Multi-unmanned aerial vehicle power transmission line inspection autonomous path planning method

Abstract

The invention discloses a multi-unmanned aerial vehicle power transmission line inspection autonomous path planning method. The method comprises the steps of performing digital processing on a power transmission line map; constructing a multi-objective optimization path planning model, and considering a plurality of optimization objectives of inspection time, inspection efficiency and unmanned aerial vehicle load balance while minimizing a total motion path; smoothing the initial path output by the path planning model by adopting an improved Dubins curve, and generating a continuous smooth path conforming to the flight ability of the unmanned aerial vehicle through track point sorting and dynamic adjustment of the maximum curvature of the Dubins curve; and performing unmanned aerial vehicle deployment optimization on the smoothed path by using an improved bait bird foraging algorithm. According to the method, the efficiency and the accuracy of multi-unmanned aerial vehicle path planning are remarkably improved, the smoothness and the stability of the path are optimized, the cooperative inspection efficiency of the multi-unmanned aerial vehicle is enhanced, the energy consumption and the cost are reduced, and safe and stable operation of the power transmission line is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle inspection and path planning, and in particular to an autonomous path planning method for multi-unmanned aerial vehicle power transmission line inspection. Background Art

[0002] With the rapid development of today's society, electricity has become an important cornerstone to support national economic construction and people's livelihood security. As the core channel for electric energy transmission, transmission lines are widely distributed throughout the country, crossing various complex terrains and climatic conditions, including mountains, lakes, basins, canyons and other diverse landforms. my country has a vast territory and the coverage of the power grid continues to expand, which makes the transmission lines face a more severe natural environment and potential operating risks. In particular, transmission and transformation towers are often set up in areas with complex geographical conditions and susceptible to meteorological changes. Frequent extreme weather not only increases the probability of transmission line failures, but also accelerates the loss and aging of electrical equipment, posing a major challenge to the safe and stable operation of the power grid.

[0003] In view of this, regular, efficient and comprehensive inspections of transmission lines have become an indispensable part of ensuring the safety of the power grid. At present, the main inspection methods for transmission lines include manual inspection, helicopter inspection, robot inspection and drone inspection. Among them, although manual inspection can find problems intuitively and quickly, it is limited by low inspection efficiency, blind spots and safety hazards in complex environments; although helicopter inspection and robot inspection can reduce the burden on personnel, the former is not economical and difficult to adapt to complex terrain, and the latter is greatly affected by the external environment and has limited inspection efficiency. In contrast, drone inspection has gradually become the mainstream method of transmission line inspection with its advantages of safety, efficiency, strong adaptability, lightness and flexibility, and high data acquisition rate. Especially in scenarios with a large inspection range, the collaborative operation of multiple drones has shown great potential.

[0004] However, as the core of drone inspection technology, the quality of the algorithm for multi-UAV autonomous path planning directly affects the efficiency and effectiveness of the inspection. Although many experts and scholars have conducted in-depth research on the path planning methods for drone transmission line inspection and proposed a variety of solutions, the existing methods are mostly based on bio-inspired algorithms and have certain limitations. Specifically, these methods fail to fully demonstrate how to flexibly adjust the path planning strategy according to different drone configurations and numbers, and often face the problems of low path planning efficiency and high computational complexity.

[0005] Therefore, developing a multi-UAV autonomous path planning method that can efficiently and flexibly respond to changing inspection needs while ensuring inspection effectiveness and efficiency is of great significance for improving the overall level of transmission line inspection and ensuring the safe and stable operation of the power grid. Summary of the invention

[0006] In view of this, the present invention proposes an autonomous path planning method for multi-UAV power transmission line inspection to solve the problems of low path planning efficiency and poor scalability faced by the UAV power transmission line inspection technology in the prior art.

