Method and device for polling photovoltaic panel by unmanned aerial vehicle

Abstract

The invention provides a method and a device for inspecting a photovoltaic panel by an unmanned aerial vehicle, and the method for inspecting the photovoltaic panel by the unmanned aerial vehicle comprises the steps: S1, obtaining a planar grid map which comprises a plurality of first grids corresponding to the photovoltaic panel; and S2, a particle swarm is generated based on a particle swarm optimization algorithm, the particle swarm comprises N particles, each particle traverses all the first grids, N traversal paths connected with all the first grids are obtained respectively, and the lengths of the N traversal paths are compared, so that the shortest path is obtained, the unmanned aerial vehicle performs routing inspection based on the shortest path, and N is greater than or equal to 2. According to the method for polling the photovoltaic panel by the unmanned aerial vehicle, the polling efficiency of the photovoltaic panel by the unmanned aerial vehicle is high, and the coverage rate is high.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaics, and particularly to a method and device for a drone to inspect photovoltaic panels. Background Art

[0002] As a key component in the field of renewable energy, the installed capacity of photovoltaic technology has been continuously increasing. Along with this trend, the maintenance and monitoring of photovoltaic power stations have become important tasks to ensure the efficient operation of the power stations. The drone photovoltaic panel inspection technology has been widely used in the regular inspection and monitoring of photovoltaic panels due to its high efficiency and practicality, which helps to detect and solve problems in a timely manner, and ensure the normal operation and power generation efficiency of the photovoltaic power generation system.

[0003] However, the current drone photovoltaic panel inspection technology has an unreasonable path planning situation. Path planning is the core technology of drone inspection, which directly affects the inspection efficiency and coverage effect. Summary of the Invention

[0004] Aiming at the above problems of the prior art, the purpose of the present invention is to provide a method and device for a drone to inspect photovoltaic panels, which can achieve high inspection efficiency and high coverage rate of the drone for photovoltaic panels.

[0005] To solve the above problems, the present invention provides a method for a drone to inspect photovoltaic panels, the method comprising:

[0006] Step S1, obtaining a planar grid map, the planar grid map including a plurality of first grids corresponding to photovoltaic panels;

[0007] Step S2, generating a particle population based on the particle swarm optimization algorithm, the particle population including N particles, each particle traversing all the first grids to obtain N traversal paths connecting all the first grids respectively, comparing the lengths of the N traversal paths, so as to obtain the shortest path, such that the drone performs inspection based on the shortest path, where N≥2.

[0008] Further, the step S1 includes:

[0009] Step S11, obtaining a real shot map and a satellite map, the real shot map being obtained by the drone's patrol shooting of the photovoltaic panels;

[0010] Step S12, comparing the real shot map and the satellite map to identify the boundaries and positions of the photovoltaic panels;

[0011] Step S13, converting the real shot map into a grid map, the grid map including the first grids, fusing the grid map with the satellite map to generate the planar grid map, the planar grid map further including the terrain and buildings around the photovoltaic panels.

[0012] Furthermore, the step S12 includes:

[0013] Identifying the boundary and position of the photovoltaic panel based on an image recognition algorithm and an edge detection algorithm.

[0014] Furthermore, the step S2 includes:

[0015] Step S21, obtaining the starting point grid in the first grid;

[0016] Step S22, generating the particle population based on the particle swarm optimization algorithm;

[0017] Step S23, obtaining the initial velocity of each particle, and allowing each particle to gradually traverse all the first grids starting from the starting point grid;

[0018] Step S24, after N particles traverse all the first grids, N traversal paths are respectively formed, and by comparing the lengths of the N traversal paths, the shortest path is obtained.

[0019] Furthermore, the step S23 includes:

[0020] Calculating the path for the particle to reach the next first grid to be traversed from the current first grid based on a heuristic function, and allowing the particle to traverse according to this path.

[0021] Furthermore, the heuristic function includes:

[0022] F(n,g) = w1·D(n,g) + w2·x i

[0023]

[0024] where F(n,g) is the heuristic function, D(n,g) is the estimated distance cost, x i is the priority function, w1 is the first weight value, w2 is the second weight value, (x n ,y n ) and (x g ,y g ) are the coordinates of the current first grid n and the next first grid g respectively.

