Unmanned aerial vehicle inspection path planning method and system for power grid lines

Abstract

The application discloses a UAV inspection path planning method and system for power grid lines, and particularly relates to the technical field of power grid inspection, comprising the following steps: step one, dynamically fusing multi-source sensors and prior data, constructing and real-time updating a comprehensive threat field and a passable degree map with a time attenuation mechanism; step two, decomposing the inspection line into sub-task sections, and generating a reference flight path for each section in combination with a static safety cost; step three, in the flight of the UAV, solving a locally optimal trajectory on-line through a rolling window optimization based on a real-time map and an adaptive weight adjustment; and step four, realizing multi-UAV cooperation and automatic transfer of blocked tasks through a blockchain-type distributed task state ledger and intelligent contract rules. The application significantly enhances the real-time avoidance capability of the UAV for dynamic obstacles, environmental adaptability and task continuity, and effectively guarantees the safety and efficiency of power grid inspection operations.

Description

Technical Field

[0001] This invention relates to the field of power grid inspection technology, and more specifically, to a method and system for unmanned aerial vehicle (UAV) inspection path planning for power grid lines. Background Technology

[0002] As a critical national infrastructure, the safe and stable operation of the power grid is of paramount importance. Traditional manual inspection methods suffer from inherent drawbacks such as low efficiency, high risk, and limited coverage. In recent years, drones have been widely used in power transmission line inspections due to their flexibility, efficiency, and ability to conduct close-range observations. The operational efficiency of drones is highly dependent on the quality of their flight path planning. Currently, most mainstream drone inspection path planning relies on fixed routes generated from preset waypoints or static three-dimensional path planning based on known geographical information (such as digital elevation models and tower coordinates). These methods work effectively in stable environments with known obstacles, but they inherently lack the ability to respond in real time to dynamic and uncertain factors in the inspection environment.

[0003] Specifically, existing technical solutions face the following challenges: First, the power grid inspection environment is complex, often encountering unmapped or temporarily appearing dynamic obstacles, such as construction machinery near power lines and migrating flocks of birds. Existing static planning methods cannot perceive and predict these dynamic threats. Once such obstacles appear on the preset flight path, drones can usually only execute simple emergency strategies (such as hovering alarms or returning to base), leading to interruptions in the inspection mission and requiring manual intervention or subsequent re-flight, severely impacting operational continuity and efficiency. Second, for sudden localized severe weather (such as gusts or airflow disturbances) or communication signal interference, existing path planning systems lack effective environmental perception fusion and online replanning mechanisms, making it difficult to dynamically adjust flight paths to avoid risks. Finally, in long-distance line or multi-drone collaborative inspection scenarios, existing solutions mostly employ centralized task scheduling, which cannot achieve automatic and reliable task handover when a single drone fails or becomes stranded, resulting in insufficient system robustness.

[0004] Therefore, existing power grid drone inspection path planning technologies suffer from significant shortcomings, including insufficient real-time obstacle avoidance capabilities, a lack of online intelligent replanning mechanisms, and unintelligent multi-drone collaborative fault takeover. There is an urgent need for an intelligent path planning method capable of real-time perception and fusion of dynamic environmental information, online autonomous replanning of local paths, and ensuring task continuity through a decentralized collaborative mechanism. This would enhance the all-weather adaptability, operational safety, and overall efficiency of power grid drone inspections. Summary of the Invention

[0005] To overcome the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a method and system for unmanned aerial vehicle (UAV) inspection path planning for power grid lines.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The method for unmanned aerial vehicle (UAV) inspection path planning for power grid lines includes the following steps:

[0008] S1. Dynamic Fusion Modeling of Inspection Environment: Integrates static basic data and real-time dynamic data to construct and update the comprehensive threat field in real time, and generates a drone accessibility map;

[0009] S2. Global Task Decomposition and Baseline Path Generation: The inspection route is decomposed into multiple sub-task segments, and a theoretically optimal baseline route is generated for each sub-task segment.

[0010] S3. Online rolling optimization and local path replanning: When the UAV flies along the baseline route, it performs rolling optimization within a limited time window based on the real-time updated local comprehensive threat field to solve and execute the optimal local flight trajectory.

