local slope value of the ground. Let x be the rate of change of elevation along the x-direction. y is the rate of change of elevation along the y-direction.

[0006] Preferably, the specific logical steps of S2 are as follows: S201: Swarm formation initialization: Based on the area of ​​the surveying area, the complexity of the terrain and the project schedule, determine the total number of drones in the swarm, build a master-slave formation with 1 master and multiple slaves. The master is responsible for global scheduling and data aggregation, and the slaves are responsible for segmented surveying. At the same time, verify the endurance, payload and communication hardware status of all drones, build a stable communication topology, and form a swarm system that can work collaboratively. S202: Determination of Survey Area Boundaries and Mesh Preprocessing: Extraction of Global Boundary Coordinates from the 3D Terrain Model The formula for calculating the total area of ​​the surveyed area is as follows: ,in , The x-axis represents the extreme values. , The extreme values ​​are represented by the vertical axis. To determine the total area of ​​the region, the base mesh size is then set to complete the region preprocessing. The formula used is: ,in Where k is the maximum safe flight distance for a single UAV, and k is the safety redundancy coefficient, with a value of 1.2. The grid density coefficient is 5-10, and N is the total number of frames in the bee colony. The base grid side length is in meters (m). S203: Adaptive Mesh Block Partitioning: First calculate the initial number of meshes. M represents the total number of grid cells, and the local terrain complexity is calculated by combining this with the terrain slope. Then dynamically adjust the grid side length. Ultimately, the entire area is divided into several non-overlapping sub-blocks that conform to the terrain, each block having an area of ​​[area missing]. ,in Let be the complexity of the i-th block, with a value between 0 and 1. The average slope of the block. , This represents the extreme slope of the region. This represents the actual grid side length. This is the terrain influence coefficient, with a value ranging from 0.1 to 0.3. S204: Task Balance Allocation and Drone Capability Matching: First calculate the maximum operating area of ​​a single drone. Then calculate the block operation cost. Finally, with task balancing as the objective, an optimization function is established. Each block is uniquely allocated, and the workload per machine is limited. This indicates that block i is assigned to drone j, and 0 otherwise, thus achieving a fair and reasonable task allocation. For cruising speed, For extended battery life on a full charge, For safety margin factor, Considering the overall operational costs, This represents the terrain cost weight, with a value ranging from 0.2 to 0.5.

[0007] Preferably, the specific logical steps of S3 are as follows: S301: Select the core constraints of power engineering surveying, complete the quantitative processing, and construct a constraint set. The constraint items include: terrain fit, flight safety distance, energy consumption cost, surveying coverage, airspace conflict-free, and line-of-sight conditions. Transform terrain undulation, obstacle location, remaining range, and meteorological wind speed into numerical constraint indicators, eliminate fuzzy qualitative indicators, and ensure that subsequent calculations are objective and accurate. S302: Standardize all constraints, using the following formula: And calculate the information entropy of each constraint. Finally, the constraint weights are obtained. ;in Let j be the standardized value of the constraint for the i-th path point. , Let j be the constrained extreme value of the j-th term. The total number of path points. Let the entropy be the constraint information of the j-th term. Let be the weight of the j-th constraint, and n be the total number of constraints. The sum of the weights is 1, thus eliminating the bias of manual weighting. S303: Integrating various constraints and their corresponding weights, a global path-optimal objective function is established, the formula of which is: ,in This represents the total cost of the global path. Let $j$ be the single cost of the j-th constraint on path $P$. S304: The total cost F is used as the heuristic function for the improved algorithm, and its formula is as follows: By introducing a terrain cost correction term, the heuristic function is improved to... The bee colony algorithm is used to optimize the node search direction, avoid local optima, and generate a 3D global path that fits the terrain throughout, is collision-free, and has low energy consumption. The total estimated value for node n. The actual cost from the starting point to node n. Estimate the cost from node n to the destination. This is the terrain penalty coefficient, with a value ranging from 0.3 to 0.6. For terrain complexity; S305: Verify the generated initial 3D path, check the UAV's safe distance, mapping overlap, and endurance capacity, eliminate excessive road sections, fine-tune the path inflection points, ensure that the flight path follows the terrain throughout, the surveying equipment's acquisition angle meets the standards, the point cloud coverage is uniform, there are no missed measurements or collision risks, and finally determine the exclusive global path for each UAV.

