Wind Farm Blade Lightning Protection Inspection UAV Task Scheduling and Path Optimization Management Method

Abstract

This invention relates to the field of resource allocation and discloses a method for scheduling and optimizing the path of unmanned aerial vehicles (UAVs) for lightning protection inspection of wind farm blades. This method aims to improve the efficiency, safety, and economy of lightning protection inspection of wind farm blades. It includes generating dynamic risk labels and sorting task queues based on the wind turbine spacing topology, the annual degradation rate of blade grounding resistance, and real-time lightning warning levels. It outputs inspection work orders with time windows by combining the remaining power of the UAVs and their return distance. Fractional differential equations are used to calculate the derivative of unit time cost, dynamically adjusting the deployment density of the UAV swarm. A three-dimensional composite potential field including the safety radius of lightning rods is constructed in a digital elevation model. The deviation between actual inspection costs and predicted values ​​is monitored in real time, and work order priorities are dynamically adjusted based on the enterprise's risk coefficient. This invention integrates dynamic risk assessment, cost optimization, obstacle avoidance planning, and intelligent decision-making, significantly improving the timeliness, economy, and reliability of lightning protection inspection of wind farm blades.

Description

Technical Field

[0001] This invention relates to the field of resource allocation, and in particular to a method for scheduling and path optimization management of unmanned aerial vehicles (UAVs) for lightning protection inspection of wind farm blades. Background Technology

[0002] With the transformation of the global energy structure and the large-scale development of renewable energy, wind power, as an important component of clean energy, has seen its installed capacity grow rapidly. Wind farms are typically located in areas with complex geographical environments and variable climates, such as coastal areas, plateaus, or mountainous regions. These areas are rich in wind resources but also face challenges such as frequent lightning activity and undulating terrain. As the core component for capturing wind energy, the structural integrity of wind turbine blades directly affects power generation efficiency and equipment safety. However, lightning strikes on blades can cause material aging, crack propagation, or even breakage, seriously threatening the stable operation of wind farms. Therefore, efficient lightning protection testing and maintenance of wind turbine blades has become a crucial link in ensuring the safety of wind farms.

[0003] Traditional wind turbine blade lightning protection inspections primarily rely on manual inspections or fixed monitoring equipment. Manual inspections suffer from low efficiency, high cost, and significant risk, especially in complex terrain or severe weather conditions. While fixed monitoring equipment can achieve real-time monitoring, its deployment costs are high, its coverage is limited, and it struggles to flexibly address dynamically changing risk scenarios. In recent years, drone technology, due to its high mobility, wide coverage, and ability to carry multiple sensors, has gradually been applied to wind turbine blade inspection. However, existing drone inspection solutions still have the following shortcomings:

[0004] Existing methods are mostly based on static risk assessment, which do not fully consider the impact of dynamic factors such as lightning warnings and equipment degradation on detection priority, resulting in high-risk areas not being covered in a timely manner, while low-risk areas are wasted resources.

[0005] Traditional path planning algorithms do not take into account real-time constraints such as the remaining battery power of the drone and the return distance, which can easily lead to mission interruption or failure to return. At the same time, they lack obstacle avoidance strategies for obstacles such as the safety radius of lightning rods and complex terrain, which pose safety hazards.

[0006] Existing drone swarm scheduling mostly adopts uniform distribution or experience-based deployment, without dynamically adjusting drone density according to the complexity of the detection task, resulting in unreasonable resource allocation and high detection costs;

[0007] The discrepancy between actual and predicted costs during the testing process is not effectively utilized, and there is a lack of a dynamic optimization mechanism based on real-time data, making it difficult to cope with sudden risks or efficiency improvement needs.

[0008] Therefore, we propose a task scheduling and path optimization management method for wind farm blade lightning protection inspection UAVs to solve the above problems. Summary of the Invention

[0009] This invention provides a method for scheduling and path optimization management of UAV tasks for lightning protection inspection of wind farm blades, which can improve the efficiency, safety and economy of lightning protection inspection of wind farm blades.

[0010] The first aspect of this invention provides a method for scheduling and path optimization management of UAVs for lightning protection inspection of wind farm blades. This method includes: calculating regional connectivity based on the topological relationship between wind turbine spacing, superimposing the annual degradation rate of blade grounding resistance and real-time lightning warning level to generate dynamic risk labels for wind turbines; using these dynamic risk labels, arranging them in descending order of risk value, and combining the remaining power and return distance of the UAVs to output inspection work order instructions with time windows; parsing the inspection work order instructions, calling the average inspection time of similar wind turbines over the past 90 days, calculating the derivative of unit time cost using fractional differential equations, and generating UAV cluster deployment density instructions; based on the UAV cluster deployment density instructions, constructing a repulsive potential field containing the safety radius of lightning rods in a digital elevation model, and generating an obstacle avoidance trajectory point sequence through gradient descent of the potential field; monitoring the trajectory execution progress of the obstacle avoidance trajectory point sequence, and when the actual inspection cost decreases from the predicted value to a preset value, calculating the strategy switching probability based on the enterprise risk coefficient, and if the probability is greater than a set value, outputting a work order priority update instruction.

[0011] Optionally, in the first implementation of the first aspect of the present invention, the method includes: establishing connecting edges with the wind turbine location as nodes to form a wind turbine adjacency graph; analyzing the interconnected unit clusters in the adjacency graph and marking clusters with more than 3 units as highly connected regions; extracting historical detection values ​​of blade grounding resistance, calculating the annual degradation percentage, and generating equipment degradation warning labels; receiving lightning warning level signals issued by the meteorological department and activating environmental risk indicators according to the warning level.

[0012] Optionally, in the second implementation of the first aspect of the present invention, the method includes: placing wind turbines with red risk labels at the top of the queue, followed by those with yellow labels, and then those with green labels at the bottom; arranging wind turbines of the same color level in descending order of grounding resistance degradation rate, and outputting a risk-ranked wind turbine queue; obtaining the real-time remaining battery power of the drones, multiplying it by the endurance conversion coefficient to obtain the safe flight time, and calculating the maximum single-trip operating distance in combination with the average cruising speed of the wind farm, and outputting the safe operating radius of each drone; matching available drones with the location of the wind farm charging pile as the center, establishing a dedicated mapping relationship table between drones and charging piles, and outputting a charging pile binding list; assigning the nearest drone to the risk-ranked wind turbine queue according to the safe operating radius of each drone, superimposing the charging pile binding list to ensure that the drone can return to the dedicated charging pile after the inspection is completed, and outputting a work order instruction with three elements; when multiple drones are assigned to the same wind turbine, prioritizing the turbine with higher remaining battery power, and automatically reassigning the replaced drone to the inspection task of the next wind turbine in the queue, and outputting an updated work order instruction set.

[0013] Optionally, in a third implementation of the first aspect of the present invention, the safe flight time is set to... The maximum single-trip working distance is ,but: ;in, This represents the remaining battery percentage. This is the range conversion factor; ;in, Average cruising speed, in km / h. The unit of safe flight time is minutes; the safe operating radius of each UAV is R. .

[0014] Optionally, in the fourth implementation of the first aspect of the present invention, the method includes: parsing the target wind turbine model and blade length in the inspection work order instruction; matching the wind turbine technical files in the historical database according to the model; outputting identification tags for wind turbines of the same type; calling the inspection time records of the same model of wind turbines in the past 90 days; removing outliers that have exceeded the time limit; outputting a standard time distribution table; based on the standard time distribution table, combined with real-time labor cost rate and drone depreciation parameters, using fractional differential equations to describe the memory effect of cost changes with inspection time; outputting a unit time cost change rate curve; when the unit time cost change rate exceeds the cost change rate threshold, increasing the number of drones per unit area; outputting a drone / square kilometer density control value; when the density control value fluctuates by more than 30% compared to the previous instruction, triggering the operation and maintenance expert review process; correcting the density control value according to the review opinions; and outputting a final deployment density instruction.

[0015] Optionally, in the fifth implementation of the first aspect of the present invention, the method includes: reading the density control value of UAVs per square kilometer, converting it into the minimum horizontal spacing standard between UAVs, and outputting the spacing constraint parameters; obtaining the geographical coordinates of all lightning rods in the wind farm, generating a cylindrical repulsion field centered on the coordinates, and outputting the lightning rod potential field layer.