[0011] S4. Multi-drone distributed collaboration and emergency task takeover: The drone swarm communicates through a self-organizing network and, based on a distributed task status ledger and smart contract rules, achieves task status synchronization and automatic handover of obstructed tasks.

[0012] Specifically, the construction of the integrated threat field in S1 includes:

[0013] The flight space is discretized into a three-dimensional grid. For each grid cell, the static obstacle threat value, dynamic obstacle threat value, meteorological threat value, and communication threat value are calculated separately, and the maximum value is taken as the comprehensive threat value of the grid.

[0014] Specifically, when generating the reference route in S2:

[0015] A static security cost term is introduced into the cost function of the path search algorithm, and the generated path is kept at a distance of not less than a preset minimum security interval from known static threat sources.

[0016] Specifically, in S3:

[0017] Rolling optimization is achieved by constructing and solving a constrained optimization problem whose objective function combines the costs of deviating from the baseline route, crossing the threat field, and approaching the sub-task endpoint. The constraints include UAV dynamics constraints and obstacle avoidance hard constraints.

[0018] Specifically, in S3:

[0019] The weight coefficients of each item in the objective function can be adaptively and dynamically adjusted according to real-time environmental risk factors, track deviation factors, and mission progress ratios.

[0020] Specifically, S3 also includes deadlock determination:

[0021] If a feasible solution cannot be found through rolling optimization in multiple consecutive decision cycles, it is determined to be in a deadlock and the collaborative takeover mechanism of S4 is triggered.

[0022] Specifically, in S4:

[0023] The distributed task state ledger is a lightweight blockchain-like structure. Each drone acts as a node to maintain a local copy of the ledger. The ledger records task state changes through transactions and achieves ledger consistency through a consensus mechanism.

[0024] Specifically, the smart contract rules are encoded in the ledger and used to monitor the task status. When a subtask segment is detected to be in an alarm state for more than a preset time, the process of resetting the task and bidding for an idle drone is automatically triggered, and a new execution drone is assigned according to the bidding results.

[0025] A drone inspection path planning system for power grid lines includes the following modules:

[0026] The dynamic environment perception and fusion module is used to fuse multi-source sensor data and prior data, and to construct and update a three-dimensional accessibility map in real time through threat value quantification and fusion.

[0027] The global task management and planning module is used to decompose the inspection route into tasks and generate a baseline route for each sub-task based on static safety costs.

[0028] The online rolling optimization and replanning module is used to solve the optimal local trajectory under adaptive weights in flight based on a real-time environment map through a constrained rolling window.

[0029] The multi-machine distributed collaboration and decision-making module is used to achieve cluster consensus through a blockchain-style task status ledger and execute smart contracts to trigger the automatic handover and takeover of tasks.

[0030] The technical effects and advantages of this invention are as follows:

[0031] Significant improvements have been achieved in the safety, intelligence, and adaptability of path planning. By constructing a "comprehensive threat field" that integrates dynamic and static multi-dimensional threats and introducing a time decay mechanism, a precise and time-varying environmental perception foundation is provided for UAVs, enabling them to proactively assess risks. Based on this, a global baseline path generation method incorporating static safety costs lays a higher initial foundation for flight safety. Furthermore, online rolling optimization and adaptive weight adjustment mechanisms allow UAVs to dynamically and intelligently solve for optimal local trajectories during flight based on real-time risks, trajectory deviations, and mission progress. This achieves an optimal balance between safe obstacle avoidance and efficient inspection in complex dynamic environments, significantly reducing mission interruptions caused by sudden obstacles.

[0032] Significant breakthroughs have been achieved in overall system robustness and task continuity. By introducing a multi-machine distributed collaborative mechanism based on a lightweight blockchain architecture, a decentralized, tamper-proof task state ledger and automatically executable smart contract rules have been constructed. When a single drone gets stuck due to dynamic obstacles, the system can automatically and reliably trigger the process of task state discovery, reset, bidding, and handover, allowing other idle drones to seamlessly take over the inspection task. This effectively overcomes the drawback of single-point failures causing the entire task to stall under the traditional centralized scheduling model, significantly improving the autonomous recovery capability and overall task completion rate of long-distance, multi-machine collaborative inspection operations. Attached Figure Description

[0033] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] like Figure 1 As shown, the steps of the drone inspection path planning method for power grid lines are as follows:

[0036] Step 1: Dynamic Environment Fusion Modeling. By fusing multi-source static and real-time sensor data, a comprehensive threat field containing spatial and temporal dimensions is constructed and continuously updated, quantifying it into a 3D accessibility map. Various dynamic and static threats (such as obstacles, severe weather, and signal interference) are uniformly quantified, and a predictive time decay mechanism is introduced to provide a precise, time-varying environmental situational basis for path planning; the process is as follows:

[0037] The onboard processor of the drone integrates and processes the following data sources:

[0038] Static base data: Pre-loaded precise 3D model of transmission lines, tower coordinates, no-fly zones (such as airspace above residential areas), and information on known fixed obstacles (such as mountains and high-rise buildings).

[0039] Real-time dynamic data: obstacle information (such as location, size, and motion vector) perceived in real time through airborne vision (RGB camera, infrared thermal imager), lidar (LiDAR), and millimeter-wave radar; real-time micro-meteorological data (such as local wind speed and direction, turbulent areas) obtained through meteorological sensors and ground meteorological service networks; and real-time communication link quality maps from ground base stations or other UAVs.

[0040] Fusion Modeling: All dynamic and static information is unified into a four-dimensional (three-dimensional space + one-dimensional time) "comprehensive threat field". Static obstacles are assigned a constant high threat value; dynamic obstacles are assigned a time-varying threat value based on their movement trends and predicted spatial occupancy within the next few seconds; areas with severe weather and weak signals are assigned corresponding threat values ​​based on their intensity / attenuation. A real-time updated drone accessibility map is generated.

[0041] Fusion modeling is achieved through the following steps:

[0042] Spatial discretization: A three-dimensional cubic grid space with a side length of L (e.g., 500 meters) is constructed centered on the current GPS coordinates of the UAV. The grid resolution is set according to the accuracy requirements (e.g., 5 meters × 5 meters × 2 meters). Each grid cell is represented by a data structure containing the three-dimensional coordinates (x, y, z) of its center point;

[0043] Threat Quantification and Mapping:

[0044] Static Threat Value : Query the preloaded digital elevation model and obstacle database. If the grid cell is located inside a geographical obstacle such as a building or mountain, then =100; if located within a no-fly zone, then =100; otherwise =0;

[0045] Dynamic obstacle threat value For radar-identified dynamic obstacles (such as cranes), calculate their outer envelope cuboid and extrapolate their linear motion over a future Δt (e.g., 5 seconds) based on their real-time velocity vector. For each grid cell, calculate the shortest distance d to the predicted envelope of that obstacle. Dynamic threat value. according to =80*exp(-0.1*d), where d is in meters. For visually identified flocks of birds, they can be considered as a dynamic threat cloud, and the threat value is calculated using a similar formula but with a different attenuation coefficient.

[0046] Weather threat level For the identified areas of strong wind, obtain their wind speed vectors. Threat value Calculated as ,in , The components of the wind speed vector in three-dimensional space. , For coefficients determined based on the aerodynamic stability of the unmanned aerial vehicle platform (e.g.) =2, =3);

[0047] Communication Threat Value Calculate based on the real-time received signal strength index RSSI and signal-to-noise ratio SNR. ,in and To maintain reliable communication And the minimum threshold for signal-to-noise ratio (SNR);

[0048] Threat Fusion: The overall threat value for each grid cell. Take the maximum value of each threat value, that is This method ensures that any threat source is sufficient to trigger an obstacle avoidance response;

[0049] Accessibility map generation: setting safety thresholds (For example, 30). [The following appears to be a separate, unrelated sentence:] Will The grid is marked as "passable" (passage cost is 1), and The grid is marked as "high risk" (passage cost is 10), and will The grid cells are marked as "impassable" (with an infinite cost of passage). Finally, a 3D, cost-based accessibility raster map is generated for the path planning module to use.