[0007] The specific technical solutions of the present invention are as follows:

[0008] A multi-UAV power transmission line inspection autonomous path planning method, comprising:

[0009] Step 1: Digitally process the transmission line map, including accurate input of height, obstacles and transmission line location information; construct a multi-objective optimization path planning model, while minimizing the total motion path, taking into account multiple optimization objectives such as inspection time, inspection efficiency and UAV load balance;

[0010] Step 2: Use the improved Dubins curve to smooth the preliminary path output by the path planning model, and generate a continuous and smooth path that meets the flight capabilities of the UAV by sorting the trajectory points and dynamically adjusting the maximum curvature of the Dubins curve;

[0011] Step 3: Use the improved decoy bird foraging algorithm to optimize the UAV deployment on the smoothed path; design a fitness function containing multiple factors and use a heuristic search algorithm to efficiently deploy UAVs.

[0012] Furthermore, in step 3, the improved decoy bird foraging algorithm is used to optimize the UAV deployment on the smoothed path, including dividing the UAVs into two groups, primary UAVs and auxiliary UAVs, and dynamically adjusting the grouping to ensure load balancing.

[0013] Furthermore, in step 3, temporary UAVs are deployed to meet the temporary needs of the main UAVs, and the task allocation and execution times of the temporary UAVs are dynamically adjusted according to the number of uninspected locations.

[0014] Furthermore, in step 3, while being efficiently deployed, the drones are allowed to flexibly switch roles as needed during the execution process, from a main drone to an auxiliary drone or from an auxiliary drone to a temporary drone, in order to cope with emergencies and mission changes.

[0015] Furthermore, step 1 also includes: constructing an energy consumption model of the UAV and using it as an energy constraint condition for path planning to ensure that the UAV does not exceed its energy limit during actual flight.

[0016] Furthermore, step 1 also includes: converting the path planning problem into a multiple traveling salesman problem mTSP, and solving it in combination with a dynamic programming method to generate a specific flight path for each drone.

[0017] Furthermore, step 2 also includes: designing a supervised learning model, taking the initial and final postures of the path as input, minimizing the loss function through training, and obtaining the optimal curve smoothing rate to ensure that the path is both smooth and in line with the actual flight capabilities of the drone.

[0018] Furthermore, the multiple factors in step 3 include: distance, number of defects, and energy consumption.

[0019] Furthermore, the heuristic search algorithm in step 3 is a genetic algorithm or a particle swarm optimization algorithm.

[0020] Furthermore, step 3 also includes: clustering the map using a K-means algorithm, determining an optimal number of clusters, and dividing the map into a plurality of non-overlapping areas.

[0021] The beneficial effects of the present invention are:

[0022] (1) The multi-UAV path planning model is constructed by mathematical modeling, which greatly simplifies the complex calculation process and effectively reduces the calculation workload. At the same time, the salesman algorithm is introduced and improved, and the energy constraint definition is innovatively added. This measure significantly improves the accuracy of the path planning model and the path tracking efficiency, laying a solid foundation for the efficient collaborative operation of the multi-UAV system.

[0023] (2) The present invention adopts an improved Dubins curve technology based on supervised learning. The generated multi-UAV path is not only smoother, but also greatly enhances the continuity and stability of the path, effectively reducing the risk of abrupt turns of the UAV during flight. In addition, this improvement also improves the efficiency of the algorithm, reduces unnecessary calculations, and further speeds up the search speed, so that the multi-UAV system can respond faster and optimize the path planning solution.

[0024] (3) This invention innovatively combines a bio-inspired multi-UAV path planning method, flexibly configures the UAV team according to the number and function of the UAVs, and executes the task with the optimal strategy. This method not only has low energy consumption, but also can efficiently discover potential defects in a shorter journey and shorter time during the inspection of transmission lines. This breakthrough has greatly improved the operating efficiency and detection accuracy of multi-UAV systems in complex environments, and provided strong technical support for the safe maintenance of key infrastructure such as electricity and communications. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0026] Figure 1 It is a schematic flow chart of the autonomous path planning method for multi-UAV power transmission line inspection of the present invention. DETAILED DESCRIPTION