[0016] Optionally, in the sixth implementation of the first aspect of the present invention, areas with a slope greater than 30° are marked as terrain obstacles in the digital elevation model, spacing constraint parameters are superimposed to generate a repulsion field between UAVs, and a lightning rod potential field layer is fused to output a three-dimensional composite potential field model; starting from the coordinates of the target wind turbine tower base, iterative search is performed along the negative gradient direction of the composite potential field, and the coordinates of the track points are output at 10-meter intervals to output an obstacle avoidance track point sequence; track points less than 10 meters away from the lightning rod are automatically marked, high-risk points are pushed to the operation and maintenance console for manual confirmation, and the confirmation results are integrated to generate a final review track sequence.

[0017] Optionally, in the seventh implementation of the first aspect of the present invention, the method includes: receiving the coordinates of the flight path points transmitted back by the UAV in real time, comparing them with the planned flight path point sequence to calculate the completion percentage, and outputting a flight path execution progress report; synchronously acquiring the power consumption of the UAV and the manual monitoring time data, superimposing the equipment depreciation rate to calculate the real-time detection cost, and outputting an actual cost flow record; calling the predicted cost benchmark value, activating a judgment flag when the actual cost flow record decreases by more than 15%, and outputting a cost decrease trigger signal; reading the risk tolerance coefficient preset by the enterprise management system, mapping the strategy switching probability according to the coefficient value, and outputting a strategy switching probability value; and outputting a work order priority update instruction according to the strategy switching probability value.

[0018] Optionally, in the eighth implementation of the first aspect of the present invention, it further includes real-time monitoring of the charging pile status: monitoring the charging pile temperature, output current and fault signals, and outputting a charging pile health status table; receiving the sandstorm level issued by the meteorological station and outputting a trajectory downgrade command; analyzing high-definition images of the blade surface taken by the UAV, and outputting a blade damage alarm packet when lightning damage or cracks are identified.

[0019] The mechanism of this invention is as follows: by fusing electrical topology connectivity, grounding resistance degradation rate and physical parameters of lightning warning, a dynamic risk label-driven detection priority is constructed, and a closed-loop decision system is realized by combining a fractional historical cost model and a potential field obstacle avoidance mechanism.

[0020] Beneficial effects: By integrating the topological relationship of wind turbine spacing, the annual degradation rate of blade grounding resistance, and the real-time lightning warning level, a dynamic risk assessment model is constructed to achieve real-time updates and accurate quantification of risk levels. The task queue is arranged in descending order based on the risk value. Combined with the constraints of the remaining power of the drone and the return distance, a detection work order instruction with a time window is generated to ensure that high-risk areas are detected first, while avoiding the risk of mission interruption or crash due to insufficient power of the drone.

[0021] Fractional differential equations are introduced to describe the memory effect of detection cost changes over time. Combined with historical detection time data of similar wind turbines over the past 90 days, the derivative of unit time cost is calculated to dynamically adjust the deployment density of drones. When the cost change rate exceeds the threshold, the density control mechanism is automatically triggered. The rationality of the decision is ensured by the review of operation and maintenance experts to avoid resource waste or insufficient coverage.

[0022] A three-dimensional composite potential field model is constructed in the digital elevation model, which integrates the lightning rod safety radius repulsion potential field, the UAV repulsion potential field and the terrain obstacle potential field. The obstacle avoidance flight path sequence is generated by the potential field gradient descent algorithm, and high-risk points are automatically marked for manual confirmation to ensure path safety and compliance.

[0023] Real-time monitoring of flight path execution progress and actual detection costs. When the cost decreases by more than 15% compared to the predicted value, the probability of strategy switching is calculated based on the enterprise risk coefficient. If the probability is greater than 0.6, the work order priority is automatically updated. External data such as charging pile health status monitoring and sandstorm level warning are integrated to dynamically adjust the flight path or suspend the task to avoid damage from extreme environments.

[0024] The system integrates five modules: dynamic risk assessment, task scheduling, path planning, cost optimization, and strategy adjustment, forming a closed-loop management system. The modules share data and make collaborative decisions. It introduces intelligent blade damage identification and automatically generates damage alarm packages through high-definition image analysis from UAVs, achieving integrated "detection-analysis-early warning". Attached Figure Description

[0025] Figure 1 This is a schematic diagram of an embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management method in this invention;

[0026] Figure 2 This is a schematic diagram of another embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management method in this invention;

[0027] Figure 3 This is a schematic diagram of an embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management device in this invention.

[0028] Figure 4This is a schematic diagram of an embodiment of the wind farm blade lightning protection detection UAV task scheduling and path optimization management device in this invention. Detailed Implementation

[0029] This invention provides a method for scheduling and path optimization management of unmanned aerial vehicles (UAVs) for lightning protection inspection of wind farm blades, which improves the efficiency, safety, and economy of lightning protection inspection of wind farm blades. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0030] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 One embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management method in this invention includes:

[0031] 101. Wind Turbine Cluster Risk Classification: Based on the topological relationship between wind turbine spacing, the regional connectivity is calculated, and the annual degradation rate (%) of blade grounding resistance and real-time lightning warning level are superimposed to generate dynamic risk labels for wind turbines (Product 1).

[0032] It is understood that the executing entity of this invention can be a task scheduling and path optimization management device for wind farm blade lightning protection inspection drones, or it can be a terminal or a server; the specific implementation is not limited here. This embodiment of the invention will be described using a server as an example.

[0033] It should be noted that the dynamic risk labels for 20 wind turbines in a certain wind farm were generated, and the basic data preparation and wind turbine topology were carried out. The wind farm layout is a 4×5 grid, with each turbine spaced approximately 150 meters apart.

[0034] Regional connectivity calculation: Taking each wind turbine as the center, count the number of adjacent wind turbines within a 200-meter radius. Example of adjacent wind turbine counts for some turbines:

[0035] Wind Turbine ID F01 F02 F03 F04 F05 Adjacent number 3 4 5 4 3

[0036] Connectivity weight: Weight = 1.2 when there are ≥ 5 adjacent connections (high connection); weight = 1.0 when there are 3-4 adjacent connections (medium connection); weight = 0.8 when there are ≤ 2 adjacent connections (low connection).

[0037] Annual degradation rate of grounding resistance (%): Measured annual degradation rate (reflecting the aging rate of the lightning protection system):

[0038] F01:8.2% (high) F02: 3.5% (Low) F03:6.7% (medium)

[0039] Real-time lightning warning levels: meteorological system input data (level 1-3):

[0040] Current level Area F01 = 3 (High Risk) Area F02 = 1 (Low Risk) Area F03 = 2 (Medium Risk)

[0041] Dynamic risk label calculation, risk value formula:

[0042] Risk value = connectivity weight (Grounding resistance degradation rate) 0.6+ lightning level 0.4);

[0043] Taking the F01 wind turbine as an example: Connectivity weight = 1.0 (4 adjacent units, connected in the middle); Grounding degradation rate = 8.2% → Normalized to 8.2 (percentage value); Lightning level = 3 → Normalized to 3;

[0044] ;

[0045] Risk classification rules: High risk (red): ≥5.0; Medium risk (yellow): 3.0-4.9; Low risk (green): <3.0;

[0046] Output Product 1 (partial wind turbine labels):

[0047] Wind Turbine ID Risk Value Dynamic risk labels F01 6.12 High risk (red) F02 2.18 Low risk (green) F03 4.35 Medium risk (yellow)

[0048] 102. Time-constrained work order generation: Using the wind turbine dynamic risk label of product 1, sorted in descending order of risk value, and combined with the remaining power of the drone and the return distance, output the detection work order instruction with time window (product 2).

[0049] It should be noted that the following is a specific implementation example of the "Time-Constrained Work Order Generation" step (step 102), based on the dynamic risk tags of 5 wind turbines and the status data of 3 drones in a wind farm:

[0050] Basic data preparation, wind turbine dynamic risk labeling (product 1).