[0050] The integrated threat field incorporates a time decay mechanism; for the threat value generated by dynamic obstacle prediction, the threat intensity decays as the prediction time increases; let the initial threat value generated by a certain dynamic obstacle to a certain grid cell be... This threat value corresponds to the predicted future The position in seconds indicates the decayed threat value. The calculation is as follows:

[0051] ;

[0052] in, This is the attenuation coefficient (set according to sensor accuracy and obstacle type, for example, η=0.15). For the predicted time (in seconds). When When the value is 0, the attenuation factor is 1, and the threat value is at its maximum; as... As the attenuation factor increases, the threat value decreases. For the same grid cell, if there are multiple dynamic obstacle threats, the maximum attenuated threat value is taken as the dynamic threat value of that grid cell, i.e.: ;

[0053] Then Used for threat fusion to obtain a comprehensive threat value. This mechanism ensures that the further into the future the obstacle's location is predicted, the lower the confidence level of its threat value, thus providing a reasonable buffer during planning and avoiding overly conservative obstacle avoidance behavior due to the uncertainty of long-term predictions.

[0054] Step Two: Global Task Decomposition and Baseline Path Generation. The long-route inspection task is intelligently decomposed into several sub-task segments, and then an initial baseline route is generated for each sub-task. A safety cost function and safety distance constraints based on static prior knowledge are introduced into the path search, enabling the baseline route to actively avoid known high-risk areas while pursuing geometric optimization, providing a safer and more efficient reference trajectory for subsequent real-time planning. The process is as follows:

[0055] Before the mission begins, the ground control center or the lead drone divides the entire route to be inspected into multiple sub-task segments according to tower sections or fixed lengths (e.g., every 5 kilometers). For each sub-task segment, combined with the static basic data from step one, a theoretically optimal path is pre-calculated as a baseline using the A* or Quick Random Tree (RRT) algorithm.

[0056] To improve the practicality and safety of the baseline route, the cost function of the path search algorithm incorporates a safety cost based on prior knowledge of the static environment, in addition to the traditional distance cost and heuristic cost. Specifically, for the A* algorithm, the evaluation function is extended as follows:

[0057] ;

[0058] in, Let be the evaluation function of the A* algorithm. From the starting point to the node The actual cost (such as the distance already flown). From node Heuristic cost to the target point (e.g., Euclidean distance). It is a node The static security cost. and These are adjustable weighting coefficients used to balance path length optimality and safety.

[0059] Static security costs The calculation is based on the static obstacle and no-fly zone information loaded in step one, with the aim of keeping the path away from known danger zones. The calculation method is as follows:

[0060] ;

[0061] in, This refers to the number of static threat sources (such as buildings, mountains, and no-fly zones); It is a node To the The shortest distance to the surface of a static threat source; It is the first The weighting coefficient of a threat category (reflecting its degree of danger); It is the distance attenuation coefficient.

[0062] At the same time, a hard constraint is set during the path search process: it must meet the following conditions. ,in A preset minimum safety interval (e.g., 15 meters) is used; otherwise, the node is considered unreachable. This is achieved by introducing... With the constraint of safe distance, the generated baseline route, based on theoretical optimization, has higher inherent security and provides better initial conditions for responding to dynamic threats.

[0063] Finally, the route is sent to the drones prepared to perform this segment of the mission. This route serves as a reference for subsequent dynamic adjustments.

[0064] Step 3: Online rolling optimization and replanning. Within the finite prediction time domain, multiple objectives such as environmental threats, trajectory tracking, and terminal approach are constructed as optimization problems, and model predictive control is used for real-time solution. By introducing a weight adaptive adjustment mechanism, the planning strategy is dynamically adjusted according to real-time risks, trajectory deviations, and mission progress to achieve the optimal balance between safety and efficiency. The process is as follows:

[0065] As the drone flies along the baseline route, it continuously executes the following cycle:

[0066] Perception and prediction: Based on the method in step one, the local "comprehensive threat field" centered on itself with a radius of R (e.g., 200 meters) is updated in real time, and the field changes in the future time T (e.g., 15 seconds) are predicted.

[0067] Rolling window optimization: In each decision cycle (e.g., per second), the UAV does not replan the entire remaining path, but instead optimizes within a forward-extending, finite time window (e.g., the flight space for the next 30 seconds). The optimization objective function comprehensively considers: a) the cost of deviating from the baseline route; b) the cumulative risk of traversing the "threat field"; and c) the proximity of the remaining key points (such as the next base tower) within this sub-task segment. The optimal flight trajectory sequence within this time window is solved using a Model Predictive Control (MPC) framework or an improved local search algorithm.