[0027] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0028] This invention proposes a multi-UAV power transmission line inspection autonomous path planning method. This method first uses an improved salesman algorithm to construct an inspection path planning model to cope with the performance differences and flight restrictions of multiple UAVs. Then, the path is smoothed by combining the Dubins curve technology based on supervised learning to ensure stable and efficient UAV flight. Finally, the inspection tasks are intelligently allocated through the improved decoy bird foraging algorithm, and the inspection areas and logic of the main, auxiliary and temporary UAVs are clarified, so that the calculated path planning model is deployed by multiple UAVs to achieve efficient collaborative operations. This series of innovative measures jointly improves inspection efficiency, reduces operating costs, and effectively ensures the safety of the power grid.

[0029] The specific steps of this method can be as follows:

[0030] Step 1: Build and optimize the multi-UAV energy-constrained path planning model.

[0031] The present invention aims at the path planning problem of multiple UAVs cooperating in power transmission line inspection and carefully constructs an efficient path planning model. In this model, the UAV will autonomously generate a series of accurate three-dimensional waypoints when performing inspection tasks. These waypoints not only calibrate the flight trajectory of the UAV, but also store them in the form of vectors for subsequent calculation and processing. u is stored as a vector, the specific formula can be expressed as Among them, w u，i =(x u，i ,y u，i , z u，i ) represents the i-th waypoint of UAV u, N u，wRepresents the number of waypoints in the feasible trajectory of the drone.

[0032] In order to achieve high efficiency and high quality of multi-UAV collaborative inspection, the core goal of the present invention is to minimize the total motion path of all UAVs. To this end, the path planning problem is cleverly transformed into a variant of the classic salesman problem (TSP), and targeted improvements are made on this basis. By introducing the solution framework of the salesman problem, the path planning of the UAV can be transformed into the problem of finding the shortest closed path, thereby effectively reducing the flight distance and time of the UAV.

[0033] However, the traditional salesman problem does not consider the actual constraint of the energy limit of the UAV. In order to make up for this deficiency, the present invention specifically adds energy constraints to the path planning model. The energy constraint is defined as the UAV u cannot choose a path that exceeds its energy δ u The specific formula for minimizing the UAV motion path can be expressed as

[0034]

[0035] Among them, U is the total number of drones, r i，j is the energy consumption of UAV i to the jth waypoint, represents the total driving distance corresponding to UAV i. The specific formula can be expressed as Among them, D i，j Represents the distance of the UAV from the i-th waypoint to the j-th waypoint.

[0036] Through mathematical modeling and algorithm optimization, we finally obtained a multi-UAV transmission line inspection path planning model that not only meets actual flight conditions but also has high efficiency. This model not only greatly simplifies the computational complexity of multi-UAV path planning, but also improves the accuracy and practicality of path planning by introducing energy constraints. In practical applications, this model can significantly improve the inspection efficiency and collaborative operation capabilities of UAVs, providing strong guarantees for the safe and stable operation of transmission lines.

[0037] Step 2: Path smoothing optimization based on supervised learning and improved Dubins curve.

[0038] Based on the multi-UAV path planning model constructed in step 1, the present invention further adopts an improved Dubins curve for path smoothing. First, the trajectory points generated by the path planning are carefully sorted to ensure that they are traversed in the best order to train a complete and efficient flight trajectory. When selecting the trajectory sequence of the next segment i+1, special consideration is given to the two trajectory points from the current position of the UAV to the near end of the segment, and taking this as the starting point, the remaining trajectory points are covered and connected in sequence.

[0039] When connecting adjacent waypoints to generate a complete and continuous flight path, the present invention innovatively introduces an improved Dubins curve. While receiving the initial and final postures as input variables, the curve also takes into account the maximum curvature ζ limit of the drone, so that the smoothest path segment can be generated under the premise of ensuring flight safety. By utilizing the dual symmetry of path planning, that is, the symmetrical relationship between the starting point and the end point, the present application cleverly maps this smoothing problem to a multi-class classification situation in a supervised learning environment. In this environment, input parameters (such as initial and final postures, curvature limits, etc.) are used as features, and a function that can minimize the expected loss is found through a training process to achieve the best curve smoothness.