[0051] Data example:

[0052] Wind Turbine ID Risk Value Risk Level F03 8.6 High risk (red) F01 7.2 High risk (red) F05 5.1 Medium risk (yellow) F02 3.8 Medium risk (yellow) F04 2.3 Low risk (green)

[0053] Drone status: Drone cluster: 3 drones (UAV1, UAV2, UAV3); Remaining battery power: UAV1 (85%), UAV2 (70%), UAV3 (90%);

[0054] Return distance (distance from each wind turbine to the base station / km):

[0055] Wind Turbine ID UAV1 distance UAV2 distance UAV3 distance F03 1.2 0.8 2.5 F01 0.5 1.5 1.0

[0056] Work order generation logic, risk priority sorting: wind turbine inspection order is arranged from high to low risk value: F03→F01→F05→F02→F04. High-risk wind turbines are inspected first to avoid lightning strike accidents (F03 risk value 8.6, requires immediate response).

[0057] Battery level and return distance constraints: Time window calculation: Single machine detection time: High-risk wind turbines require an average of 25 minutes (including flight, detection, and obstacle avoidance), while medium- and low-risk wind turbines require 15 minutes. UAV1 working time = minute.

[0058] Return flight time reserved: . minute.

[0059] Task allocation rules: Dynamic binding: When assigning a wind turbine to each drone, the following condition must be met: Detection time + return time ≤ maximum operable time. Nearby dispatch: Prioritize drones with shorter return distances (e.g., assigning UAV2 to F03 because its return distance is only 0.8km).

[0060] Output Product 2: Inspection work order instruction with time window:

[0061] Work order number Wind Turbine ID Risk Level Execute drones Start Time Window End Time Window Priority 1 F03 High risk UAV2 T+0min T+25min urgent 2 F01 High risk UAV1 T+5min T+30min urgent 3 F05 Medium risk UAV3 T+10min T+25min high 4 F02 Medium risk UAV1 T+35min T+50min middle 5 F04 Low risk UAV3 T+30min T+45min Low

[0062] 103. Historical Dependency-Based Cost Control: Analyze the inspection work order instructions of Product 2, call the average inspection time of the same type of wind turbine in the past 90 days, use fractional differential equations to calculate the derivative of unit time cost, and generate the drone cluster deployment density instruction (Product 3).

[0063] It should be noted that this data is based on the inspection work orders and historical data of 5 wind turbines in a certain wind farm:

[0064] Basic data preparation, inspection work order instructions (product 2): including the sequence of fans to be inspected: F03 (high risk), F01 (high risk), F05 (medium risk), F02 (medium risk), F04 (low risk).

[0065] Planned inspection time windows for each wind turbine (unit: minutes):

[0066] Wind Turbine ID Start time End time F03 T+0 T+25 F01 T+5 T+30

[0067] Historical testing time: The average testing time for the same type of wind turbine (2.0MW model) over the past 90 days is as follows:

[0068] Wind Turbine ID Historical average time (minutes) F03 28.5 F01 26.2 F05 18.7 F02 17.9 F04 15.3

[0069] The derivative of unit time cost is calculated, and the cost function is defined as follows: Based on the historical average time consumption, and combined with the real-time work order time window deviation (the difference between the actual time consumption and the planned time consumption), a time cost function is constructed. ,in =0.8 (real-time deviation weight) =0.2 (historical benchmark weight).

[0070] Fractional differential calculation: Calculating the derivative of unit time cost using fractional differential equations (order 0.7). This reflects the sensitivity of costs to changes over time. Take F03 as an example:

[0071] The planned time is 25 minutes, with a historical average of 28.5 minutes → real-time deviation of 3.5 minutes; fractional derivative (A positive value indicates a significant upward trend in costs).

[0072] Drone swarm deployment density command generation, derivative and deployment density mapping rules:

[0073] High-density deployment ( ): The spacing between drones needs to be shortened to ≤15m to improve detection efficiency and curb cost increases.

[0074] Medium density deployment ( ): Maintain a standard spacing of 20m.

[0075] Low-density deployment ( ): Relax the spacing to ≥25m and optimize resource allocation.

[0076] Output Product 3 (Deployment Density Instruction):

[0077] Wind Turbine ID derivative of unit time cost Deployment density instructions F03 +1.24 High density (≤15m) F01 +0.92 Medium density (20m) F05 -0.38 Medium density (20m) F02 -0.61 Low density (≥25m) F04 -0.43 Medium density (20m)

[0078] 104. Safe potential field path planning: Based on the deployment density instructions of product 3, construct a repulsive potential field containing the safety radius of the lightning rod (≥5m) in the digital elevation model (DEM), and generate an obstacle avoidance path point sequence (product 4) through potential field gradient descent.

[0079] It should be noted that this is based on the deployment density instructions and geographical data of 5 wind turbines in a certain wind farm:

[0080] Basic data preparation, input data: Deployment density instruction (product 3): output from step 103 (F03 wind turbines require high-density deployment ≤15m, F01 requires medium density 20m); Digital elevation model (DEM): includes wind turbine coordinates, altitude and lightning rod location (F03 lightning rod coordinates (120.5, 38.2), safety radius ≥5m); Obstacle data: wind turbine tower, surrounding high-voltage lines (coordinates already marked);

[0081] Potential field construction and overlay, terrain potential field: generating elevation potential field based on DEM, steep slope area (slope > 15) Set a high potential energy value (potential energy in the abrupt elevation change zone = 8).

[0082] Lightning rod repulsive potential field: A spherical repulsive field is constructed with the lightning rod as the center. Within a safe radius of 5m, the potential energy increases exponentially (potential energy = 12 at a distance of 4m from the lightning rod; potential energy ≈ 0 at a distance of 6m).

[0083] Target gravitational potential field: Set a low potential energy (potential energy = -10) at the target point (leading edge of the blade) to guide the drone closer.

[0084] Density-constrained potential field: A repulsive potential field between UAVs is added in the high-density region (F03), and the potential energy increases sharply when the spacing is <15m (potential energy = 5 when the spacing is 10m).

[0085] Gradient descent generates waypoints. Taking the leading edge detection of F03 blade as an example: Starting point: UAV hovering point (coordinates (120.48, 38.18), altitude 90m);

[0086] Potential field superposition calculation: Lightning rod (distance 3m) → repulsive potential energy = 15; Target point (distance 2m) → gravitational potential energy = -8; Terrain slope 8 → Potential energy = 3;

[0087] The direction of the resultant force is northeast-east (the direction in which the potential energy gradient decreases the fastest).

[0088] Waypoint sequence generation:

[0089] Serial Number longitude latitude Altitude (m) Obstacle Avoidance Instructions 1 120.482 38.183 92 Avoid the west side of the lightning rod 2 120.487 38.186 95 Climbing along the gentle slope 3 120.491 38.190 98 Cut into the leading edge of the blade

[0090] The spacing between track points is dynamically adjusted according to density commands (F03 point spacing ≤ 15m, F04 point spacing = 20m) to ensure cluster collision avoidance.

[0091] Output Product 4: Obstacle Avoidance Waypoint Sequence

[0092] Wind Turbine ID Number of waypoints Total path length (m) Minimum lightning rod distance F03 18 240 5.2 F01 12 180 6.8

[0093] 105. Strategy Shift Audit Decision: Monitor the execution progress of product 4. When the actual detection cost decreases by more than 15% compared to the predicted value, calculate the strategy switching probability based on the enterprise risk coefficient. If the probability is greater than 0.6, output the work order priority update instruction (product 5).

[0094] It should be noted that this data is based on the flight path execution data of five wind turbines in a certain wind farm and the company's risk parameters:

[0095] Basic data preparation, track execution progress (product 4): Actual detection time (minutes):

[0096] Wind Turbine ID Predicted time Actual time consumed Cost reduction F03 25 22 15.4%↓ F01 25 20 22.1%↓ F05 15 14 8.3%↓

[0097] Cost reduction formula: ;

[0098] Enterprise risk coefficient: set according to the wind farm safety level: high wind speed area: risk coefficient = 0.7 (high risk, low tolerance); low wind speed area: risk coefficient = 0.4 (low risk, high tolerance); this example uses 0.7 (currently it is the high wind speed season);

[0099] Strategy switching decision-making process, Step 1: Monitor actual cost reduction. Trigger condition: Strategy switching calculation is initiated only when the cost reduction is >15%. F03's cost reduction is 15.4% → meets the condition and enters the calculation process. Although F01's cost reduction is 22.1%, it is higher, but only the first wind turbine that meets the condition needs to be processed.