[0025] Furthermore, the step S2 further includes:

[0026] Step S25, calculating the fitness value of the particle based on a fitness function and performing iteration, the iteration includes the particle traversing all the first grids once, and the fitness function is:

[0027]

[0028] Among them, n1, n2, … n k represents the first grid traversed by the particle in sequence, where n1 is the starting grid and n k is the ending grid, d(n i , n i+1 ) The Euclidean distance between every two adjacent first grids n i and n i+1 If two first grids are adjacent horizontally or vertically, it is 1. If two first grids are adjacent diagonally, it is

[0029] Step S26, update the velocity and position of the particle;

[0030] Step S27, when the termination iteration condition is reached, output the shortest path based on the lowest fitness value in all traversal paths, so that the drone performs inspection based on the shortest path. The termination iteration condition includes that the iteration reaches a predetermined number of iterations or a predetermined iteration time.

[0031] Further, the planar grid map further includes a second grid corresponding to an obstacle. When the particle needs to pass through the second grid to reach the next first grid, the second grid is avoided based on the gravitational potential field function.

[0032] Further, the method further includes:

[0033] Step S3, control the drone to perform inspection on the photovoltaic panel according to the shortest path and collect real-time images, and analyze the real-time images. The real-time images include visible light images and infrared images.

[0034] The present invention further provides a device for a drone to inspect a photovoltaic panel. The device includes:

[0035] An acquisition module, which is used to acquire a planar grid map. The planar grid map includes a plurality of first grids corresponding to photovoltaic panels;

[0036] A path planning module, which is used to generate a particle population based on the particle swarm optimization algorithm. The particle population includes N particles. Each particle traverses all the first grids and respectively obtains N traversal paths connecting all the first grids. By comparing the lengths of the N traversal paths, the shortest path is obtained, so that the drone performs inspection based on the shortest path, where N≥2.

[0037] Due to the above technical solutions, the present invention has the following beneficial effects:

[0038] The method for inspecting a photovoltaic panel by a drone according to the present invention generates a planar grid map including a first grid corresponding to the photovoltaic panel, generates a particle population including N particles based on a particle swarm optimization algorithm, allows each particle to traverse all the first grids to form N traversal paths, and compares the N paths to obtain the shortest path for the drone to perform the inspection. Thus, the shortest path for the drone to perform the inspection can be obtained through an intelligent algorithm, the path planning is relatively reasonable, the inspection efficiency and coverage rate of the drone for the photovoltaic panel are high, and automatic inspection of the photovoltaic panel can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the present invention, 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 invention, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts.

[0040] Figure 1 is a flowchart of a method for inspecting a photovoltaic panel by a drone according to an embodiment of the present invention;

[0041] Figure 2 is a planar grid map according to an embodiment of the present invention;

[0042] Figure 3 is a schematic diagram of the traversal path of a particle according to an embodiment of the present invention;

[0043] Figure 4 is a schematic diagram of a device for inspecting a photovoltaic panel by a drone according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0045] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present invention are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion.

[0046] Next, a method for a drone of an embodiment of the present invention to inspect a photovoltaic panel will be described.

[0047] As Figures 1 to 3 shown, the method for a drone of an embodiment of the present invention to inspect a photovoltaic panel includes:

[0048] Step S1, obtaining a planar grid map, where the planar grid map includes a plurality of first grids corresponding to photovoltaic panels.

[0049] For example, a planar grid map as Figure 2 shown is formed, and the gray grids are the first grids, that is, the grids corresponding to the photovoltaic panels.

[0050] Step S2, generating a particle swarm based on the particle swarm optimization algorithm. The particle swarm includes N particles. Each particle traverses all the first grids to obtain N traversal paths connecting all the first grids, and compares the lengths of the N traversal paths to obtain the shortest path, so that the drone performs inspection based on the shortest path, where N≥2.

[0051] For example, 100 particles are generated, and each particle traverses all the first grids, so there are 100 paths. The lengths of the 100 paths are compared to obtain the shortest path (for example, the shortest path as Figure 3 shown). The drone performs inspection based on the shortest path, so that the drone can efficiently inspect the photovoltaic panels. Among them, the specific particle swarm optimization algorithm is a prior art. A general particle swarm optimization algorithm can be used, or the specific algorithm below can also be used.