[0008] Preferably, the specific logical steps of S4 are as follows: S401: Real-time status and environmental awareness: The host UAV collects flight data from each slave UAV at fixed intervals, including positioning coordinates, flight altitude, speed, remaining battery power, and communication signal strength. At the same time, it uses airborne radar and visual sensors to detect sudden obstacles in the surrounding area in real time, obtains information on the location, size, and speed of the obstacles, and collects real-time meteorological data to form a global awareness data pool. S402: Abnormal Situation Identification and Risk Assessment: Set safety thresholds, compare real-time data with the thresholds to determine if there are any abnormalities such as obstacle intrusion, insufficient power, signal obstruction, or excessive weather conditions, and calculate the risk coefficient. ,in To assess the overall risk factor, , , For each risk weight, and , E represents the real-time distance between the drone and the obstacle, and E represents the remaining battery power. Full charge R represents the signal attenuation rate, ranging from 0 to 1; the larger the value of R, the higher the risk. S403: Swarm Airspace Conflict Detection: The host calculates the distance between each UAV in real time to determine whether there is a risk of intersecting flight paths or overlapping airspace. The formula for the safe distance between UAVs is: ,in The minimum safe distance is expressed in meters (m), and k is the safety factor, which ranges from 8 to 12. The average cruising speed of the bee colony, in m / s. If the real-time interval is less than... It was determined that there was a risk of conflict. S404: Local Dynamic Path Replanning: To address anomaly risks, a fast path correction algorithm is activated, preserving valid segments of the original path and generating local detour paths in obstacle areas. The local path cost function is... ,in For the optimal cost of the local path, To correct the path length, The risk penalty coefficient ranges from 0.4 to 0.7, and R is the comprehensive risk coefficient. The route with the lowest cost is selected as the corrected route. S405: Swarm Coordinated Scheduling and Command Issuance: The host unit uniformly controls the flight speed, heading, and altitude of each drone, avoids conflicting flight paths, plans priority return routes for drones with low battery power, schedules backup drones to fill in the gaps, maintains the continuity of swarm operations, and the command response time does not exceed 1 second. S406: Corrected Path Tracking and Closed-Loop Verification. Each UAV executes a corrected route, and the host compares the actual trajectory with the planned route in real time to calculate the trajectory deviation. The formula is as follows: If the deviation exceeds the limit, the flight path will be fine-tuned again until it returns to a safe path. This is the deviation distance, in meters. , For actual coordinates, , For planning coordinates.

[0009] Preferably, the specific logical steps of S5 are as follows: S501: Real-time data transmission and preprocessing: Each UAV will transmit the collected laser point cloud and image data back to the back-end processing terminal in real time. The terminal will perform noise reduction and stitching preprocessing on the raw data, remove invalid data and blurry images, classify and organize them according to the surveying blocks, and synchronously record the collection coordinates, flight attitude and collection time of the corresponding data to establish a complete data ledger. S502: Quantitative Verification of Surveying and Mapping Quality Indicators: Four core indicators—point cloud density, forward overlap, lateral overlap, and data integrity—are selected and quantitatively assessed item by item against the threshold values ​​of power engineering surveying standards. The formula for calculating point cloud density is as follows: The formula for calculating the heading overlap is: The formula for calculating lateral overlap is: ,in The value represents the local point cloud density, expressed in points per square meter (m²). N represents the total number of point clouds collected in the local area, and S represents the area of ​​the collected region, also expressed in square meters (m²). For heading overlap, This refers to the lateral overlap rate. The length of overlap in the heading of adjacent flight strips. For single-sail length, The lateral overlap width of adjacent flight strips. For single-pass width; S503: Determination and Location of Non-conforming Areas: Compare each measured indicator with the pass / fail threshold to determine non-conforming surveying areas, and set the comprehensive quality judgment formula as follows: When Q is less than the set passing score, the corresponding area is determined to be a non-compliant retest area, and the three-dimensional coordinate range of that area is locked. For the overall quality score, , , These are the indicator weights, and the sum of the weights is 1. For standard point cloud density, For standard heading overlap, Standard lateral overlap; S504: Supplementary Survey Route Planning and Fine-tuning: For non-compliant areas, generate localized, refined supplementary survey routes, and calculate the length of the supplementary survey route following the shortest flight path principle. ,in To supplement the measurement of the flight segment length, , , To remeasure the coordinates of the starting point, , , To supplement the measurement of the endpoint coordinates, the original flight path was slightly adjusted, the drone's terrain-following flight altitude and data collection angle were maintained, obstacles and airspace conflicts were avoided, and the supplementary measurement area was fully covered without omissions. S505: Supplementary Survey Data Verification and Route Closure: After completing the local supplementary survey, re-verify the quality of the supplementary survey data until all indicators meet the standards and the comprehensive score is qualified, confirming that there are no missed surveys or distorted areas; update the global route file, record the supplementary survey points and fine-tuned flight segments, form a complete survey route and data closure loop, and prevent rework and re-survey in the later stage.

[0010] Preferably, in step S6, after the entire area is surveyed and the data is verified, the host UAV issues a return command, plans a unified safe return path, guides each UAV to return in an orderly manner and land accurately, and after recovery, summarizes all survey data, removes redundant information, and splices together to form a complete power engineering survey result for subsequent power engineering design and construction.

[0011] Preferably, in step S4, during the swarm collaborative dynamic path correction and obstacle avoidance process, the host drone refreshes the entire swarm status at a cycle of no more than 0.5 seconds to identify sudden obstacles, signal blockage, insufficient power, and sudden weather changes in real time.