[0100] Step 2: Calculate the probability of strategy switching, formula logic: F03 calculation: Decision rule: Output update instruction when probability > 0.6.

[0101] Step 3: Generate a work order priority update instruction (product 5). Update logic: Due to the significant improvement in F03 detection efficiency (high cost reduction), it is speculated that there may be similar optimization potential for wind turbines (F05) in the same area. Increase the priority of F05 from "High" to "Urgent" and insert it into the F01 post-detection sequence.

[0102] Wind Turbine ID Original priority New priority Detection order adjustment F05 high urgent F01→F05→F03

[0103] In this embodiment of the invention, a dynamic risk label for each wind turbine is generated by comprehensively considering the topological relationship between turbine spacing, the annual degradation rate of blade grounding resistance, and the real-time lightning warning level. This multi-factor integrated risk assessment method overcomes the limitations of single-factor assessment, more accurately reflects the actual risk status of each wind turbine, provides a scientific basis for subsequent task scheduling, and helps to prioritize high-risk wind turbines, reducing the possibility of lightning strike accidents. Based on the dynamic risk label of the wind turbine, combined with the remaining power of the drone and the return distance, a detection work order instruction with a time window is generated. On the basis of satisfying the risk priority ranking, the constraints of power and return distance are fully considered to achieve dynamic binding and nearby scheduling. This innovative task allocation rule ensures that high-risk wind turbines are prioritized for detection while making reasonable use of drone resources, improving detection efficiency, and avoiding delays in detection tasks due to insufficient power or excessive return distance. The average detection time of similar wind turbines over the past 90 days is used to calculate the derivative of unit time cost using fractional differential equations, thereby generating a drone swarm deployment density instruction. This innovative method combines historical data with real-time work orders, accurately calculating the cost derivative through fractional differential equations. This allows for timely reflection of cost sensitivity over time, enabling reasonable adjustment of UAV swarm deployment density. While ensuring detection quality, it effectively controls costs and optimizes resource allocation. A repulsive potential field, including the safety radius of lightning rods, is constructed within a digital elevation model. Obstacle avoidance path point sequences are generated through potential field gradient descent. This method comprehensively considers various factors such as terrain, lightning rods, target gravity, and density constraints among UAVs, constructing a complex and comprehensive potential field model to ensure UAVs can effectively avoid obstacles during flight, guaranteeing flight safety. Simultaneously, the path point spacing is dynamically adjusted based on the deployment density commands of different wind turbines, further improving the safety and efficiency of swarm flight. Monitoring path execution progress, when the actual detection cost decreases by more than 15% compared to the predicted value, the strategy switching probability is calculated based on the enterprise risk coefficient. If the probability is greater than 0.6, a work order priority update command is output. This step, through real-time monitoring of cost changes and combined with enterprise risk factors, dynamically adjusts work order priorities, achieving flexible switching of detection strategies. This innovative decision-making mechanism can seize opportunities to reduce costs in a timely manner, further optimize the testing process, improve overall testing efficiency, and adapt to the actual operational needs of different wind farms.

[0104] Please see Figure 2 Another embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management method in this invention includes:

[0105] 201. Wind Turbine Cluster Risk Classification: Based on the topological relationship between wind turbine spacing, the regional connectivity is calculated, and the annual degradation rate (%) of blade grounding resistance and real-time lightning warning level are superimposed to generate dynamic risk labels for wind turbines (Product 1).

[0106] Specifically, establish a wind turbine adjacency network: with the wind turbine location as a node, establish connecting edges within a 500-meter interval to form a wind turbine adjacency relationship graph; output the adjacency relationship graph.

[0107] Identify highly connected regions: Analyze the interconnected clusters of units in the adjacency graph, and mark clusters with more than 3 units as highly connected regions; output a highly connected region labeling table.

[0108] Quantify grounding resistance degradation: Extract historical detection values ​​of blade grounding resistance and calculate the annual degradation percentage; generate equipment degradation warning labels when the annual degradation rate exceeds 5%; output degradation warning labels.

[0109] Integrate lightning warning signals: Receive lightning warning level signals issued by meteorological departments; activate environmental risk indicators when the warning level reaches orange (level 4) or above; output environmental risk indicators.

[0110] Generate dynamic risk labels: Generate red risk labels for wind turbines that simultaneously meet the following conditions: located in a highly connected area, carrying an equipment degradation warning label, and currently having an active environmental risk indicator; generate yellow risk labels for wind turbines that meet any two conditions; generate green risk labels for the remaining wind turbines; output three-color risk labels.

[0111] It should be noted that the following example uses a wind farm (containing 10 2.0MW units) to illustrate the specific implementation and data flow of the wind turbine cluster risk classification steps:

[0112] Establish a wind turbine adjacency network, inputting the following data: wind turbine coordinates (unit: meters):

[0113] Fan 1: (0,0);

[0114] Fan 2: (300, 100);

[0115] Fan 3: (700, 400);

[0116] ...(Coordinates of wind turbine 4-10 omitted);

[0117] Connection rules: Establish connection edges with a radius of 500 meters. Example connection relationship: Fan 1 is connected to Fan 2 (spacing 352m), Fan 3 is connected to Fan 4 (spacing 420m), and Fan 5 is isolated.

[0118] Output adjacency graph (partial): Node connections: 1-2, 2-3, 3-4, 6-7, 7-8, 8-9; Isolated nodes: 5, 10;

[0119] Identify highly connected regions and perform cluster analysis: Cluster 1: Fans 1-4 (4 units); Cluster 2: Fans 6-9 (4 units); Fans 5 and 10 are single units. Output a table of highly connected regions:

[0120] Cluster ID Includes wind turbine High connectivity (>3 units) C1 1,2,3,4 yes C2 6,7,8,9 yes

[0121] Quantify grounding resistance degradation. Input data: Grounding resistance values ​​for 2023-2024 (unit: ... Fan 1: ( Fan 3: (Degradation rate 1.4%); Warning rule: Generate a label if degradation rate > 5%. Output degradation warning labels:

[0122] Wind Turbine ID Deterioration rate Warning label 1 5.9% yes 3 1.4% no

[0123] Integrating lightning warning signals, input: Meteorological observatory issues orange lightning warning (level IV). Rule: Orange and above warnings activate environmental risk indicators. Output: Environmental risk indicator:

[0124] Warning Level Activate or not? orange color yes

[0125] Generate dynamic risk labels and apply rules (taking wind turbine 1 and wind turbine 3 as examples):

[0126] Wind Turbine ID Highly connected regions Degradation warning Environmental risks Number of conditions satisfied Risk Label 1 Yes (C1) yes yes 3 items red 3 Yes (C1) no yes 2 items yellow 5 no no yes 1 item green

[0127] Output three-color risk labels: Red: Fans 1, 4; Yellow: Fans 2, 3, 6, 7; Green: Fans 5, 8, 9, 10;

[0128] In this embodiment, wind turbine 1 is marked as the highest risk (red) and needs to be prioritized for detection because it is simultaneously located in a highly connected cluster, its grounding resistance deteriorates beyond the threshold, and it is under an orange lightning warning. Wind turbine 5 is the lowest risk (green) because it operates in isolation and has no deterioration. Dynamic tag generation enables a three-dimensional coupled assessment of equipment status, cluster topology, and meteorological threats.

[0129] 202. Time-constrained work order generation: Using the wind turbine dynamic risk label of product 1, sorted in descending order of risk value, and combined with the remaining power of the drone and the return distance, output the detection work order instruction with time window (product 2).

[0130] Specifically, a risk priority sequence is generated: wind turbines with red risk labels are placed at the top of the queue, followed by those with yellow labels, and then those with green labels; wind turbines of the same color level are arranged in descending order of their grounding resistance degradation rate; and a risk-ranked wind turbine queue is output.

[0131] Calculate the effective operating radius of the drone: obtain the drone's real-time remaining battery power, multiply it by the endurance conversion factor of 0.8 to obtain the safe flight time; combine the average cruising speed of the wind farm to calculate the maximum single-trip operating distance; output the safe operating radius of each drone.

[0132] Bind to nearby charging stations: Using the location of the wind farm charging station as the center, match available drones within 20 kilometers; establish a dedicated mapping table between drones and charging stations; output a list of bound charging stations.