[0052] In the above method for a drone to inspect a photovoltaic panel, a planar grid map including the first grids corresponding to the photovoltaic panels is generated, a particle swarm including N particles is generated based on the particle swarm optimization algorithm, each particle traverses all the first grids to form N traversal paths, and the N paths are compared to obtain the shortest path for the drone to perform inspection. Thus, the shortest path for the drone to perform inspection can be obtained through an intelligent algorithm, the path planning is relatively reasonable, the inspection efficiency and coverage rate of the drone for the photovoltaic panels are high, and automatic inspection of the photovoltaic panels can be realized.

[0053] For a planar grid map, it can be generated through the following method for acquisition:

[0054] Step S1 includes: Step S11, obtaining a real-shot image and a satellite map, where the real-shot image is obtained by the drone's patrol shooting of the photovoltaic panels. Step S12, comparing the real-shot image and the satellite map to identify the boundaries and positions of the photovoltaic panels. Step S13, converting the real-shot image into a grid map. The grid map includes the first grid. The grid map is fused with the satellite map to obtain a planar grid map, and the planar grid map also includes the terrain and buildings around the photovoltaic panels.

[0055] Obtain the specific real-shot image of the photovoltaic panels through the real-shot image, obtain the corresponding satellite map (satellite image or online map (such as Amap, Baidu Map, etc.)) according to the positions of the photovoltaic panels, compare the real-shot image and the satellite map to identify the boundaries and positions of the photovoltaic panels, perform distortion removal and contrast adjustment processing on the real-shot image, and use image registration technology to align and register the real-shot image and the satellite map plan in the same coordinate system. Convert the real-shot image into a grid map including the first grid, and fuse the grid map with the satellite map to generate a planar grid map that also includes the terrain and buildings around the photovoltaic panels. Thus, an accurate and comprehensive planar grid map can be generated, and the influence of the terrain and buildings around the photovoltaic panels on the drone flight can be considered.

[0056] Furthermore, the boundaries and positions of the photovoltaic panels are identified based on an image recognition algorithm and an edge detection algorithm.

[0057] Thus, the boundaries and positions of the photovoltaic panels can be accurately identified. Among them, the image recognition algorithm and the edge detection algorithm are known technologies and will not be elaborated here.

[0058] In some embodiments of the present invention, Step S2 includes: Step S21, obtaining the starting point grid in the first grid. Step S22, generating a particle population based on the particle swarm optimization algorithm. Step S23, obtaining the initial velocity of each particle, and letting each particle gradually traverse all the first grids starting from the starting point grid. Step S24, after N particles traverse all the first grids, N traversal paths are respectively formed. Compare the lengths of the N traversal paths to obtain the shortest path.

[0059] For example, randomly generate a particle population, and the particle population includes 50 particles. A real number vector x with a dimension of 126 can be defined as a particle (decision variable). The value range of each element in this particle is [0,1]. Each element of the particle represents the priority of the corresponding first grid being selected in the traversal path. The smaller the value, the higher the priority of the first grid in the traversal order. Before the start of the algorithm loop, each element in the particle is initialized to a random number in [0,1].

[0060] Set the initial velocity of the particles, and let N particles traverse all the first grids step by step from the starting grid with the initial velocity, obtain and compare N traversal paths, so as to obtain the shortest path.

[0061] Further, step S23 includes calculating the path for the particle to reach the next first grid to be traversed from the current first grid based on the heuristic function, and enabling the particle to traverse according to this path.

[0062] The heuristic function can select a traversal path with a smaller estimated distance cost from the current first grid to the next first grid.

[0063] Specifically, the heuristic function includes:

[0064] F(n,g) = w1·D(n,g) + w2·x i

[0065]

[0066] Among them, F(n,g) is the heuristic function, D(n,g) is the estimated distance cost, x i is the priority function, w1 is the first weight value, w2 is the second weight value, (x n ,y n ) and (x g ,y g ) are the coordinates of the current first grid n and the next first grid g respectively.

