A multi-unmanned aerial vehicle nest cooperative site selection method for forest area inspection

Abstract

The application discloses a multi-unmanned aerial vehicle nest cooperative site selection method for forest area inspection, acquires geographic information data and carries out data preprocessing; extracts main ridge lines of a forest area, preliminarily preselects multiple unmanned aerial vehicle nest candidate points by comprehensively considering three factors of geographic environment, infrastructure and safety guarantee; a multi-target optimization site selection model is constructed by comprehensively considering unmanned aerial vehicle electric quantity, coverage rate, anti-interference and terrain influence, and target functions and constraint conditions are determined; the improved NSGA-II algorithm is adopted to solve the multi-target optimization site selection model, and an optimal site selection layout scheme of the unmanned aerial vehicle nest is obtained; and the Pareto optimal solution set is calculated based on the fuzzy set theory, optimal indexes of each solution are calculated and sorted, and the solution with the maximum optimal value is the compromise solution of the multi-target problem. The application effectively improves cooperative operation efficiency of the unmanned aerial vehicle group, significantly reduces construction and operation costs of the nest, and improves use efficiency and resource allocation capability of the nest.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of unmanned aerial vehicle nest planning and site selection, and particularly relates to a multi-unmanned aerial vehicle nest cooperative site selection method for forest inspection. BACKGROUND

[0002] Forests, as an important part of the Earth's ecosystem, play a crucial role in water conservation, soil preservation, climate regulation, and biodiversity maintenance. However, with increasing human activities and the impact of global climate change, forests are facing growing threats such as fires, pests, and illegal logging. These problems not only harm forest resources but also disrupt ecological balance. Therefore, timely and effective forest inspection and monitoring have become particularly important.

[0003] Traditional forest inspection methods mainly rely on manual ground patrols and aerial monitoring by fixed-wing aircraft. Although these methods can achieve some level of forest monitoring, they have obvious limitations, such as limited inspection range, significant terrain impact, high cost, and difficulty in achieving real-time monitoring. Especially in complex terrain conditions, the effectiveness of traditional methods is often unsatisfactory.

[0004] In recent years, with the development of unmanned aerial vehicle technology and its widespread application in various fields, unmanned aerial vehicles have become an ideal tool for forest inspection due to their rapid deployment, high flexibility, relatively low cost, and ability to achieve real-time data transmission. In particular, in mountainous forest environments, unmanned aerial vehicles can overcome terrain obstacles and efficiently perform large-area inspection tasks, promptly discovering and responding to forest fires, pest infestations, illegal activities, and other issues. Therefore, using multiple unmanned aerial vehicles for collaborative work to automate forest inspection is gradually becoming a trend to replace traditional inspection methods.

[0005] The efficient operation of an unmanned aerial vehicle inspection system cannot be achieved without reasonable nest deployment. The nest is not only a key hub for unmanned aerial vehicle takeoff and landing but also a main facility for charging and storage. The reasonable selection of nest location directly affects the inspection efficiency and overall performance of the system. Although scholars at home and abroad have conducted a lot of research in the field of unmanned aerial vehicle nest layout optimization and have made some progress, current research mostly focuses on urban environments and less on complex forest terrains. Existing site selection optimization models have limited processing capacity under multiple objectives and constraints, especially in dealing with the impact of terrain on unmanned aerial vehicle flight performance. In addition, existing methods often run independently, without fully considering multi-machine collaboration, resulting in unsatisfactory application results in mountainous forest environments.

[0006] Therefore, further research and improvement are still needed for the site selection optimization problem of forest area inspection unmanned aerial vehicle nests to ensure the efficient operation of unmanned aerial vehicle inspection systems in complex natural environments. SUMMARY

[0007] The application provides a multi-unmanned aerial vehicle nest cooperative site selection method for forest area inspection.

[0008] The application provides a multi-unmanned aerial vehicle nest cooperative site selection method for forest area inspection.