[0012] Compared with existing technologies, the beneficial effects of this invention are: It adopts a refined three-dimensional terrain-following path planning, completely abandoning the traditional two-dimensional planar flight path. It relies on a high-precision three-dimensional terrain model to design a flight path that fits the terrain undulation. It can accurately match the ground elevation difference, effectively eliminate elevation deviation, uneven point cloud density, and local missed measurement problems. At the same time, it avoids obstacles such as mountains and forests, reducing the risk of collision from the root and greatly improving the surveying accuracy and operational safety in complex terrain. It is especially suitable for harsh terrain scenarios such as mountainous areas and dense forests. By adopting a drone swarm collaborative operation mode, the limitations of small operating range and insufficient battery life of single drones are broken. By carrying out surveying and mapping work in sections by multiple drones simultaneously, the coverage of a single operation is greatly expanded, and the surveying and mapping period of large-scale and long-distance power transmission lines is significantly shortened. Compared with traditional single drone surveying, the operation efficiency is increased several times, effectively solving the pain points of excessive time consumption and low efficiency, and reducing manpower and time costs. By incorporating real-time environmental perception and dynamic replanning mechanisms, it can quickly respond to unexpected situations in the field, such as GPS signal obstruction, strong winds and severe weather, temporary obstacles, and equipment failures. It can correct flight routes in real time, avoid risks, ensure uninterrupted surveying and mapping operations, avoid data loss and rework, and greatly improve the stability of field operations. By centrally managing and balancing tasks, the UAV mapping area is rationally divided, and independent routes are precisely planned. This completely solves the problems of chaotic task division, overlapping routes, and airspace conflicts among multiple UAVs, and eliminates the situation of repeated data collection and large positioning errors. It ensures that the swarm operation has a clear division of labor, good cooperation, and seamless data collection. 5. It quantitatively integrates multiple constraints such as terrain, obstacles, weather, and range, abandons the drawbacks of subjective weight setting, and automatically allocates weights and optimizes paths through scientific algorithms to achieve global path optimization, effectively reducing the probability of rework and surveying costs, making path planning more reasonable and more in line with actual engineering needs, and adapting to various power engineering surveying scenarios. This invention comprehensively addresses many shortcomings of existing UAV power surveying by employing three-dimensional terrain-following path planning, UAV swarm collaborative operation, dynamic adaptive path correction, unified swarm scheduling, and multi-constraint intelligent fusion planning. It not only achieves high-precision surveying that conforms to terrain undulations, eliminating the risk of missed measurements and collisions and improving operational safety, but also significantly improves surveying efficiency through multi-UAV collaboration. Furthermore, it can respond to various emergencies in the field in real time, ensuring continuous and uninterrupted operation, achieving orderly swarm operation, no airspace conflicts, and no duplicate data collection. Moreover, it can achieve global path optimization through scientific weight allocation, reducing rework rates and surveying costs, and perfectly adapts to the power engineering surveying needs of various complex terrains such as mountainous areas and dense forests. Attached Figure Description

[0013] Figure 1 This is a flowchart of a path planning method for power engineering surveying tasks based on UAV swarms, as proposed in this invention. Detailed Implementation

[0014] The present invention will be further explained below with reference to specific embodiments. Example