[0133] Generate time-constrained work orders: sort the wind turbine queue by risk, and assign the nearest drone to each drone according to its safe operating radius; overlay the charging pile binding list to ensure that the drone can return to its dedicated charging pile after the inspection is completed; output work order instructions with three elements: <target wind turbine ID, executing drone ID, latest start time>.

[0134] Real-time conflict resolution mechanism: When multiple drones are assigned to the same wind turbine, the turbine with the highest remaining power is selected first; the replaced drone is automatically reassigned to the next wind turbine in the queue for detection tasks; and an updated work order instruction set is output.

[0135] It should be noted that the following is a specific example of the "time-constrained work order generation" steps in a wind farm (including 10 2.0MW units), and the data is based on the previous risk classification results (product 1):

[0136] Risk priority sequence generation, input product 1 data: Red label fans: ID01 (deterioration rate 7.2%), ID04 (deterioration rate 6.8%), ID02 (deterioration rate 5.9%); Yellow label fans: ID03 (deterioration rate 4.5%), ID06 (deterioration rate 3.9%), ID07 (deterioration rate 3.2%), ID09 (deterioration rate 2.8%); Green label fans: ID05, ID08, ID10 (no deterioration warning);

[0137] Sorting rules:

[0138] ;

[0139] ;

[0140] ;

[0141] Output queue: [ID01,ID04,ID02,ID03,ID06,ID07,ID09,ID05,ID08,ID10];

[0142] Set safe flight time as The maximum single-trip working distance is ,but: ;in, This represents the remaining battery percentage. This is the range conversion factor; ;in, Average cruising speed, in km / h. The unit of safe flight time is minutes; the safe operating radius of each UAV is R. ;

[0143] Effective operating radius calculation for UAVs, UAV parameters (3 UAVs): Remaining battery power: UAV1 (85%), UAV2 (70%), UAV3 (90%); Endurance conversion factor: 0.8 → Safe flight time: UAV1 (34 minutes), UAV2 (28 minutes), UAV3 (36 minutes); Average cruising speed: 10m / s (36km / h);

[0144] One-way operating distance: UAV1: km (safe radius 10.2km); UAV2: km (safe radius 8.4km); UAV3: km (safe radius 10.8km);

[0145] Charging pile binding relationship, charging pile location: pile A (coordinates X1, Y1), pile B (coordinates X2, Y2).

[0146] Binding rules: Matching drones within 20km: UAV1 distance to beacon A: 15km → Binding beacon A; UAV2 distance to beacon B: 18km → Binding beacon B; UAV3 distance to beacon A: 12km → Binding beacon A;

[0147] Output binding list:

[0148] Drone ID Connect to charging station UAV1 Pile A UAV2 Pile B UAV3 Pile A

[0149] Time-constrained work order generation and allocation logic:

[0150] High-risk wind turbines will be assigned based on proximity: ID01 (coordinate P1) is closest to UAV1 (1.2km) → assigned to UAV1; ID04 (coordinate P4) is closest to UAV3 (0.8km) → assigned to UAV3; ID02 (coordinate P2) is closest to UAV1 (1.5km), but UAV1 needs to return to pile A ( → Reassign UAV3 (distance 1.8km) );

[0151] Conflict resolution: ID02 is simultaneously contested by UAV1 / UAV3 → UAV3 with higher remaining battery power (90%>85%) takes priority → UAV3 executes; UAV1 is reassigned to the next task ID03.

[0152] Output work order instructions:

[0153] Target wind turbine Execute drones Latest start time (assuming current time T0) ID01 UAV1 T0+0min (Execute immediately) ID04 UAV3 T0+0min ID02 UAV3 T0+15min (UAV3 needs to complete ID04 first) ID03 UAV1 T0+20min (UAV1 recharges after returning to base)

[0154] Work orders ensure that the total mission distance for each drone does not exceed the safe operating radius (UAV3 executes ID04+ID02). It is also equipped with a dedicated charging station to ensure a safe return trip.

[0155] 203. Historical Dependency-Based Cost Control: Analyze the inspection work order instructions of Product 2, call the average inspection time of similar wind turbines in the past 90 days, use fractional differential equations to calculate the derivative of unit time cost, and generate the drone cluster deployment density instruction (Product 3).

[0156] Specifically, extract the fan type characteristics from the work order: parse the target fan model and blade length in the detection work order instruction; match the fan technical files in the historical database according to the model; and output identification tags for fans of the same type.

[0157] Construct a historical time consumption distribution matrix: retrieve the detection time records of the same model of wind turbine in the past 90 days; remove outliers that exceed the time limit (data points > twice the average value); output a standard time consumption distribution table.

[0158] Calculate fractional-order cost dynamics: Based on the standard time consumption distribution table, combined with real-time labor cost rate and drone depreciation parameters; use fractional-order differential equations to describe the memory effect of cost changes with detection time; output the cost change rate curve per unit time.

[0159] Generate deployment density instructions: Set the cost change rate threshold to 5% increase per hour; when the cost change rate curve per unit time exceeds the threshold, increase the number of drones per unit area; output the drone / square kilometer density control value.

[0160] Manual review and correction mechanism: When the density control value fluctuates by more than 30% compared to the previous instruction, the operation and maintenance expert review process is triggered; the density control value is corrected according to the review opinions; and the final deployment density instruction is output.

[0161] It should be noted that the following example uses a wind farm (containing 10 2.0MW units):

[0162] Extract the fan type characteristics of the work order and input product 2 data: The detection work order instruction includes the target fan ID and model: ID01~ID08: Model A (blade length 60m); ID09~ID10: Model B (blade length 45m).

[0163] Matching historical technical files: Model A corresponds to technical file number TECH-A (down conductor type: copper cable; number of lightning arresters: 6); Model B corresponds to TECH-B (down conductor type: aluminum strip; number of lightning arresters: 4).

[0164] Output labels of the same type: Model A labels: ID01~ID08; Model B labels: ID09~ID10;

[0165] Construct a historical time distribution matrix and retrieve historical data (detection time over the past 90 days):

[0166] Model A fan: Average time 65 minutes (Example data points: 55 min, 70 min, 60 min, 120 min*, 62 min; *Note: (It is retained because it does not exceed the standard).

[0167] Model B fan: Average time 45 minutes (data points: 40min, 48min, 50min, 100min*; excluded) ).

[0168] Output standard time distribution table:

[0169] Fan Model Valid data points (minutes) Average time A 55,60,62,70 62±5min B 40,48,50 46±4min

[0170] Calculate the fractional-order cost dynamics, with the following cost parameters: labor cost rate: 200 yuan / hour; drone depreciation: 50 yuan / hour.

[0171] Cost change rate calculation: Based on the historical time distribution of Model A, for every 10-minute increase in testing time, the unit time cost increases by 3.5 yuan / minute (due to the combined effect of labor and depreciation); when the testing exceeds 70 minutes, the cost increase exceeds the threshold of 5% / hour. Yuan, % / h).

[0172] Generate deployment density instructions and threshold determination: When the detection time of Model A fan is >70 minutes, the cost change rate exceeds 5% / h → trigger the density increase mechanism;

[0173] Density control: The original basic density of 5 aircraft / square kilometer was increased to 7 aircraft / square kilometer (an increase of 40%), and the inspection range of a single aircraft was shortened to control the time consumption.

[0174] Manual review and correction mechanism, fluctuation judgment: if the density control value fluctuates by 40% > 30% compared to the previous value (5 aircraft / square kilometer) → expert review is triggered;

[0175] Correction process: Based on the real-time wind speed (15m / s) and the complexity of the blade structure of Model A, the operation and maintenance experts approved the density adjustment to 6 units / square kilometer (final approval order).

[0176] Output final deployment density instructions: Model A wind turbine area: 6 units / km²; Model B wind turbine area: maintain 5 units / km².

[0177] 204. Safe potential field path planning: Based on the deployment density instructions of product 3, construct a repulsive potential field containing the safety radius of the lightning rod (≥5m) in the digital elevation model (DEM), and generate an obstacle avoidance path point sequence (product 4) through potential field gradient descent.

[0178] Specifically, the deployment density constraint is analyzed by: reading the drone / square kilometer density control value; converting it into the minimum horizontal spacing standard between drones; and outputting the spacing constraint parameters.