[0040] In the framework of supervised learning, the core task of the present invention is to solve a loss function that can accurately reflect the smoothness of the curve. k ,y k , z k} is N u，w For a dataset of samples, using the maximum likelihood principle, assuming that the samples in D are independent and identically distributed, the goal of the path planning supervised learning task formed by the improved Dubins curve is to train the function f that minimizes the expected loss. The specific formula can be expressed as

[0041]

[0042] Among them, w i is the weight coefficient, a i，j is the training function for supervised learning, f = a i，j .

[0043] After the above processing, the UAV path planning result after the improved Dubins curve smoothing is finally obtained. This result not only significantly improves the continuity and smoothness of the path, reduces the probability of abrupt turns of the UAV during flight, but also further improves the overall performance of the algorithm by reducing unnecessary calculations and optimizing search efficiency. Therefore, the improved Dubins curve smoothing method based on supervised learning adopted in this step is of great significance for improving the execution efficiency and safety of multi-UAV power transmission line inspection tasks.

[0044] Step 3: Intelligent deployment of multiple UAVs based on improved decoy bird foraging algorithm.

[0045] After the path planning is completed, the present invention further uses an innovative decoy bird foraging algorithm to intelligently deploy multiple drones to optimize the execution of inspection tasks. First, the central control algorithm uses the K-means clustering algorithm to accurately divide the inspection map into a predetermined number of non-overlapping areas to ensure that each area can be inspected independently. Then, according to a specific formula or strategy, the drones are divided into two groups: main drones and auxiliary drones, each of which undertakes different inspection tasks. The specific formula can be expressed as

[0046] Step 3.1 Deployment of primary drones: Each primary drone is explicitly assigned to a non-overlapping available area, and no multiple primary drones can be assigned to the area to ensure the uniqueness of task assignment. Within the specified area, the primary drone will randomly select a starting position based on the first inspection, or select an optimal position based on the existing information. In addition, the primary drone has the authority to send a joining request to the auxiliary drone for additional support. If the remaining uninspected locations in the current area exceed the preset threshold, the primary drone will request a certain number of temporary drones to join, and the number is calculated by the minimum-maximum normalization formula, which comprehensively considers the area size and the total number of locations to ensure the reasonable allocation of resources.

[0047] Specifically, the primary drone will select a location within the specified area (randomly if it is the first time to check the area). In addition, each primary drone is allowed to send a request to the auxiliary drone to join the inspection process of the specified area. Even if the request has been met, the primary drone cannot send more requests to the auxiliary drone. Once the current position of the primary drone is checked, the drone will check whether the entire status of the area has been checked. If the area is still not checked, it will check whether the remaining positions are greater than a certain threshold. If so, the primary drone will request temporary drones. Each primary drone is allowed to request a certain number of temporary drones. The number of required temporary drones is calculated based on the size of the area allocated to the primary drone. The minimum-maximum normalization is used to normalize the total number of positions in the specified area within the range [1, U]. The specific formula can be expressed as Among them, Z total is the total number of positions in the specified partition, minvalueofZ total and maxvalueofZ total are the sizes of the minimum and maximum regions, Z′ total is a normalized value of the total number of locations within the specified region in the range [1, U].

[0048] Step 3.2 Deployment of auxiliary drones: The auxiliary drones are assigned tasks by the central control algorithm in the initial stage, but they will be dynamically adjusted according to the actual situation during the execution process. When receiving a joining request from the main drone, the auxiliary drone will evaluate the fitness value of each request, which is calculated by comprehensively considering factors such as distance and number of defects, and assign corresponding weights to different components to reflect their importance. Auxiliary drones tend to choose the main drones with close distance and large number of defects for support to improve inspection efficiency. After completing the inspection of the current area, the auxiliary drone will switch roles according to the regional status and overall needs, such as becoming a main drone or a temporary drone.