[0068] The specific optimization problem is structured as follows:

[0069] State variables and control variables: Define the decision time domain as Step (corresponding to the future) Time, such as =30, =1 second). The state variable is the drone's position at each step. and speed The control variable is acceleration. ;

[0070] The specific form of the objective function J:

[0071] ;

[0072] in: It is the reference position of the baseline route at step n; It is the preset cruising speed; Let be the predicted speed of the drone at step n; The access cost for this location is obtained by querying the accessibility map generated in step one. Let n be the predicted position of the drone at step n; It is the endpoint of the current subtask segment (the preset hovering point of the next tower). , , , , These are pre-calibrated weighting coefficients used to balance different objectives such as trajectory tracking, speed maintenance, control smoothing, obstacle avoidance, and terminal approach; to enable rolling window optimization to adapt to different flight environments and mission phases, the weighting coefficients... , , , , The strategy is dynamically adjusted based on real-time environmental conditions. The adjustment strategy is based on current environmental risks, track deviation, and mission progress; the process is as follows:

[0073] Define environmental risk factors , representing the average travel cost in the predicted time domain;

[0074] Define track deviation ;

[0075] Define the proportion of remaining tasks = Remaining distance of the current subtask / Total length of the subtask.

[0076] The rules for dynamic weight adjustment are as follows:

[0077] Environmental risk adjustment: when (risk threshold, for example) When the value is 5, increase the obstacle avoidance cost weight. Adjusted to ,in for The base value, To adjust the gain (e.g.) =0.1);

[0078] Track deviation adjustment: when (deviation threshold, for example) When the distance is less than 10 meters, enhance track tracking. Adjusted to ,in for The base value, To adjust the gain (e.g.) =0.05);

[0079] Task progress adjustment: When (Progress threshold, for example) When the distance reaches 0.2 (meaning 20% ​​of the distance remains), the enhanced terminal approaches. Adjusted to ,in for The base value, To adjust the gain (e.g.) =0.2).

[0080] Through the aforementioned adaptive adjustment mechanism, rolling optimization can automatically enhance obstacle avoidance in high-risk areas, strengthen tracking when deviating from the flight path, and accelerate towards the endpoint at the end of the mission, thereby dynamically optimizing the flight trajectory while ensuring safety and improving overall inspection efficiency and adaptability.

[0081] Constraints: Dynamic constraints: The relationship between velocity and acceleration is similar;

[0082] Control constraints: (Maximum acceleration of the drone);

[0083] Speed ​​constraints: (Maximum speed of the drone);

[0084] Obstacle avoidance hard constraints: (That is, you cannot enter the "impassable" grid).

[0085] Solving and Deadlock Detection: In the onboard computing unit, an efficient C code solver is generated using the ACADO toolbox or CasaADi for real-time solving. If the solver returns "no feasible solution" for K consecutive decision cycles (e.g., K=5), the UAV is determined to be in a "deadlock," immediately triggering the collaborative takeover mechanism in step four.

[0086] Decision-making and execution: Execute the first trajectory obtained from the optimization (such as an instruction for the next second), then slide the time window and repeat the above steps. If local replanning fails to find a path that meets the safety threshold multiple times (i.e., gets stuck in a "dead end"), then the collaborative mechanism in step four is triggered.

[0087] Step Four: Multi-drone Distributed Collaboration and Emergency Takeover. Based on a lightweight blockchain architecture, the drone cluster maintains a distributed, consensus-based task state ledger. Through smart contract rules encoded in the ledger, when a drone malfunctions or is stranded for an extended period, the task is automatically re-tendered and handed over, achieving decentralized task continuity and ensuring the continuity of inspection operations. The process is as follows:

[0088] The drone swarm maintains communication through a self-organizing network. Each drone acts as a node, maintaining a shared, lightweight blockchain-style "task status ledger." The ledger records information in the form of blocks, including: subtask segment ID, currently executing drone ID, task status (pending / in progress / completed / alarmed), progress percentage, and any special threats encountered (such as the location of strong wind areas).

[0089] State synchronization and consensus: When a drone updates its own mission status (such as starting execution, progress update, triggering an alarm), it broadcasts this information to the network as a transaction. After neighboring drones verify this, they package it into a new block and attach it to their respective ledger chains through a simplified consensus mechanism (such as a simplified version of Practical Byzantine Fault Tolerance PBFT) to achieve state consistency.