[0067] Adding the priority function to the heuristic function can improve the diversity and accuracy of the algorithm, and the priority function and the estimated distance cost can be adjusted through the first weight and the second weight. Moreover, pruning operations can be performed on the redundancy in the traversal path to further optimize the traversal path.

[0068] Further, step S2 also includes: step S25, calculating the fitness value of the particle based on the fitness function and performing iteration. The iteration includes the particle traversing all the first grids once. The fitness function is:

[0069]

[0070] Among them, n1,n2,…n k represents the first grids traversed by the particle in sequence, where n1 is the starting grid and n k is the ending grid, d(n i ,n i+1 ) is the Euclidean distance between every two adjacent first grids n i and n i+1 . If two first grids are horizontally or vertically adjacent, it is 1. If two first grids are diagonally adjacent, it is Step S26, update the velocity and position of the particle. Step S27, when the termination iteration condition is reached, output the shortest path based on the lowest fitness value among all traversed paths, so that the drone performs inspection based on the shortest path. The termination iteration condition includes that the iteration reaches a predetermined number of iterations or a predetermined iteration time.

[0071] The fitness value aims to optimize the shortest traversed path. Calculating the fitness value can better evaluate and compare each traversed path. Let the particle traverse all the first grids multiple times, that is, each time the first grid is traversed, it is iterated once, and the fitness value will gradually decrease. The velocity and position of the particle are updated each time an iteration is performed. When iterating to the termination iteration condition (reaching the predetermined number of iterations or the predetermined iteration time), the iteration is terminated, and the shortest path with the lowest fitness is output. Thereby, the accuracy and comprehensiveness of the traversed path can be improved, the path planning efficiency can be enhanced, and the operation and maintenance efficiency and the reliability of the system can be effectively improved.

[0072] In some embodiments of the present invention, the planar grid map further includes a second grid corresponding to the obstacle. When the particle needs to pass through the second grid to reach the next first grid, the second grid is avoided based on the gravitational potential field function. Among them, the gravitational potential field function is a known technology and will not be elaborated here.

[0073] For example, there are relatively high columns, mounds, large garbage, etc. in the area of the photovoltaic panels that may be touched by the drone during inspection. Such obstacles are marked as the second grid, and such obstacles are avoided based on the gravitational potential field function. The obstacle avoidance path is short, and efficient obstacle avoidance can be achieved.

[0074] In some embodiments of the present invention, the method further includes step S3, controlling the drone to inspect the photovoltaic panels according to the shortest path and collect real-time images, and analyzing the real-time images. The real-time images include visible light images and infrared images.

[0075] The drone is equipped with a dual camera (visible light camera and infrared camera) to inspect the photovoltaic panels according to the shortest path, and the collected data information is transmitted in real time for analysis to identify and warn potential fault points, realizing automatic and efficient inspection of the photovoltaic panels.

[0076] Next, the device 1000 for the drone to inspect the photovoltaic panels according to the embodiments of the present invention will be described.

[0077] As Figure 4 shown, the device for the drone to inspect the photovoltaic panels includes an acquisition module 1001 and a path planning module 1002.

[0078] The acquisition module 1001 is used to acquire a planar grid map, and the planar grid map includes a plurality of first grids corresponding to photovoltaic panels. The path planning module 1002 is used to generate a particle population based on the particle swarm optimization algorithm. The particle population includes N particles, and each particle traverses all the first grids to obtain N traversal paths connecting all the first grids respectively. By comparing the lengths of the N traversal paths, the shortest path is obtained, so that the drone can perform inspection based on the shortest path, where N≥2.

[0079] Since the method for the drone to inspect the photovoltaic panels according to the embodiment of the present invention has the above technical effects, therefore, the device for the drone to inspect the photovoltaic panels according to the embodiment of the invention also has corresponding technical effects.

[0080] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for an unmanned aerial vehicle to inspect a photovoltaic panel, characterized in that, The method includes: Step S1, obtaining a planar grid map, where the planar grid map includes a plurality of first grids corresponding to photovoltaic panels; Step S2, generating a particle population based on the particle swarm optimization algorithm, where the particle population includes N particles, each particle traverses all the first grids, respectively obtaining N traversal paths connecting all the first grids, comparing the lengths of the N traversal paths, so as to obtain the shortest path, such that the drone performs inspection based on the shortest path, where N≥2.