[0049] Specifically, the auxiliary UAVs are initially assigned by the central control algorithm. Once they start executing, they will check the number of joining requests from the primary UAV. If there are multiple requests, they will evaluate these requests and select one that maximizes the fitness value. Assume that the auxiliary UAV is located at position i, the primary UAV that joins the request is located at position j, and the number of transmission line defects detected in the area corresponding to the primary UAV is D. num , the distance between positions i and j is given by Cost i，j Given, the formula for calculating the fitness value of each available join request can be expressed as req fitness =w1(D num )-w2(Cost i，j ), where w i (i=1, 2) is the weight of each component, reflecting the essential differences when evaluating candidate requests. When calculating the weights of the fitness function components, the more important terms in the fitness calculation are given greater weights, and the auxiliary drone seeks to select the main drone within a shorter distance (minimization goal) while looking for the main drone that detects more defects (maximization goal). Once the inspection algorithm checks the current position of the auxiliary drone, it will check whether the entire area status has been checked. If the area is still not checked, the auxiliary drone will remain in the current area for further position inspections. However, if the area status is checked, the auxiliary drone will select the applicable option from the following: 1. Evaluate the joining request (if any) and then select the request with the highest fitness value; 2. If there is an available area, change the role to the main drone; 3. If there is a temporary drone request, change the role to a temporary drone.

[0050] Step 3.3 Deployment of temporary drones: Temporary drones are not assigned specific tasks during the initialization phase, but can switch roles and receive tasks at any time according to actual needs. The number of times its tasks are executed is dynamically calculated based on the number of uninspected locations in the designated area of ​​the main drone, ensuring that sufficient resources can be quickly mobilized when the inspection task is heavy. Once the task is started, the temporary drone will receive the task request in FIFO mode and go to the location of the requested main drone for support inspection. During the inspection process, the temporary drone will record the inspection data in real time and upload it to the central control system.

[0051] Specifically, temporary drones are not assigned during the initialization phase. In addition, any primary or secondary drone that is not needed during any execution phase can change its role to a temporary drone and meet the demand for temporary drones. The number of rounds that the temporary drone will execute is calculated based on the number of unchecked locations in the area specified by the primary drone. This value is calculated once for each temporary drone mission. The more unchecked locations there are, the more likely it is that more temporary drones will be assigned. The specific formula can be expressed as

[0052]

[0053] Among them, Z uninspected is the number of unchecked locations within the specified area, minvalueofZ uninspected and maxvalueofZ uninspected are zero and the total number of positions in the region, Z′ uninspected is the normalized value of the number of unchecked locations in the specified area within the range [1, U]. Once the temporary drone starts executing, it will receive the first temporary request in a FIFO manner. Then, it will find the nearest location of the primary drone that requested the temporary drone and start checking the area. Once the inspection algorithm checks the current location of the temporary drone, the temporary drone will check whether the maximum sub-area has been reached. The final result is the multi-drone inspection result after deployment using the improved decoy bird foraging algorithm.

[0054] By deploying an improved decoy bird foraging algorithm, multiple drones can efficiently and collaboratively complete inspection tasks. This method not only improves inspection efficiency, but also reduces energy consumption, allowing drones to find defects in transmission lines in a shorter time and with a shorter journey. At the same time, the algorithm also fully considers the configuration of drones with different numbers and functions, making the entire inspection system more flexible and adaptable.

[0055] The beneficial effects of the present invention are:

[0056] (1) Significantly improve the efficiency and accuracy of path planning: This invention constructs a path planning model for multiple UAVs through innovative mathematical modeling, effectively simplifies the complex calculation process, and significantly reduces the calculation workload of path planning. At the same time, the salesman algorithm is introduced and improved, and the energy constraint definition of multiple UAVs is added. This innovation not only improves the accuracy of path planning modeling, but also greatly improves the tracking efficiency of path planning, ensuring that UAVs can fly more accurately according to the planned path when performing tasks.