[0090] Smart Contract Triggering and Task Transfer: Predefined smart contract rules are encoded in the ledger. For example, a rule could be defined as: "If a drone marks a sub-task segment as 'alarm' (stuck due to dynamic obstacles) for more than RE minutes and its progress is not updated, the sub-task segment's status will be automatically reset to 'pending execution,' and a task bid will be sent to the nearest available / soon-to-be-available drone in the network that is closest to the segment's start point." A qualified drone can automatically respond and accept the task, flying from the nearest safe rendezvous point to the faulty segment to resume inspection from the breakpoint or start point. The original faulty drone will then either return to base or land safely, depending on the situation.

[0091] The specific implementation of the lightweight blockchain-based task state ledger and smart contracts is as follows:

[0092] Data structure definition: Transaction : in, For transaction types (such as status updates, task bidding); The specific content is in JSON format (e.g., {"Status": "Alert", "Progress": 45, "Location": [x, y, z]}). The private key of the sender's UAV is used to sign the transaction hash;

[0093] Block : . Include: (Hash of the previous block), MerkleRoot (Merkle root of the transaction list in this block), Timestamp, Nonce, ConsensusInfo (consensus information).

[0094] Simplified PBFT consensus process (per round):

[0095] Preparatory Phase: Valid transactions collected by the drones (which can be elected by ID or based on signal strength) that are currently the master node in the view are packaged into block B and broadcast.<PRE-PREPARE,V,N,H(B),B> The message is defined as follows: V is the view number, and N is the sequence number.

[0096] Preparation Phase: After receiving the pre-preparation message from the master node, node i verifies the signature and transaction validity. If valid, it broadcasts the message.<PREPARE,V,N,H(B),i> The message is sent to all other nodes;

[0097] Commit Phase: After node i receives 2f ready messages from different nodes that match the pre-ready message (f is the number of allowed faulty nodes in the cluster), it enters the commit phase and broadcasts the message.<COMMIT,V,N,H(B),i> information;

[0098] Reply and accounting: After receiving 2f+1 commit messages matching the pre-prepared messages, node i considers block B to have reached consensus, appends it to its local blockchain, and executes in-block transactions to update its local task state view. Simultaneously, it sends a message to the master node or relevant drones.<REPLY,V,N,Result> information.

[0099] The following are modules of the drone inspection path planning system for power grid lines:

[0100] The dynamic environment perception and fusion module receives and fuses obstacle information, meteorological sensor data, and communication link status from vision, lidar, millimeter-wave radar, and other sources. Through quantization and fusion algorithms, it constructs a dynamically updated "comprehensive threat field" and a three-dimensional accessibility map, providing real-time and accurate environmental situation perception for path planning.

[0101] The global task management and planning module intelligently decomposes the transmission lines to be inspected into multiple sub-task segments. Combining static geographic information, it uses enhanced A* / RRT and other algorithms to calculate a baseline route for each sub-task segment that takes into account both path length and static safety, and distributes it to the designated UAVs as a global reference for subsequent real-time adjustments.

[0102] The online rolling optimization and replanning module, based on a real-time accessibility map, performs rolling optimization within a finite future time window using a model predictive control (MPC) framework in each decision cycle. By adaptively adjusting the objective function weights, it dynamically solves for the optimal flight trajectory that satisfies dynamic constraints and obstacle avoidance constraints, and drives the UAV to execute the flight path.

[0103] The multi-machine distributed collaboration and decision-making module, based on a lightweight blockchain architecture, maintains a consistent and tamper-proof distributed task state ledger in the cluster. It synchronizes task states through a simplified consensus mechanism and executes preset smart contract rules. In the event of a single machine failure or obstruction, it automatically triggers the redistribution and takeover of tasks, ensuring the robustness and continuity of inspection tasks.