[0009] (1) obtaining geographic information data and performing data preprocessing;

[0010] (2) extracting main ridge lines of the forest area, comprehensively considering three factors of geographic environment, infrastructure and safety guarantee, and preliminarily preselecting multiple unmanned aerial vehicle nest candidate points;

[0011] (3) comprehensively considering unmanned aerial vehicle power, coverage, anti-interference and terrain influence, constructing a multi-objective optimization site selection model, and determining a target function and a constraint condition;

[0012] (4) solving the multi-objective optimization site selection model by using an improved NSGA-II algorithm to obtain an optimal site selection layout scheme of the unmanned aerial vehicle nest;

[0013] (5) calculating a Pareto optimal solution set solved in the step (4) based on a fuzzy set theory, calculating and sorting preferred indexes of each solution, and taking a solution with the maximum preferred value as a compromise solution of the multi-objective problem.

[0014] Further, the geographic information data in the step (1) comprises high-precision DEM data, annual meteorological information, power supply network and signal coverage network.

[0015] Further, the data preprocessing in the step (1) is implemented as follows:

[0016] Z-Score and IQR of the DEM data height value are calculated, deviation degrees of data and average values of a sample set are measured, and the IQR is a difference between upper quartiles and lower quartiles; the Z-Score calculation formula is as follows:

[0017]

[0018] In the formula, h is a single observation height value, mu is a sample height average value, and sigma is a sample height standard deviation.

[0019] For small-range abnormal values and missing values, a bilinear interpolation method is used, that is, four known pixel values around the small-range abnormal values and missing values are used to replace or estimate the values:

[0020]

[0021] In the formula, (lon, lat) represents the horizontal and vertical coordinates of the pixel point, and h(lon, lat) represents the height value of the pixel point (lon, lat); the small range of abnormal values and missing values is the data whose number of pixels affected is less than 5% of the total number of pixels;

[0022] For large range of abnormal values and missing values, use a drone to take a photo in the range, and then use GIS software to generate new high-precision DEM data and replace the original error data;

[0023] Extract the latitude and longitude coordinate points of the forest boundary from the open street map data to form an initial boundary point set P = {P1, P2, …, P n} where P i =(lon i ,lat i ), and use the system function to obtain the convex hull point set C = {C1, C2, …, C m} of the boundary point set, where C i =(lon i ,lat i ), and the point set relationship The regular convex polygon region formed by the convex hull point set is taken as the forest area, and the original DEM data is cut according to the convex hull region; the minimum circumscribed rectangle of the convex hull region is calculated to obtain the coordinates of its four vertices

[0024] Grid the forest area, define the side length of the square grid as L according to the accuracy requirement of the inspection task, and ensure that each grid can reasonably represent the geographical area in the range; divide the DEM data into a series of square grids with side length L, and aggregate the elevation values in each grid to represent the elevation value of the grid.

[0025] Further, the extraction process of the main ridge line of the forest area in step (2) is as follows:

[0026] Define the characteristics of the key grid as the opposite trend of the terrain before and after the start and end of the flat terrain; perform cross-section scanning on the terrain grid in multiple directions to identify the grid that meets the key grid characteristics; the cross-section scanning is performed along the main direction, and the diagonal direction is considered to capture more details; mark and extract all grid points that meet the key grid characteristics; use Kruskal algorithm to find the minimum spanning tree of the extracted key grid point set to connect all key points while ensuring that the total weight of the tree is minimized; analyze the branches in the minimum spanning tree, remove the branches that do not meet the characteristics of the main ridge line, and retain the main path; retain the point set of the main ridge line after the minimum spanning tree processing, MST = {M1, M2, … M t} where M i =(lon ilat i )。

[0027] Further, the preliminary pre-selection of multiple unmanned aerial vehicle nest candidate points in step (2) is implemented as follows:

[0028] The geographic environment factor processes digital elevation data to obtain the height and slope information of each point to evaluate the suitability of the terrain; at the same time, the local annual wind direction and wind speed are analyzed to simulate the ground wind speed to ensure that the climate conditions of the site selection area are suitable for unmanned aerial vehicle flight;

[0029] The infrastructure factor obtains power supply data to ensure that the site selection area is within the power supply reachable area to meet the energy demand of the nest; at the same time, signal coverage data is obtained to consider the signal coverage area to ensure that the communication between the unmanned aerial vehicle and the control center is smooth and unobstructed;

[0030] The security guarantee factor obtains no-fly zone data to avoid setting up nests in no-fly zones to ensure flight safety and comply with relevant regulations.