[0015] Reference Figure 1 This embodiment proposes a path planning method for power engineering surveying tasks based on UAV swarms, including the following steps: S1: Multi-source data acquisition and 3D modeling: Acquire multi-source geographic data of the target mapping area, process the data to construct a refined 3D terrain model, and mark no-fly zones, obstacles and terrain parameters; The resolution of the 3D terrain model is no less than 0.5 meters, and the point cloud modeling is encrypted for complex terrain to eliminate terrain distortion and data holes; The specific logical steps are as follows: S101: Select the target survey area and collect full-coverage, multi-dimensional geographic data. The collected data includes 0.5 and higher resolution digital elevation model data, aerial satellite remote sensing imagery, ground laser point cloud data, existing power facility vector data, surface obstacle data, medium- and long-term meteorological data, airspace control information, and topographic and geological survey data. All data are uniformly collected using the National Geodetic Coordinate System 2000 to ensure coordinate consistency in subsequent data fusion and modeling, and to prevent coordinate offset and misalignment errors. S102: Performs noise reduction, filtering, and normalization on the raw acquired data, removing outliers, redundant data, and noise points, repairing data holes, and improving data quality. It uses a Gaussian filtering algorithm to remove point cloud noise, with the following formula: ,in Here, x and y are the Gaussian filter weights, and x and y are the plane coordinate offsets of the point cloud data. This is the filtering smoothing coefficient, with a value range of 0.5-1.0; When performing data normalization, various data types are mapped to a unified numerical range. The formula used is as follows: ,in The data is normalized and ranges from [0,1]. x represents the original collected data. The minimum value among the collected data. The maximum value in the collected data; S103: Based on preprocessed high-quality DEM and point cloud data, an irregular triangular mesh 3D terrain model is constructed to restore the real surface undulations, gullies, steep slopes, and vegetation cover morphology, ensuring complete terrain details without distortion or voids. It employs a triangular mesh growing algorithm for modeling, and the plane elevation fitting formula is: ,in Let (x, y) be the ground elevation value at coordinates (x, y). , , These are the plane fitting coefficients. Ground plane coordinates; S104: In the 3D terrain model, no-fly zones, obstacles, and hazardous flight areas are marked, including steep slopes, dense forests, rivers, buildings, existing high-voltage lines, and signal blind spots. Core terrain parameters are extracted for subsequent path planning constraint calculations, with a focus on extracting terrain slope parameters. Surface roughness, visibility conditions, and safe altitude threshold parameters are also extracted simultaneously. These parameters are compiled into a constraint dataset and stored in the path planning database to provide data support for subsequent swarm flight path planning. The path planning constraint calculation formula is as follows: ,in This represents the local slope value of the ground. Let x be the rate of change of elevation along the x-direction. This represents the rate of change of elevation along the y-direction; S2: Initialization and task allocation of UAV swarm formation: Construct a UAV swarm formation in master-slave mode, use adaptive grid partitioning method to divide the survey area, and complete the balanced allocation of tasks for each UAV; The specific logical steps are as follows: S201: Swarm formation initialization: Based on the area of ​​the surveying area, the complexity of the terrain and the project schedule, determine the total number of drones in the swarm, build a master-slave formation with 1 master and multiple slaves. The master is responsible for global scheduling and data aggregation, and the slaves are responsible for segmented surveying. At the same time, verify the endurance, payload and communication hardware status of all drones, build a stable communication topology, and form a swarm system that can work collaboratively. S202: Determination of Survey Area Boundaries and Mesh Preprocessing: Extraction of Global Boundary Coordinates from the 3D Terrain Model The formula for calculating the total area of ​​the surveyed area is as follows: ,in , where are the extreme values ​​of the x-axis. , The extreme values ​​are represented by the vertical axis. To determine the total area of ​​the region, the base mesh size is then set to complete the region preprocessing. The formula used is: ,in Where k is the maximum safe flight distance for a single UAV, and k is the safety redundancy coefficient, with a value of 1.2. The grid density coefficient is 5-10, and N is the total number of frames in the bee colony. The base grid side length is in meters (m). S203: Adaptive Mesh Block Partitioning: First calculate the initial number of meshes. M represents the total number of grid cells, and the local terrain complexity is calculated by combining this with the terrain slope. Then dynamically adjust the grid side length. Ultimately, the entire area is divided into several non-overlapping sub-blocks that conform to the terrain, each block having an area of ​​[area missing]. ,in Let be the complexity of the i-th block, with a value between 0 and 1. The average slope of the block. , This represents the extreme slope of the region. This represents the actual grid side length. This is the terrain influence coefficient, with a value ranging from 0.1 to 0.3. S204: Task Balance Allocation and Drone Capability Matching: First calculate the maximum operating area of ​​a single drone. Then calculate the block operation cost. Finally, with task balancing as the objective, an optimization function is established. Each block is uniquely allocated, and the workload per machine is limited. This indicates that block i is assigned to drone j, and 0 otherwise, thus achieving a fair and reasonable task allocation. For cruising speed, For extended battery life on a full charge, For safety margin factor, Considering the overall operational costs, This represents the terrain cost weight, with a value ranging from 0.2 to 0.5. S3: Multi-constraint fusion 3D global path planning: Construct a multi-constraint objective function, use the entropy weight method to determine the constraint weights, and combine the improved algorithm and the bee colony algorithm to plan a 3D global terrain-following path; The specific logical steps are as follows: S301: Select the core constraints of power engineering surveying, complete the quantitative processing, and construct a constraint set. The constraint items include: terrain fit, flight safety distance, energy consumption cost, surveying coverage, airspace conflict-free, and line-of-sight conditions. Transform terrain undulation, obstacle location, remaining range, and meteorological wind speed into numerical constraint indicators, eliminate fuzzy qualitative indicators, and ensure that subsequent calculations are objective and accurate. S302: Standardize all constraints, using the following formula: And calculate the information entropy of each constraint. Finally, the constraint weights are obtained. ;in Let j be the standardized value of the constraint for the i-th path point. , Let j be the constrained extreme value of the j-th term. The total number of path points. Let the entropy be the constraint information of the j-th term. Let be the weight of the j-th constraint, and n be the total number of constraints. The sum of the weights is 1, thus eliminating the bias of manual weighting. S303: Integrating various constraints and their corresponding weights, a global path-optimal objective function is established, the formula of which is: ,in This represents the total cost of the global path. Let $j$ be the single cost of the $j$ constraint on path $P$. S304: The total cost F is used as the heuristic function for the improved algorithm, and its formula is as follows: By introducing a terrain cost correction term, the heuristic function is improved to... The bee colony algorithm is used to optimize the node search direction, avoid local optima, and generate a 3D global path that fits the terrain throughout, is collision-free, and has low energy consumption. The total valuation for node n. The actual cost from the starting point to node n. Estimate the cost from node n to the destination. This is the terrain penalty coefficient, with a value ranging from 0.3 to 0.6. For terrain complexity; S305: Verify the generated initial 3D path, check the UAV's safe distance, mapping overlap, and endurance capacity, eliminate excessive road sections, fine-tune the path inflection points, ensure that the flight path follows the terrain throughout the entire route, that the surveying equipment's acquisition angle meets the standards, that the point cloud coverage is uniform, that there are no missed measurements or collision risks, and finally determine the exclusive global path for each UAV. S4: Swarm Cooperative Dynamic Path Correction and Obstacle Avoidance: During swarm operations, the host monitors the status of each machine and the surrounding environment in real time, and initiates dynamic replanning in response to emergencies to achieve cooperative obstacle avoidance and flight path correction; The specific logical steps are as follows: S401: Real-time status and environmental awareness: The host UAV collects flight data from each slave UAV at fixed intervals, including positioning coordinates, flight altitude, speed, remaining battery power, and communication signal strength. At the same time, it uses airborne radar and visual sensors to detect sudden obstacles in the surrounding area in real time, obtains information on the location, size, and speed of the obstacles, and collects real-time meteorological data to form a global awareness data pool. S402: Abnormal Situation Identification and Risk Assessment: Set safety thresholds, compare real-time data with the thresholds to determine if there are any abnormalities such as obstacle intrusion, insufficient power, signal obstruction, or excessive weather conditions, and calculate the risk coefficient. ,in To assess the overall risk factor, , , For each risk weight, and , E represents the real-time distance between the drone and the obstacle, and E represents the remaining battery power. Full charge R represents the signal attenuation rate, ranging from 0 to 1; the larger the value of R, the higher the risk. S403: Swarm Airspace Conflict Detection: The host calculates the distance between each UAV in real time to determine whether there is a risk of intersecting flight paths or overlapping airspace. The formula for the safe distance between UAVs is: ,in The minimum safe distance is expressed in meters (m), and k is the safety factor, which ranges from 8 to 12. The average cruising speed of the bee colony, in m / s. If the real-time interval is less than... It was determined that there was a risk of conflict. S404: Local Dynamic Path Replanning: To address anomaly risks, a fast path correction algorithm is activated, preserving valid segments of the original path and generating local detour paths in obstacle areas. The local path cost function is... ,in For the optimal cost of the local path, To correct the path length, The risk penalty coefficient ranges from 0.4 to 0.7, and R is the comprehensive risk coefficient. The route with the lowest cost is selected as the corrected route. S405: Swarm Coordinated Scheduling and Command Issuance: The host unit uniformly controls the flight speed, heading, and altitude of each drone, avoids conflicting flight paths, plans priority return routes for drones with low battery power, schedules backup drones to fill in the gaps, maintains the continuity of swarm operations, and the command response time does not exceed 1 second. S406: Corrected Path Tracking and Closed-Loop Verification. Each UAV executes a corrected route, and the host compares the actual trajectory with the planned route in real time to calculate the trajectory deviation. The formula is as follows: If the deviation exceeds the limit, the flight path will be fine-tuned again until it returns to a safe path. This is the deviation distance, in meters. , For actual coordinates, , For planning coordinates; During the swarm collaborative dynamic path correction and obstacle avoidance process, the host drone refreshes the status of the entire swarm at a cycle of no more than 0.5 seconds, and identifies sudden obstacles, signal blockage, low battery and sudden weather changes in real time. S5: Real-time verification of surveying quality and fine-tuning of the route: Real-time verification of surveying data quality, local supplementary surveying for unqualified areas, and fine-tuning of the route; The specific logical steps are as follows: S501: Real-time data transmission and preprocessing: Each UAV will transmit the collected laser point cloud and image data back to the back-end processing terminal in real time. The terminal will perform noise reduction and stitching preprocessing on the raw data, remove invalid data and blurry images, classify and organize them according to the surveying blocks, and synchronously record the collection coordinates, flight attitude and collection time of the corresponding data to establish a complete data ledger. S502: Quantitative Verification of Surveying and Mapping Quality Indicators: Four core indicators—point cloud density, forward overlap, lateral overlap, and data integrity—are selected and quantitatively assessed item by item against the threshold values ​​of power engineering surveying standards. The formula for calculating point cloud density is as follows: The formula for calculating the heading overlap is: The formula for calculating lateral overlap is: ,in The value represents the local point cloud density, expressed in points per square meter (m²). N represents the total number of point clouds collected in the local area, and S represents the area of ​​the collected region, also expressed in square meters (m²). For heading overlap, This refers to the lateral overlap rate. The overlap length of adjacent flight strips in the heading direction. For single-sail length, The lateral overlap width of adjacent flight strips. For single-pass width; S503: Determination and Location of Non-conforming Areas: Compare each measured indicator with the pass / fail threshold to determine non-conforming surveying areas, and set the comprehensive quality judgment formula as follows: When Q is less than the set passing score, the corresponding area is determined to be a non-compliant retest area, and the three-dimensional coordinate range of that area is locked. For the overall quality score, , , These are the indicator weights, and the sum of the weights is 1. For standard point cloud density, For standard heading overlap, Standard lateral overlap; S504: Supplementary Survey Route Planning and Fine-tuning: For non-compliant areas, generate localized, refined supplementary survey routes, and calculate the length of the supplementary survey route following the shortest flight path principle. ,in To supplement the measurement of the flight segment length, , , To remeasure the coordinates of the starting point, , , To supplement the measurement of the endpoint coordinates, the original flight path was slightly adjusted, the drone's terrain-following flight altitude and data collection angle were maintained, obstacles and airspace conflicts were avoided, and the supplementary measurement area was fully covered without omissions. S505: Supplementary Survey Data Verification and Route Closure: After completing the local supplementary survey, re-verify the quality of the supplementary survey data until all indicators meet the standards and the comprehensive score is qualified, confirming that there are no missed surveys and no distorted areas; update the global route file, record the supplementary survey points and fine-tuning segments, form a complete survey route and data closure, and prevent rework and re-survey in the later stage. S6: Work Completion and Swarm Recovery: After the survey is completed, guide the swarm to return in an orderly manner, summarize the survey results. After the entire area is surveyed and the data is verified, the host UAV issues a return command, plans a unified safe return path, guides each UAV to return in an orderly manner and land accurately. After recovery, summarize all survey data, remove redundant information, and stitch together to form a complete power engineering survey result for subsequent power engineering design and construction. This embodiment comprehensively addresses many shortcomings of existing UAV power surveying by employing 3D terrain-following path planning, UAV swarm collaborative operation, dynamic adaptive path correction, unified swarm scheduling, and multi-constraint intelligent fusion planning. It not only achieves high-precision surveying that conforms to terrain undulations, eliminating the risk of missed measurements and collisions and improving operational safety, but also significantly improves surveying efficiency through multi-UAV collaboration. Furthermore, it can respond to various emergencies in the field in real time, ensuring continuous and uninterrupted operation, achieving orderly swarm operation, no airspace conflicts, and no duplicate data collection. Moreover, it can achieve global path optimization through scientific weight allocation, reducing rework rates and surveying costs, perfectly adapting to the power engineering surveying needs of various complex terrains such as mountainous areas and dense forests.