[0104] The above formulas are all dimensionless calculations. Dimensionless calculations can be performed using various methods such as standardization, which will not be elaborated here. The formulas are derived from software simulations based on a large amount of collected data, and the preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, ATA hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0106] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0107] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0108] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0109] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0110] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0111] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion 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 this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable ATA hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for unmanned aerial vehicle inspection path planning for power grid lines, characterized in that, Comprise the following steps: S1, patrol environment dynamic fusion modeling: fusion static basic data and real-time dynamic data, build and real-time update integrated threat field, generate unmanned aerial vehicle accessible degree map, including: The flight space is discretized into a three-dimensional grid. For each grid cell, the static obstacle threat value, dynamic obstacle threat value, weather threat value, and communication threat value are calculated respectively, and the maximum value of the four threat values is taken as the comprehensive threat value of the grid. For the threat value generated by dynamic obstacle prediction, the threat intensity decays with the extension of the prediction time. For the same grid cell, if there are multiple dynamic obstacle threats, the maximum value of the multiple decayed dynamic obstacle threat values is taken as the dynamic obstacle threat value of the grid. S2, global task decomposition and benchmark path generation: the patrol route is divided into multiple sub-tasks, and a theoretically optimal benchmark flight path is generated for each sub-task, including: For each sub-task segment, a theoretical optimal path is calculated as a benchmark route in advance by using path search algorithm combined with static basic data, and the cost function of the path search algorithm is based on traditional distance cost and heuristic cost, and the safety cost based on static environment prior knowledge is integrated, and the calculation method of static safety cost s(n) is: wherein M is the number of static threat sources, d j，n is the shortest distance from node n to the surface of the jth static threat source, ω j is the weight coefficient of the jth threat reflecting the degree of danger, and φ j is the distance attenuation coefficient. And make the generated path keep a distance of not less than the preset minimum safety interval from the known static threat source; S3, online rolling optimization and local path replanning: when the unmanned aerial vehicle flies along the benchmark flight path, based on the real-time updated local integrated threat field, rolling optimization is performed within a limited time domain window to solve and execute the optimal local flight trajectory; Wherein, the rolling optimization is realized by constructing and solving a constrained optimization problem, the objective function of the problem integrates the cost of deviating from the benchmark flight path, the cost of crossing the threat field, and the cost of approaching the sub-task endpoint, the constraint conditions include the unmanned aerial vehicle dynamics constraint and the obstacle avoidance hard constraint; The weight coefficients in the objective function can be adaptively and dynamically adjusted according to the real-time environmental risk factor, the flight path deviation factor and the task progress ratio; Real-time environmental risk factors wherein, is the position passing cost obtained according to the passability map, N is the number of flight steps of the UAV, and n is the step number. Course deviation factor wherein, is the predicted position of the UAV at the n-th step, is the reference position of the reference air line at the n-th step; Task progress ratio = current sub-task remaining distance / sub-task total length; S4, multi-machine distributed cooperation and task emergency takeover: the unmanned aerial vehicle group communicates through self-organizing network, based on distributed task state ledger and smart contract rules, realizes task state synchronization and automatic transfer of blocked tasks, including: The smart contract rule is encoded in the ledger for monitoring task state; When it is detected that the sub-task segment is in the alarm state for more than a preset time, the task reset and bidding process to the idle unmanned aerial vehicle are automatically triggered, and a new execution unmanned aerial vehicle is assigned according to the bidding result. 2.The method for planning a UAV inspection path for power grid lines according to claim 1, wherein, The S3 also includes dead end judgment: If the rolling optimization cannot find a feasible solution in continuous multiple decision cycles, it is judged to be in a dead end, and the cooperation mechanism of S4 is triggered. 3.The method for planning a UAV inspection path for power grid lines according to claim 1, wherein, In the S4: The distributed task state ledger is a lightweight blockchain structure, and each unmanned aerial vehicle acts as a node to maintain a local ledger copy. The ledger records the task state changes through transactions, and achieves ledger consistency through consensus mechanism.

4. The system for applying the method of path planning for drone inspection of power grid lines according to any one of claims 1-3, characterized in that, Comprise the following modules: Dynamic environment perception and fusion module, for fusing multi-source sensors and prior data, constructing and real-time updating three-dimensional accessibility map through threat value quantization and fusion; Global task management and planning module, for task decomposition of patrol route, and generating benchmark flight path for each sub-task combined with static safety cost; An online rolling optimization and re-planning module for solving optimal local trajectory under adaptive weights by constrained rolling window optimization based on real-time environment map in flight; A multi-machine distributed coordination and decision module for realizing cluster consensus through a block-chain task state ledger and executing an intelligent contract to trigger automatic handover and takeover of tasks.

Images