[0031] Further, the objective function of the multi-objective optimization site selection model in step (3) includes:

[0032] The total number of actual nests is minimized:

[0033]

[0034] In the formula, x i represents the use of the nest candidate point, which is a 0-1 variable, 1 indicating that the nest is built at the candidate point i, and 0 indicating that the nest is not built at the candidate point i;

[0035] The total coverage of the selected nests on the terrain is maximized:

[0036]

[0037] In the formula, Cover i represents the coverage of the candidate nest i on the terrain.

[0038] Further, the constraint conditions of the multi-objective optimization site selection model in step (3) include:

[0039] The overlap rate between any two selected nests is less than β:

[0040]

[0041] In the formula, O ij represents the overlap rate between candidate nests i and j;

[0042] The proportion of the area blocked by any selected nest by significant terrain is less than γ:

[0043]

[0044] In the formula, D k (2) This indicates the proportion of the candidate nest area that is blocked by terrain lines;

[0045] The effective coverage area ratio of any selected nest is greater than

[0046]

[0047] In the formula, D k (3) Indicates the proportion of the effective area of ​​the candidate nest;

[0048] Each selected nest has at least one other selected nest within its coverage area:

[0049]

[0050] The drone starts from any point in the terrain and travels to ηd max Within the range, there exists at least one nest:

[0051]

[0052] In the formula, d max It refers to the drone's battery life. kp <η·d max} is an indicator function, when d kp ≤η·d max The value is 1 if the condition is met, and 0 otherwise. p is any point in the terrain.

[0053] Furthermore, the implementation process of step (4) is as follows:

[0054] (41) Population initialization: Each candidate nest point has only two states: selected to build a nest and not selected to build a nest, using binary encoding; each nest selection scheme corresponds to an individual in the population, and each gene value of the individual's chromosome corresponds to the state of the candidate nest; the chromosome length is set to the number of candidate nest points, i.e., N; the chromosome genotype X k = (x1, x2, ..., x N ), x i The state corresponding to the i-th nest candidate point; the population size is M, which is initialized by random generation, and each individual represents a nest location scheme;

[0055] (42) Fitness evaluation: fitness evaluation index is set according to two objective functions of maximizing coverage rate and minimizing nest number, and a penalty term is added in the fitness function, the overlap rate of nests exceeding the set threshold is accumulated for penalty, the nests blocked by terrain are added for penalty, the nests with low effective coverage are accumulated for penalty, each nest coverage range needs to contain at least one standby nest, otherwise the penalty is accumulated:

[0056] P = P overlap + P blocked + P effective + P redundancy (10)

[0057]

[0058]

[0059] In the formula, P is the sum of the penalty term, P overlap , P blocked , P effective , P redundancy are overlap rate penalty, terrain blocking penalty, effective coverage penalty, and redundant coverage penalty, respectively; I{overlap i,j > β} is an indicator function, which takes 1 when the overlap rate of nests i and j is greater than β, otherwise 0; I{blocked i > γ} indicates that the nest is blocked by the terrain area when it is greater than γ, otherwise 0; indicates that the effective coverage rate of the nest is less than , otherwise 0; I{nearbycount i <1} indicates that the nest coverage range has no other nests, otherwise 0; F1 is the fitness value of the target maximum coverage rate, and F2 is the fitness value of the target minimum nest number; the greater the F1 value and the smaller the F2 value, the better the site selection scheme;