2. The method for a drone to inspect a photovoltaic panel according to claim 1, wherein The step S1 includes: Step S11, obtaining a real-shot image and a satellite map, where the real-shot image is obtained by the drone's inspection shooting of the photovoltaic panels; Step S12, comparing the real-shot image and the satellite map, and identifying the boundaries and positions of the photovoltaic panels; Step S13, converting the real-shot image into a grid map, where the grid map includes the first grids, fusing the grid map with the satellite map to generate the planar grid map, and the planar grid map also includes the terrain and buildings around the photovoltaic panels.

3. The method for the drone to inspect the photovoltaic panels according to claim 2, wherein, The step S12 includes: Identifying the boundaries and positions of the photovoltaic panels based on an image recognition algorithm and an edge detection algorithm.

4. The method for the drone to inspect the photovoltaic panels according to claim 1, characterized in that, The step S2 includes: Step S21, obtaining a starting point grid in the first grids; Step S22, generating the particle population based on the particle swarm optimization algorithm; Step S23, obtaining the initial velocity of each particle, and allowing each particle to gradually traverse all the first grids starting from the starting point grid; Step S24, after the N particles traverse all the first grids, respectively forming N traversal paths, comparing the lengths of the N traversal paths to obtain the shortest path.

5. The method for the drone to inspect the photovoltaic panels according to claim 4, characterized in that, The step S23 includes: Calculating the path for the particle to reach the next first grid to be traversed from the current first grid based on a heuristic function, and allowing the particle to traverse according to this path.

6. The method for an unmanned aerial vehicle to inspect a photovoltaic panel according to claim 5, characterized in that, The heuristic function includes: F(n,g) = w1·D(n,g) + w2·x i Among them, F(n, g) is the heuristic function, D(n, g) is the estimated distance cost, and x i is the priority function, w1 is the first weight value, w2 is the second weight value, and (x n , y n ) and (x g , y g ) are the coordinates of the current first grid n and the next first grid g respectively.

7. The method for the drone to inspect the photovoltaic panels according to claim 5, characterized in that, The step S2 further includes: Step S25, calculating the fitness value of the particle based on a fitness function and performing iteration, where the iteration includes the particle traversing all the first grids once, and the fitness function is: Among them, n1, n2, …, n k represent the first grids traversed successively by the said particle, where n1 is the starting point grid and n k is the ending point grid, and d(n i , n i+1 ) is the Euclidean distance between every two adjacent said first grids n i and n i+1 . If two said first grids are horizontally or vertically adjacent, it is 1; if two said first grids are diagonally adjacent, it is Step S26, updating the velocity and position of the particle; Step S27, when the termination iteration condition is reached, outputting the shortest path based on the lowest fitness value among all the traversal paths, such that the drone performs inspection based on the shortest path, and the termination iteration condition includes that the iteration reaches a predetermined number of iterations or a predetermined iteration time.

8. The method for the drone to inspect the photovoltaic panels according to claim 7, characterized in that, The planar grid map further includes second grids corresponding to obstacles. When the particle needs to pass through the second grids to reach the next first grid, the second grids are avoided based on a gravitational potential field function.

9. The method for an unmanned aerial vehicle to inspect a photovoltaic panel according to claim 1, characterized in that, The method further includes: Step S3, controlling the drone to perform inspection on the photovoltaic panels according to the shortest path and collect real-time images, and analyzing the real-time images, where the real-time images include visible light images and infrared images.

10. A device for an unmanned aerial vehicle to inspect a photovoltaic panel, characterized in that, The device includes: An acquisition module, the acquisition module is used to acquire a planar grid map, and the planar grid map includes a plurality of first grids corresponding to photovoltaic panels; A path planning module, the path planning module is used to generate a particle population based on the particle swarm optimization algorithm, the particle population includes N particles, each of the particles traverses all the first grids, respectively obtaining N traversal paths connecting all the first grids, comparing the lengths of the N traversal paths, so as to obtain the shortest path, so that the drone performs inspection based on the shortest path, where N≥2.

Images