[0016] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for path planning in power engineering surveying tasks based on UAV swarms, characterized in that, Includes the following steps: S1: Multi-source data acquisition and 3D modeling: Acquire multi-source geographic data of the target mapping area, process the data to construct a refined 3D terrain model, and mark no-fly zones, obstacles and terrain parameters; S2: Initialization and task allocation of UAV swarm formation: Construct a UAV swarm formation in master-slave mode, use adaptive grid partitioning method to divide the survey area, and complete the balanced allocation of tasks for each UAV; S3: Multi-constraint fusion 3D global path planning: Construct a multi-constraint objective function, use the entropy weight method to determine the constraint weights, and combine the improved algorithm and the bee colony algorithm to plan a 3D global terrain-following path; S4: Swarm Cooperative Dynamic Path Correction and Obstacle Avoidance: During swarm operations, the host monitors the status of each machine and the surrounding environment in real time, and initiates dynamic replanning in response to emergencies to achieve cooperative obstacle avoidance and flight path correction; S5: Real-time verification of surveying quality and fine-tuning of the route: Real-time verification of surveying data quality, local supplementary surveying for unqualified areas, and fine-tuning of the route; S6: Work Completion and Bee Swarm Recovery: After the survey is completed, guide the bee swarm back in an orderly manner and summarize the survey results.

2. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 1, characterized in that, In S1, the resolution of the three-dimensional terrain model is not less than 0.5 meters, and the point cloud modeling is encrypted for complex terrain to eliminate terrain distortion and data gaps. The specific logical steps of S1 are as follows: S101: Select the target survey area and collect full-coverage, multi-dimensional geographic data. The collected data includes 0.5 and higher resolution digital elevation model data, aerial satellite remote sensing imagery, ground laser point cloud data, existing power facility vector data, surface obstacle data, medium- and long-term meteorological data, airspace control information, and topographic and geological survey data. All data are uniformly collected using the National Geodetic Coordinate System 2000 to ensure coordinate consistency in subsequent data fusion and modeling, and to prevent coordinate offset and misalignment errors. S102: Performs noise reduction, filtering, and normalization on the raw acquired data, removing outliers, redundant data, and noise points, repairing data holes, and improving data quality. It uses a Gaussian filtering algorithm to remove point cloud noise, with the following formula: ,in Here, x and y are the Gaussian filter weights, and x and y are the plane coordinate offsets of the point cloud data. This is the filtering smoothing coefficient, with a value range of 0.5-1.0; When performing data normalization, various data types are mapped to a unified numerical range. The formula used is as follows: ,in The data is normalized and ranges from [0,1]. x represents the original collected data. The minimum value among the collected data. The maximum value in the collected data; S103: Based on preprocessed high-quality DEM and point cloud data, an irregular triangular mesh 3D terrain model is constructed to restore the real surface undulations, gullies, steep slopes, and vegetation cover morphology, ensuring complete terrain details without distortion or voids. It employs a triangular mesh growing algorithm for modeling, and the plane elevation fitting formula is: ,in Let (x, y) be the ground elevation value at coordinates (x, y). , , These are the plane fitting coefficients. Ground plane coordinates; S104: In the 3D terrain model, no-fly zones, obstacles, and hazardous flight areas are marked, including steep slopes, dense forests, rivers, buildings, existing high-voltage lines, and signal blind spots. Core terrain parameters are extracted for subsequent path planning constraint calculations, with a focus on extracting terrain slope parameters. Surface roughness, visibility conditions, and safe altitude threshold parameters are also extracted simultaneously. These parameters are compiled into a constraint dataset and stored in the path planning database to provide data support for subsequent swarm flight path planning. The path planning constraint calculation formula is as follows: ,in This represents the local slope value of the ground. Let x be the rate of change of elevation along the x-direction. y is the rate of change of elevation along the y-direction.

3. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 2, characterized in that, The specific logical steps of S2 are as follows: S201: Swarm formation initialization: Based on the area of ​​the surveying area, the complexity of the terrain and the project schedule, determine the total number of drones in the swarm, build a master-slave formation with 1 master and multiple slaves. The master is responsible for global scheduling and data aggregation, and the slaves are responsible for segmented surveying. At the same time, verify the endurance, payload and communication hardware status of all drones, build a stable communication topology, and form a swarm system that can work collaboratively. S202: Determination of Survey Area Boundaries and Mesh Preprocessing: Extraction of Global Boundary Coordinates from the 3D Terrain Model The formula for calculating the total area of ​​the surveyed area is as follows: ,in , The x-axis represents the extreme values. , The extreme values ​​are represented by the vertical axis. To determine the total area of ​​the region, the base mesh size is then set to complete the region preprocessing. The formula used is: ,in Where k is the maximum safe flight distance for a single UAV, and k is the safety redundancy coefficient, with a value of 1.

2. The grid density coefficient is 5-10, and N is the total number of frames in the bee colony. The base grid side length is in meters (m). S203: Adaptive Mesh Block Partitioning: First calculate the initial number of meshes. M represents the total number of grid cells, and the local terrain complexity is calculated by combining this with the terrain slope. Then dynamically adjust the grid side length. Ultimately, the entire area is divided into several non-overlapping sub-blocks that conform to the terrain, each block having an area of ​​[area missing]. ,in Let be the complexity of the i-th block, with a value between 0 and 1. The average slope of the block. , This represents the extreme slope of the region. This represents the actual grid side length. This is the terrain influence coefficient, with a value ranging from 0.1 to 0.

3. S204: Task Balance Allocation and Drone Capability Matching: First calculate the maximum operating area of ​​a single drone. Then calculate the block operation cost. Finally, with task balancing as the objective, an optimization function is established. Each block is uniquely allocated, and the workload per machine is limited. This indicates that block i is assigned to drone j, and 0 otherwise, thus achieving a fair and reasonable task allocation. For cruising speed, For extended battery life on a full charge, For safety margin factor, Considering the overall operational costs, This represents the terrain cost weight, with a value ranging from 0.2 to 0.

5.

4. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 3, characterized in that, The specific logical steps of S3 are as follows: S301: Select the core constraints of power engineering surveying, complete the quantitative processing, and construct a constraint set. The constraint items include: terrain fit, flight safety distance, energy consumption cost, surveying coverage, airspace conflict-free, and line-of-sight conditions. Transform terrain undulation, obstacle location, remaining range, and meteorological wind speed into numerical constraint indicators, eliminate fuzzy qualitative indicators, and ensure that subsequent calculations are objective and accurate. S302: Standardize all constraints, using the following formula: And calculate the information entropy of each constraint. Finally, the constraint weights are obtained. ;in Let j be the standardized value of the constraint for the i-th path point. , Let j be the constrained extreme value of the j-th term. The total number of path points. Let the entropy be the constraint information of the j-th term. Let be the weight of the j-th constraint, and n be the total number of constraints. The sum of the weights is 1, thus eliminating the bias of manual weighting. S303: Integrating various constraints and their corresponding weights, a global path-optimal objective function is established, the formula of which is: ,in This represents the total cost of the global path. Let $j$ be the single cost of the $j$ constraint on path $P$. S304: The total cost F is used as the heuristic function for the improved algorithm, and its formula is as follows: By introducing a terrain cost correction term, the heuristic function is improved to... The bee colony algorithm is used to optimize the node search direction, avoid local optima, and generate a 3D global path that fits the terrain throughout, is collision-free, and has low energy consumption. The total valuation for node n. The actual cost from the starting point to node n. Estimate the cost from node n to the destination. This is the terrain penalty coefficient, with a value ranging from 0.3 to 0.

6. For terrain complexity; S305: Verify the generated initial 3D path, check the UAV's safe distance, mapping overlap, and endurance capacity, eliminate excessive road sections, fine-tune the path inflection points, ensure that the flight path follows the terrain throughout, the surveying equipment's acquisition angle meets the standards, the point cloud coverage is uniform, there are no missed measurements or collision risks, and finally determine the exclusive global path for each UAV.

5. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 4, characterized in that, The specific logical steps of S4 are as follows: S401: Real-time status and environmental awareness: The host UAV collects flight data from each slave UAV at fixed intervals, including positioning coordinates, flight altitude, speed, remaining battery power, and communication signal strength. At the same time, it uses airborne radar and visual sensors to detect sudden obstacles in the surrounding area in real time, obtains information on the location, size, and speed of the obstacles, and collects real-time meteorological data to form a global awareness data pool. S402: Abnormal Situation Identification and Risk Assessment: Set safety thresholds, compare real-time data with the thresholds to determine if there are any abnormalities such as obstacle intrusion, insufficient power, signal obstruction, or excessive weather conditions, and calculate the risk coefficient. ,in To assess the overall risk factor, , , For each risk weight, and , E represents the real-time distance between the drone and the obstacle, and E represents the remaining battery power. Full charge R represents the signal attenuation rate, ranging from 0 to 1; the larger the value of R, the higher the risk. S403: Swarm Airspace Conflict Detection: The host calculates the distance between each UAV in real time to determine whether there is a risk of intersecting flight paths or overlapping airspace. The formula for the safe distance between UAVs is: ,in The minimum safe distance is expressed in meters (m), and k is the safety factor, which ranges from 8 to 12. The average cruising speed of the bee colony, in m / s. If the real-time interval is less than... It was determined that there was a risk of conflict. S404: Local Dynamic Path Replanning: To address anomaly risks, a fast path correction algorithm is activated, preserving valid segments of the original path and generating local detour paths in obstacle areas. The local path cost function is... ,in For the optimal cost of the local path, To correct the path length, The risk penalty coefficient ranges from 0.4 to 0.7, and R is the comprehensive risk coefficient. The route with the lowest cost is selected as the corrected route. S405: Swarm Coordinated Scheduling and Command Issuance: The host unit uniformly controls the flight speed, heading, and altitude of each drone, avoids conflicting flight paths, plans priority return routes for drones with low battery power, schedules backup drones to fill in the gaps, maintains the continuity of swarm operations, and the command response time does not exceed 1 second. S406: Corrected Path Tracking and Closed-Loop Verification. Each UAV executes a corrected route, and the host compares the actual trajectory with the planned route in real time to calculate the trajectory deviation. The formula is as follows: If the deviation exceeds the limit, the flight path will be fine-tuned again until it returns to a safe path. This is the deviation distance, in meters. , For actual coordinates, , For planning coordinates.

6. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 5, characterized in that, The specific logical steps of S5 are as follows: S501: Real-time data transmission and preprocessing: Each UAV will transmit the collected laser point cloud and image data back to the back-end processing terminal in real time. The terminal will perform noise reduction and stitching preprocessing on the raw data, remove invalid data and blurry images, classify and organize them according to the surveying blocks, and synchronously record the collection coordinates, flight attitude and collection time of the corresponding data to establish a complete data ledger. S502: Quantitative Verification of Surveying and Mapping Quality Indicators: Four core indicators—point cloud density, forward overlap, lateral overlap, and data integrity—are selected and quantitatively assessed item by item against the threshold values ​​of power engineering surveying standards. The formula for calculating point cloud density is as follows: The formula for calculating the heading overlap is: The formula for calculating lateral overlap is: ,in The value represents the local point cloud density, expressed in points per square meter (m²). N represents the total number of point clouds collected in the local area, and S represents the area of ​​the collected region, also expressed in square meters (m²). For heading overlap, This refers to the lateral overlap rate. The length of overlap in the heading of adjacent flight strips. For single-sail length, The lateral overlap width of adjacent flight strips. For single-pass width; S503: Determination and Location of Non-conforming Areas: Compare each measured indicator with the pass / fail threshold to determine non-conforming surveying areas, and set the comprehensive quality judgment formula as follows: When Q is less than the set passing score, the corresponding area is determined to be a non-compliant retest area, and the three-dimensional coordinate range of that area is locked. For the overall quality score, , , These are the indicator weights, and the sum of the weights is 1. For standard point cloud density, For standard heading overlap, Standard lateral overlap; S504: Supplementary Survey Path Planning and Fine-tuning: For non-compliant areas, generate localized, refined supplementary survey paths, following the shortest flight distance principle, and calculate the length of the supplementary survey path. ,in To supplement the measurement of the flight segment length, , , To remeasure the coordinates of the starting point, , , To supplement the measurement of the endpoint coordinates, the original flight path was slightly adjusted, the drone's terrain-following flight altitude and data collection angle were maintained, obstacles and airspace conflicts were avoided, and the supplementary measurement area was fully covered without omissions. S505: Supplementary Data Verification and Path Closure: After completing the local supplementary measurement, re-verify the quality of the supplementary measurement data until all indicators meet the standards, the comprehensive score is qualified, and it is confirmed that there are no missed measurements or distorted areas. Update the global route file, record the locations of supplementary survey points and minor adjustments to the flight segments, forming a complete surveying route and data loop to prevent rework and resurveying later.

7. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 6, characterized in that, In step S6, after the entire area is mapped and the data is verified, the host UAV issues a return command, plans a unified safe return path, guides each UAV to return in an orderly manner and land accurately, and after recovery, summarizes all mapping data, removes redundant information, and splices together to form a complete power engineering mapping result for subsequent power engineering design and construction.

8. The method for path planning of power engineering surveying tasks based on UAV swarms according to claim 1, characterized in that, In S4, during the swarm collaborative dynamic path correction and obstacle avoidance process, the host drone refreshes the entire swarm status every 0.5 seconds to identify sudden obstacles, signal blockage, insufficient power, and sudden weather changes in real time.

Images