[0060] (43) Environment selection: first, the individuals are non-dominated sorted, and the individual with the lowest non-dominated level is selected as the priority reservation object; when the number of optimal solution set exceeds the preset external archive capacity, the selection is based on the crowding distance of the individual to ensure diversity; in the selection process of the solution set, the individual with larger crowding distance will be reserved first to ensure the uniform distribution of the solution set on the Pareto frontier and avoid excessive concentration in the local area;

[0061] The crowding distance of individual i for objective function f j :

[0062]

[0063] In the formula, F j(i) represents the value of the jth fitness function of the ith individual;

[0064] Total crowding distance of individual i:

[0065]

[0066] (44) Double-point crossover and binary bit mutation: a double-point crossover method is adopted, which randomly selects two crossover points of parent chromosomes, and exchanges the gene fragments between the two points; after the crossover operation, a mutation operation is performed, and a binary bit mutation operation is adopted; that is, one or more bits in the individual chromosome are randomly selected for mutation, and gene 0 is changed to 1, or gene 1 is changed to 0;

[0067] (45) Update population: select the newly generated offspring together with the parent individuals to construct the next generation population; first, non-dominated sorting is performed on all individuals to determine the dominance level of each individual; for individuals in the same dominance level, an elite reservation tournament selection method is adopted, and in the same dominance level, the individual with greater diversity is reserved through tournament selection to ensure the diversity of the solution set, and the individual with higher non-dominated level is preferentially reserved;

[0068] (46) Terminate iteration: when the preset maximum number of iterations is reached or when the Pareto optimal solution set does not change significantly for a plurality of consecutive generations, the algorithm is terminated; the final output solution set is the Pareto optimal solution set, which is selected by the decision maker to select the optimal scheme.

[0069] Further, the step (5) is implemented as follows

[0070] Calculate the preferred index u(i) of each solution and sort, wherein the solution with the largest preferred value is the compromise solution of the multi-objective problem:

[0071]

[0072] u(i) = u1(i) + u2(i) (21)

[0073] In the formula, u j (i) is the preferred value of the jth objective function value of the ith solution, f j (i) is the objective function value of the jth objective function value of the ith solution, f jmin and f jmax are the minimum value and the maximum value in the jth objective function value.

[0074] Advantages: compared with the prior art, the advantages of the present application are:

[0075] 1、The present application not only considers the coverage rate of the unmanned aerial vehicle nest, but also considers the number of nests, the overlap rate, the terrain sheltering and other factors, and establishes a multi-objective site selection optimization model; by minimizing the number of nests and maximizing the coverage rate, the maximization of resource utilization and the improvement of inspection efficiency are realized; at the same time, the method also considers the influence of terrain sheltering on the flight of the unmanned aerial vehicle, and by analyzing the blocking of the ridge line to the nest coverage area, unnecessary resource waste is avoided, and the inspection effect is improved.

[0076] 2、The improved NSGA-II algorithm is used for solving the multi-objective optimization model, compared with the traditional NSGA-II algorithm, the initial order is constructed based on the site selection standard screening of the candidate points, and by introducing the punishment mechanism, for the area blocked by the terrain, the penalty term is increased, and the fitness value is reduced, so that the convergence speed and solution accuracy of the algorithm are effectively improved; the improved algorithm can better handle the multi-objective optimization problem, and avoid the appearance of local optimal solution;

[0077] 3、Based on fuzzy set theory, the preferred indicators of each scheme in the Pareto optimal solution set are sorted, and the scheme with the maximum preferred value is selected as the final compromise solution; considering the preference and actual demand of the decision maker, the nest layout scheme more in line with the actual situation is selected, the scientificity and rationality of the decision are improved. BRIEF DESCRIPTION OF DRAWINGS

[0078] Figure 1 is a flowchart of the present application. DETAILED DESCRIPTION

[0079] The present application will be further described in detail below in combination with the drawings.