[0179] Constructing the lightning rod repulsion potential field: Obtain the geographic coordinates of all lightning rods in the wind farm; generate a cylindrical repulsion field with a radius of 5 meters centered on the coordinates; output the lightning rod potential field layer.

[0180] Integrating terrain and density potential field: Mark areas with slope > 30° as terrain obstacles in the digital elevation model (DEM); superimpose spacing constraint parameters to generate repulsion field between UAVs; merge lightning rod potential field layers; output a three-dimensional composite potential field model.

[0181] Generate gradient descent waypoints: Start from the target wind turbine tower base coordinates; iteratively search along the negative gradient direction of the composite potential field; output waypoint coordinates (including altitude) every 10 meters; output obstacle avoidance waypoint sequence.

[0182] Manual verification of high-risk flight path segments: Automatically mark flight path points less than 10 meters from the lightning rod; push high-risk points to the operation and maintenance console for manual confirmation; integrate the confirmation results to generate the final review flight path sequence.

[0183] It should be noted that the following example uses a wind farm (containing 10 2.0MW units):

[0184] Input data and parameter initialization, deployment density command (product 3): Model A area (ID01-ID08): 6 aircraft / square kilometer; Model B area (ID09-ID10): 5 aircraft / square kilometer;

[0185] Lightning rod coordinates: Lightning rod P1 (near ID01): (120.5, 38.2, 85.0); Lightning rod P2 (near ID04): (122.1, 39.8, 86.5);

[0186] Digital Elevation Model (DEM): Areas with slope > 30°: Coordinates (121.3, 40.1) are marked as terrain obstacles (elevation change of 20m).

[0187] Potential field construction process, analysis of deployment density constraints: Model A area density 6 aircraft / km 2 →Minimum drone spacing 40 meters; Model B area density 5 aircraft / km 2 →Minimum Spacing 45 meters

[0188] Output Spacing Constraint Parameter Table:

[0189] area Minimum spacing Model A 40 meters Model B 45 meters

[0190] Construct the lightning rod repulsion potential field: Generate a cylindrical repulsion field with a radius of 5 meters centered on P1. The potential energy intensity decreases with distance (potential energy = 0.8 when the distance to P1 is 6 meters; potential energy = 2.5 when the distance to P1 is 3 meters); Output the potential field layer: Mark the repulsion range and intensity gradient of P1 and P2.

[0191] Fusion of three-dimensional composite potential fields: Ground potential field: Mark (121.3, 40.1) in DEM as an obstacle (slope 35°), potential energy value = 3.0; UAV repulsion field: Generate a repulsion grid with a spacing of 40 meters in the model A area (potential energy value = 1.5); Lightning rod potential field superposition: P1 repulsion field covers the area around the ID01 tower base;

[0192] Output of a three-dimensional composite potential field model (partial):

[0193] Coordinates (x, y, z) Potential energy type Potential energy value (120.5，38.2，85) Lightning rod repulsion 5.0 (121.3，40.1，72) Terrain obstacles 3.0 (120.8，38.5，80) Inter-drone exclusion 1.5

[0194] Track generation and manual verification, gradient descent track generation (starting from ID01 base): starting point: (120.0, 38.0, 80.0); iterative search: along the negative gradient direction of the potential field (avoiding P1 and terrain obstacles);

[0195] Output waypoint sequence (partial):

[0196] Point 1: (120.0, 38.0, 80.0);

[0197] Point 2: (120.3, 38.3, 82.0) → Detour around the west side of P1;

[0198] Point 3: (120.6, 38.6, 83.5) → Avoid areas with abrupt changes in terrain;

[0199] ... (output every 10 meters);

[0200] High-risk flight path segment marking and manual verification: Automatic marking: Point (120.4, 38.3, 81.5) is only 8.2 meters away from P1 (<10-meter threshold) → push to the control console;

[0201] Operations confirmed: passage is permitted (because P1's altitude is below the waypoint) → the waypoint is retained;

[0202] Final review track sequence, final review track output (ID01 part): Track point sequence:

[0203] [(120.0,38.0,80.0),(120.3,38.3,82.0),(120.4,38.3,81.5)*,(120.6,38.6,83.5)...]

[0204] *Note: Points marked with * are high-risk points confirmed manually; please maintain the original route.

[0205] 205. Strategy Shift Audit Decision: Monitor the execution progress of Product 4. When the actual detection cost decreases by more than 15% compared to the predicted value, calculate the strategy switching probability based on the enterprise risk coefficient. If the probability is greater than 0.6, output the work order priority update instruction (Product 5).

[0206] Specifically, the tracking progress monitoring includes: receiving the coordinates of the track points already flown by the UAV in real time; comparing the progress with the planned track point sequence to calculate the percentage of completion; and outputting a track execution progress report.

[0207] Dynamic data collection of actual costs: synchronously acquire data on drone power consumption and manual monitoring hours; overlay equipment depreciation rates to calculate real-time monitoring costs; and output actual cost transaction records.

[0208] Cost deviation threshold determination: Call the predicted cost baseline value; when the actual cost record decreases by more than 15% compared with the predicted value, activate the determination flag; output the cost decrease trigger signal.

[0209] Risk Quantification Strategy Decision: Read the risk tolerance coefficient (0-1.0) preset in the enterprise management system; map the strategy switching probability according to the coefficient value: coefficient ≥ 0.8 → switching probability = 0.9; coefficient 0.5-0.8 → switching probability = 0.7; coefficient < 0.5 → switching probability = 0.4; output the strategy switching probability value.

[0210] Work order priority update execution: When the strategy switching probability value is >0.6: reduce the detection weight of high-risk wind turbines by 20%; increase the detection frequency of medium-risk wind turbines by 30%; output work order priority update instruction.

[0211] It should be noted that the following example uses a wind farm (containing 10 2.0MW units):

[0212] Input data and monitoring initialization, trajectory plan data (product 4):

[0213] Target wind turbines: ID01 (red risk), ID04 (red risk), ID03 (yellow risk);

[0214] Planned waypoints: 100 in total (40 points for ID01, 35 points for ID04, and 25 points for ID03).

[0215] Real-time data transmission (UAV1 in progress): Waypoints flown: all 40 points of ID01 + the first 30 points of ID04 (total 70 points); Progress completed:

[0216] ;

[0217] Cost parameters: Labor cost: 200 yuan / hour; Drone depreciation: 50 yuan / hour; Forecasted cost baseline (total): 3,500 yuan;

[0218] Actual cost is dynamically collected, with real-time consumption data (up to 70% progress): Drone power consumption: 8.4 kWh (electricity price 0.8 yuan / kWh → 6.72 yuan); Manual monitoring hours: 1.2 hours → 240 yuan; Equipment depreciation: Yuan; Total actual cost: Yuan;

[0219] Output actual cost transaction record: Timestamp T+35min: Actual cost 305.02 yuan ( Yuan);

[0220] Cost deviation threshold determination and deviation calculation: The cost has decreased (far exceeding the 15% threshold); activation judgment flag: output cost reduction trigger signal (reason: the drone avoids terrain obstacles to shorten the path, saving 40% of flight time);

[0221] Risk quantification strategy decision-making; Enterprise risk tolerance coefficient: 0.7 (medium risk appetite); Strategy switching probability mapping: coefficient ; Decision logic: ;

[0222] Work order priority update execution, update rules: high-risk wind turbines (red label) detection weight 20%: Original weights: ID01=40%, ID04=35% → Updated: ID01=32%, ID04=28%; Frequency of detection for medium-risk wind turbines (yellow label) 30%: Original frequency: ID03=25% → After update: ID03=32.5% (25%×1.3);

[0223] Output work order priority update command:

[0224] Wind Turbine ID Original weight Updated weights Adjustment basis ID01 40% 32% High-risk weighting reduced ID04 35% 28% High-risk weighting reduced ID03 25% 32.5% Increased frequency of medium-risk

[0225] 206. It also includes real-time monitoring of the charging station status:

[0226] Monitor the charging pile's temperature, output current, and fault signals; mark it as a high-risk pile when the temperature is >60℃ or the current fluctuation is >15%; output a charging pile health status table.

[0227] Sandstorm stratified response mechanism: Receives sandstorm level (Level I-IV) issued by the meteorological station; Level I: Reduces flight altitude to within 5 meters of the blades; Level II: Reduces trackpoint spacing to 5 meters; Level III: Suspends detection and initiates the nearest landing; Outputs track downgrade command.