[0057] (2) Optimizing path smoothness and stability: This patent uses an improved Dubins curve technology based on supervised learning. This technological innovation makes the generated multi-UAV paths smoother and more fluent, greatly improving the continuity and stability of the paths. By reducing the possibility of abrupt turns, it not only improves the stability and safety of UAV flight, but also further optimizes flight efficiency and reduces energy consumption. In addition, the application of this technology also significantly improves the execution efficiency of the algorithm and reduces unnecessary calculations, thereby speeding up the search speed and enhancing the responsiveness of the overall system.

[0058] (3) Enhance the efficiency of multi-UAV collaborative inspection: This invention innovatively combines the bio-inspired multi-UAV path planning method, and achieves efficient collaborative operation by flexibly configuring UAVs of different numbers and functions. This method not only greatly improves the efficiency of multi-UAV configuration, but also completes the comprehensive inspection of transmission lines with shorter flight distances and shorter time, significantly improves the ability to detect defects, reduces inspection costs, and provides a strong guarantee for the safe and stable operation of transmission lines. At the same time, due to the excellent performance of this method in energy consumption control, it further reduces operating costs, extends the operation time of UAVs, and enhances the overall economy and sustainability of the system.

[0059] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An autonomous path planning method for multi-UAV power transmission line inspection, It is characterized in that include: Step 1, digitizing the transmission line map, including accurate input of height, obstacles and transmission line location information; Construct a multi-objective optimization path planning model to minimize the total motion path while considering multiple optimization objectives such as inspection time, inspection efficiency, and UAV load balance; Step 2: Use the improved Dubins curve to smooth the preliminary path output by the path planning model, and generate a continuous and smooth path that meets the flight capabilities of the UAV by sorting the trajectory points and dynamically adjusting the maximum curvature of the Dubins curve; Step 3: Use the improved decoy bird foraging algorithm to optimize the UAV deployment on the smoothed path; design a fitness function containing multiple factors and use a heuristic search algorithm to efficiently deploy UAVs.

2. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that In step 3, the improved decoy bird foraging algorithm is used to optimize the UAV deployment on the smoothed path, including dividing the UAVs into two groups: main UAVs and auxiliary UAVs, and dynamically adjusting the grouping to ensure load balancing.

3. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 2, It is characterized in that In step 3, temporary drones are deployed to meet the temporary needs of the main drones, and the task allocation and execution times of the temporary drones are dynamically adjusted according to the number of uninspected locations.

4. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 3, It is characterized in that In step 3, while being efficiently deployed, the drone is allowed to flexibly switch roles as needed during the execution process, from a main drone to an auxiliary drone or from an auxiliary drone to a temporary drone, so as to cope with emergencies and mission changes.

5. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that The step 1 also includes: constructing an energy consumption model of the UAV and using it as an energy constraint condition for path planning to ensure that the UAV does not exceed its energy limit during actual flight.

6. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that Step 1 also includes: converting the path planning problem into a multiple traveling salesman problem mTSP, and solving it in combination with a dynamic programming method to generate a specific flight path for each UAV.

7. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that The step 2 also includes: designing a supervised learning model, taking the initial and final postures of the path as input, and obtaining the optimal curve smoothing rate by training to minimize the loss function, thereby ensuring that the path is both smooth and in line with the actual flight capabilities of the drone.

8. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that The multiple factors in step 3 include: distance, number of defects, and energy consumption.

9. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that The heuristic search algorithm in step 3 is a genetic algorithm or a particle swarm optimization algorithm.

10. The autonomous path planning method for multi-UAV power transmission line inspection as claimed in claim 1, It is characterized in that The step 3 also includes: using the K-means algorithm to cluster the map, determine the optimal number of clusters, and divide the map into multiple non-overlapping areas.

Images