[0080] As shown in Figure 1 , the present application proposes a multi-unmanned aerial vehicle nest cooperative site selection method for forest area inspection, which specifically includes the following steps:

[0081] Step 1: geographic information data acquisition and preprocessing

[0082] According to the latitude and longitude of the target area, the geographic information data is queried, including high-precision DEM data, annual meteorological information, power supply network and signal coverage network. Integrate and import into the GIS system.

[0083] Outlier and missing value identification: calculate the Z-Score and IQR of the DEM data elevation value, measure the deviation degree of the data from the average value of the sample set, and the IQR is the difference between the upper quartile (Q3) and the lower quartile (Q1), and the Z-Score calculation formula is as follows:

[0084]

[0085] In the formula: h is a single observation elevation value, μ is the sample elevation mean, σ is the sample elevation standard deviation, if the elevation of a point satisfies |Z| > 3 or is lower than Q1-1.5xIQR or higher than Q3+1.5xIQR, the value is marked as an outlier. Identify NaN values and -9999 in the data and mark them as missing values.

[0086] Outlier and missing value processing: For small range outliers and missing values (the number of affected pixels is less than 5% of the total number of pixels), bilinear interpolation method is used, that is, according to the surrounding four known pixel values to replace or estimate the value:

[0087]

[0088] In the formula: (lon, lat) represents the horizontal and vertical coordinates of the pixel point, and h(lon, lat) represents the height value of the pixel point (lon, lat).

[0089] For large range outliers and missing values (the number of affected pixels is not less than 5% of the total number of pixels), use the UAV to take a photo in this range, and then use GIS software to generate new high-precision DEM data and replace the original error data.

[0090] Regional convex hull: Use the Overpass plug-in in QGIS to extract the latitude and longitude coordinate points of the forest area boundary from OpenStreetMap (OSM) data to form an initial boundary point set P = {P1, P2, …, P n} where P i = (lon i ,lat i ), and use the system function to obtain the convex hull point set C = {C1, C2, …, C m} where C i = (lon i ,lat i ), and the point set relationship The regular convex polygon region formed by the convex hull point set is taken as the forest area range, and the original DEM data is cut according to this range. Calculate the minimum bounding box of the convex hull region to obtain the coordinates of its four vertices

[0091] Regional gridding: Gridding the forest area, define the side length of the square grid as L according to the accuracy requirement of the inspection task, ensure that each grid can reasonably represent the geographical area within the range. Divide the DEM data into a series of square grids with side length L. Aggregate the elevation values in each grid (take the maximum value) to represent the elevation value of the grid.

[0092] Step 2: Extract the main ridge line of the forest area, and comprehensively analyze and determine the pre-selected location of the UAV nest.

[0093] Extract the main ridge line of the forest area: define the characteristics of the key grid as a) the terrain trend before and after is opposite (from rising to falling); b) the starting point and end point of flat terrain. Perform cross-section scanning of the terrain grid in multiple directions to identify grids that meet the key grid characteristics. Cross-section scanning is performed along the main direction (from south to north, from east to west), and diagonal directions are considered to capture more details. Mark and extract all grid points that meet the key grid characteristics. Use Kruskal's algorithm to find the minimum spanning tree (MST) of the extracted key grid point set to connect all key points while ensuring the total weight of the tree is minimized. Analyze the branches in the minimum spanning tree and remove those that do not meet the characteristics of the main ridge line, leaving the main path. Keep the point set of the main ridge line after the minimum spanning tree processing MST = {M1, M2, … M t} where M i = (lon i , lat i ).

[0094] Comprehensively consider accessibility, geographical environment, communication conditions, and security guarantees to preliminarily pre-select multiple UAV nest candidate points. Consider factors such as terrain, vegetation, and climate to improve communication quality and flight safety; ensure that the selected location has the necessary infrastructure support, such as power supply and communication network coverage, which are the basis for the normal operation of the UAV nest; among multiple alternative locations, conduct cost-benefit analysis to select locations with relatively low construction and maintenance costs and good coverage.