[0228] Closed-loop feedback of blade damage data: Analyze high-definition images of the blade surface taken by the UAV; when lightning damage or cracks (length > 10cm) are detected, output a blade damage alarm package.

[0229] It should be noted that the following uses a wind farm (including 10 2.0MW units) as an example to illustrate the specific implementation and data flow of the charging pile status monitoring, sandstorm response, and blade damage feedback steps (206):

[0230] Real-time monitoring of charging pile status, input data (real-time parameters of 2 charging piles):

[0231] Charging station A: Temperature 58℃, Output current 50A (fluctuation rate 8%), Fault signal: None;

[0232] Charging station B: Temperature 63℃ (>60℃), Output current 48A (fluctuation rate 18%>15%), Fault signal: Poor contact;

[0233] Monitoring rules: Temperature > 60℃ or current fluctuation > 15% → marked as high-risk pile.

[0234] Output health status table:

[0235] Charging station ID temperature Current fluctuation Fault signal Health status A 58℃ 8% none normal B 63℃ 18% Poor contact High risk

[0236] Sandstorm stratified response mechanism, input signal: Meteorological observatory issues sandstorm level II warning (visibility < 500 meters).

[0237] Response rules: Level I: Flight altitude drops to within 5 meters of the blade; Level II: Trackpoint spacing is reduced to 5 meters (from 10 meters); Level III: Detection is suspended and the aircraft lands at the nearest available location.

[0238] Output track downgrade command: Scope of effect: All UAVs; Track adjustment: ID01 track point spacing from 10 meters to 5 meters (coordinate sequence density doubled); Execution time: Effective immediately until warning is lifted;

[0239] Closed-loop feedback of blade damage data. Input data: High-resolution images of the ID03 blade taken by a UAV1 drone (0.5 mm / pixel resolution).

[0240] Damage identification: Lightning damage: Carbon fiber ablation at the blade tip (area 15cm²) 2 ); Crack characteristics: The longitudinal crack on the suction surface is 12cm long > 10cm (8 meters from the leaf root).

[0241] Output damage alarm packet:

[0242] Fan ID: ID03;

[0243] Damage type: Crack (suction surface);

[0244] Damage dimensions: Length 12cm × Width 0.8cm;

[0245] Location coordinates: (121.5, 38.7, 82.0);

[0246] Risk level: High risk (requires repair within 72 hours).

[0247] In this embodiment of the invention, the wind turbine spacing topology, equipment status (blade grounding resistance degradation rate), and meteorological threat (lightning warning level) are combined to achieve a three-dimensional coupled assessment of equipment status, cluster topology, and meteorological threat, generating dynamic risk labels. This multi-dimensional risk assessment method can more comprehensively and accurately reflect the actual risk status of wind turbines, providing a scientific basis for subsequent task scheduling and improving the targeting and effectiveness of detection. When generating detection work order instructions, not only risk priority is considered, but also the remaining power of the UAV and the return distance are combined to ensure that the task is completed within the time window and the UAV can return safely. At the same time, through historically dependent cost control, fractional differential equations are used to describe the memory effect of cost changes with detection time, generating UAV cluster deployment density instructions to achieve cost optimization control. This task scheduling method, which comprehensively considers timeliness and cost, improves resource utilization efficiency and reduces detection costs. A repulsive potential field containing the safety radius of lightning rods is constructed in the digital elevation model, and the terrain and the repulsive potential field between UAVs are integrated. Obstacle avoidance flight path sequence is generated through potential field gradient descent. This method fully considers the actual environmental factors of wind farms, effectively avoiding collisions between drones and lightning rods, terrain obstacles, and other drones, ensuring drone flight safety and improving the reliability of inspection tasks. Through strategy transition audit decision-making, it monitors the progress and cost of flight path execution in real time. When the actual inspection cost decreases by more than a threshold compared to the predicted value, it dynamically adjusts the inspection strategy based on the enterprise's risk coefficient and outputs work order priority update instructions. In addition, auxiliary mechanisms such as real-time monitoring of charging pile status, a sandstorm stratified response mechanism, and closed-loop feedback of blade damage data further enhance the adaptability, safety, and intelligence of the inspection system, forming a complete closed-loop feedback system that can adjust and optimize the inspection process in a timely manner according to actual conditions.

[0248] The above describes the task scheduling and path optimization management method for wind farm blade lightning protection inspection UAVs in embodiments of the present invention. The following describes the task scheduling and path optimization management device for wind farm blade lightning protection inspection UAVs in embodiments of the present invention. Please refer to [link to relevant documentation]. Figure 3An embodiment of the wind farm blade lightning protection inspection UAV task scheduling and path optimization management device of the present invention includes: a tag module 301, used to calculate regional connectivity based on the topological relationship between wind turbine spacing, superimpose the annual degradation rate of blade grounding resistance and real-time lightning warning level, and generate dynamic risk tags for wind turbines; a scheduling module 302, used to use the dynamic risk tags of wind turbines, sort them in descending order of risk value, and combine the remaining power of the UAV and the return distance to output inspection work order instructions with time windows; and an optimization module 303, used to parse the inspection work order instructions and call up similar wind turbines from the past 90 days. The average detection time is calculated using fractional differential equations to determine the derivative of the unit time cost and generate a drone swarm deployment density command. The path module 304 is used to construct a repulsive potential field containing the safety radius of the lightning rod in the digital elevation model based on the drone swarm deployment density command, and generate an obstacle avoidance path point sequence through gradient descent of the potential field. The monitoring module 305 is used to monitor the execution progress of the obstacle avoidance path point sequence. When the actual detection cost decreases by more than 15% compared to the predicted value, the strategy switching probability is calculated based on the enterprise risk coefficient. If the probability is greater than 0.6, a work order priority update command is output.

[0249] In this embodiment of the invention, dynamic risk labels are generated by comprehensively considering multiple factors such as wind turbine spacing topology, blade grounding resistance degradation, and lightning warning. This enables more accurate assessment of wind turbine risks, and the scheduling of testing tasks is arranged in descending order of risk value to ensure that high-risk wind turbines are prioritized for testing. This improves the targeting and timeliness of lightning protection testing and effectively reduces the risk of blade damage to wind farms caused by lightning strikes. The scheduling module combines the remaining power of the drone with the return distance to output testing work order instructions with time windows, preventing the drone from being unable to return or complete the testing task due to insufficient power, improving the efficiency of drone use and task completion rate, and reducing resource waste. The optimization module calls nearly 90 The average inspection time for wind turbines of the same type is calculated using fractional differential equations to determine the derivative of the unit time cost. This scientifically and rationally generates drone swarm deployment density instructions, optimizing drone resource allocation, improving inspection efficiency, and reducing overall inspection costs. The path module constructs a repulsive potential field containing the safety radius of lightning rods in the digital elevation model. By generating obstacle avoidance path point sequences through gradient descent of the potential field, it ensures safe flight of drones in complex environments, avoids collisions with obstacles, and guarantees the smooth progress of inspection tasks. The monitoring module monitors the progress of path execution. When the actual inspection cost decreases by more than 15% compared to the predicted value, it calculates the strategy switching probability based on the enterprise's risk coefficient and dynamically adjusts the work order priority, making inspection task scheduling more flexible and adaptable to changes in actual inspection conditions, further improving inspection management efficiency and effectiveness.

[0250] above Figure 3The wind farm blade lightning protection detection UAV task scheduling and path optimization management device in this embodiment of the invention is described in detail from the perspective of modular functional entities. The wind farm blade lightning protection detection UAV task scheduling and path optimization management device in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0251] Figure 4 This is a schematic diagram of a wind farm blade lightning protection inspection UAV task scheduling and path optimization management device 400 provided in an embodiment of the present invention. The wind farm blade lightning protection inspection UAV task scheduling and path optimization management device 400 can vary considerably due to differences in configuration or performance. Device 400 includes a transmitter 401, a receiver 402, and a processor 403. The processor 403 can also be a controller. Figure 4 The device is designated as "controller / processor 403". Optionally, the device 400 may also include a modem processor 405, which may include an encoder 406, a modulator 407, a decoder 408, and a demodulator 409.

[0252] In one example, transmitter 401 modulates (e.g., analog-to-analog conversion, filtering, amplification, and up-conversion, etc.) the output sample and generates an uplink signal that is transmitted via an antenna to an access network device. On the downlink, the antenna receives the downlink signal transmitted by the access network device. Receiver 402 modulates (e.g., filtering, amplification, down-conversion, and digitization, etc.) the signal received from the antenna and provides an input sample. In modem processor 405, encoder 406 receives service data and signaling messages to be transmitted on the uplink and processes (e.g., formatting, encoding, and interleaving) the service data and signaling messages. Modulator 407 further processes (e.g., symbol mapping and modulation) the encoded service data and signaling messages and provides an output sample. Demodulator 409 processes (e.g., demodulates) the input sample and provides a symbol estimate. Decoder 408 processes (e.g., deinterleaving and decoding) the symbol estimate and provides decoded data and signaling messages to device 400. Encoder 406, modulator 407, demodulator 409, and decoder 408 can be implemented by a combined modem processor 405. These units perform processing according to the radio access technology adopted by the radio access network (e.g., LTE and other evolved systems access technologies). It should be noted that when device 400 does not include modem processor 405, the above-mentioned functions of modem processor 405 can also be performed by processor 403.

[0253] The processor 403 controls and manages the operation of the device 400, and is used to execute the processing procedures performed by the device 400 in the above embodiments of this disclosure. For example, the processor 403 is also used to execute various steps of the transmitting or receiving device in the above method embodiments, and / or other steps of the technical solutions described in the embodiments of this disclosure.

[0254] Furthermore, the device 400 may also include a memory 404 for storing program code and data for the device 400.

[0255] Understandable, Figure 4 Only a simplified design of device 400 is shown. In practical applications, device 400 can include any number of transmitters, receivers, processors, modem processors, memory, etc., and all devices that can implement the embodiments of this disclosure are within the protection scope of the embodiments of this disclosure.

[0256] The present invention also provides a task scheduling and path optimization management device for wind farm blade lightning protection inspection drones. The wind farm blade lightning protection inspection drone task scheduling and path optimization management device includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor performs the steps of the wind farm blade lightning protection inspection drone task scheduling and path optimization management method in the above embodiments.

[0257] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the wind farm blade lightning protection detection UAV task scheduling and path optimization management method.

[0258] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0259] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0260] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for scheduling and path optimization management of unmanned aerial vehicles (UAVs) for lightning protection inspection of wind farm blades, characterized in that, The method for scheduling and path optimization management of UAV tasks for lightning protection inspection of wind farm blades includes: Based on the topological relationship of wind turbine spacing, the regional connectivity is calculated, and the annual degradation rate of blade grounding resistance and real-time lightning warning level are superimposed to generate dynamic risk labels for wind turbines. Using dynamic risk tags for wind turbines, sorted in descending order of risk value, and combined with the remaining battery power and return distance of the drone, output detection work order instructions with time windows; The system analyzes the inspection work order instructions, calls the average inspection time of similar wind turbines over the past 90 days, calculates the derivative of unit time cost using fractional differential equations, and generates instructions for the deployment density of drone swarms. Based on the deployment density instructions of the UAV swarm, a repulsive potential field containing the safety radius of the lightning rod is constructed in the digital elevation model, and an obstacle avoidance track point sequence is generated through gradient descent of the potential field; Monitor the progress of obstacle avoidance path point sequence. When the actual detection cost drops to the preset value compared to the predicted value, calculate the strategy switching probability based on the enterprise risk coefficient. If the probability is greater than the set value, output the work order priority update instruction.

2. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 1, characterized in that, include: Using the locations of the wind turbines as nodes, establish connecting edges to form a wind turbine adjacency graph; Analyze the interconnected clusters of units in the adjacency graph and mark clusters with more than 3 units as highly connected regions; Extract historical detection values ​​of blade grounding resistance, calculate the annual deterioration percentage, and generate equipment deterioration warning labels; Receive lightning warning level signals issued by the meteorological department and activate environmental risk indicators according to the warning level.

3. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 2, characterized in that, include: Place the wind turbines with red risk labels at the top of the queue, followed by those with yellow labels, and then those with green labels at the bottom. Wind turbines of the same color level are arranged in descending order of their grounding resistance degradation rate, and the risk-ranked wind turbine queue is output. Obtain the real-time remaining battery power of the drone, multiply it by the endurance conversion factor to obtain the safe flight time, combine it with the average cruising speed of the wind farm to calculate the maximum single-trip operating distance, and output the safe operating radius of each drone; Using the location of the wind farm charging pile as the center, match available drones, establish a dedicated mapping relationship table between drones and charging piles, and output a list of charging pile bindings. For the risk-ranked wind turbine queue, the nearest drone is assigned according to the safe operating radius of each drone, and a charging pile binding list is superimposed to ensure that the drone can return to the dedicated charging pile after the inspection is completed, and a work order instruction with three elements is output. When multiple drones are assigned to the same wind turbine, the turbine with the highest remaining power is selected first. The replaced drone is automatically reassigned to the next wind turbine in the queue for testing tasks, and an updated work order instruction set is output.

4. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 3, characterized in that, Set safe flight time as The maximum single-trip working distance is ,but: ； in, This represents the remaining battery percentage. This is the range conversion factor; ； in, Average cruising speed, in km / h. Safe flight time is measured in minutes; The safe operating radius of each drone is R: 。 5. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 4, characterized in that, include: The system analyzes the target fan model and blade length in the detection work order instruction, matches the fan technical files in the historical database according to the model, and outputs identification tags for the same type of fan. Retrieve the testing time records of the same model of fan from the past 90 days, remove outliers that exceed the time limit, and output a standard time distribution table; Based on the standard time distribution table, combined with real-time labor cost rate and drone depreciation parameters, a fractional differential equation is used to describe the memory effect of cost change with detection time, and the curve of cost change rate per unit time is output. When the rate of change of cost per unit time exceeds the cost change rate threshold, the number of drones per unit area is increased, and the drone / square kilometer density control value is output. When the density control value fluctuates by more than 30% compared to the previous instruction, the operation and maintenance expert review process is triggered. Based on the review opinions, the density control value is corrected, and the final deployment density instruction is output.

6. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 5, characterized in that, include: Read the drone density control value per square kilometer, convert it into the minimum horizontal spacing standard between drones, and output the spacing constraint parameters; Obtain the geographic coordinates of all lightning rods in the wind farm, generate a cylindrical repulsion field centered on the coordinates, and output the lightning rod potential field layer.

7. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 6, characterized in that, In the digital elevation model, areas with a slope greater than 30° are marked as terrain obstacles. Spacing constraint parameters are superimposed to generate a repulsion field between UAVs. The potential field layer of lightning rods is then fused to output a three-dimensional composite potential field model. Starting from the coordinates of the target wind turbine tower base, iteratively search along the negative gradient direction of the composite potential field, outputting the coordinates of the track point every 10 meters, and outputting the obstacle avoidance track point sequence; Automatically mark track points less than 10 meters from the lightning rod, push high-risk points to the operation and maintenance console for manual confirmation, and integrate the confirmation results to generate a final review track sequence.

8. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 7, characterized in that, include: Receive the coordinates of the flight path points that the UAV has passed back in real time, compare them with the planned flight path point sequence, calculate the percentage of completion, and output a flight path execution progress report. Simultaneously acquire drone power consumption and manual monitoring time data, overlay equipment depreciation rate to calculate real-time detection costs, and output actual cost flow records; Call the predicted cost baseline value, and when the actual cost flow record decreases by more than 15%, activate the judgment flag and output the cost decrease trigger signal; Read the risk tolerance coefficient preset in the enterprise management system, map the strategy switching probability according to the coefficient value, and output the strategy switching probability value; Based on the switching probability value of the strategy, output the work order priority update instruction.

9. The method for task scheduling and path optimization management of wind farm blade lightning protection inspection UAVs according to claim 8, characterized in that, It also includes real-time monitoring of the charging station status: Monitor the charging pile temperature, output current, and fault signals, and output a charging pile health status table; Receive the sandstorm level report issued by the meteorological station and output a track downgrade command; The system analyzes high-resolution images of the blade surface captured by drones and outputs a blade damage alarm packet when lightning damage or cracks are detected.

Images