[0095] Read and preprocess the acquired terrain digital elevation data (DEM) to obtain the height and slope information of each point. Perform statistical analysis on local annual wind direction and wind speed, and simulate the surface wind speed of the terrain. Based on the UAV nest site selection principles, comprehensively study the power supply map information, signal coverage map information, no-fly zone information, and terrain data information to preliminarily screen the UAV nest candidate point set. The requirements that the UAV nest location must meet are as follows: power supply accessible area, signal coverage area, no-fly restricted area, flat ground, and smooth airflow area.

[0096] Step 3: Consider the UAV power, coverage, anti-interference, and terrain impact to establish a multi-objective optimization site selection model with the minimum total number of UAV nests and the maximum total coverage of the UAV nests on the terrain.

[0097] The first objective function in the model represents the minimum total number of actual UAV nests, as shown in the following formula:

[0098]

[0099] In the formula: x i indicates the use of the nest candidate point, where 1 indicates that a nest is built at candidate point i, and 0 indicates that no nest is built at candidate point i.

[0100] The second objective function indicates that the total coverage of the selected nests on the terrain is maximized, as shown in the following formula:

[0101]

[0102] In the formula: Cover i indicates the coverage of candidate nest i on the terrain.

[0103] The multi-objective optimization site selection model needs to satisfy the following constraints:

[0104] 1) The overlap rate between any two selected nests is less than β:

[0105]

[0106] In the formula: O ij indicates the overlap rate between candidate nests i and j. Reducing the overlap of monitoring areas between different nests optimizes resource allocation and use efficiency, avoids unnecessary repeated coverage, thereby improving the inspection efficiency and reducing the operating cost.

[0107] 2) The proportion of the area of any selected nest blocked by significant terrain (main ridge line) is less than γ:

[0108]

[0109] In the formula: D k (2) indicates the proportion of the area of the candidate nest blocked by the terrain line. When a UAV crosses a ridge, it needs to climb rapidly in a short time, which will consume more power, thereby reducing the total endurance time of the inspection; the wind speed is larger at the ridge, and the flight stability of the UAV will be affected in the high wind speed environment. Strong wind not only makes the UAV flight more difficult, but also can cause the flight trajectory to deviate, the vibration to intensify, thereby affecting the quality of image acquisition and data transmission. At the same time, the flight in high wind speed also causes more hardware wear and tear of the UAV, increasing the cost of maintenance and replacement. If the UAV frequently encounters ridge terrain requiring high climbing and strong wind during the inspection process, it may cause part of the tasks to be unable to be completed as planned, affecting the reliability and stability of the overall inspection task.

[0110] 3) The proportion of the effective coverage area of any selected nest is greater than

[0111]

[0112] D k (3) The ratio of the effective area of the candidate nest (the part in the nest coverage circle that does not exceed the target area range). When selecting the nest location, more effective patrol area coverage is preferred to avoid unnecessary coverage of non-patrol areas, ensure efficient operation of each nest within its coverage range, reduce resource waste caused by covering non-effective areas, and improve the efficiency of the entire system and the completion quality of the patrol task. Thus, the overall deployment and planning of the nest are optimized.

[0113] 4) Each selected nest coverage area has at least one other selected nest:

[0114]

[0115] If a nest fails, the other two backup nests can quickly receive the affected drones and take over the tasks of the failed nest. This redundant design ensures that the drone patrol task is not interrupted due to a single nest failure, enhances the system's anti-interference ability and robustness, and improves the overall reliability of the system.

[0116] 5) There is at least one nest within the range of ηd max from any point in the terrain:

[0117]

[0118] where d max is the endurance of the drone, I{d kp ≤ηd max} is an indicator function that takes the value 1 when d kp ≤η·d max , and 0 otherwise; p is an arbitrary point in the terrain. This ensures that if the drone encounters an accident during the patrol, it can quickly and safely fly to a nest before running out of power, ensuring safe landing and charging, increasing the system's continuous working ability, and avoiding patrol interruption and drone loss due to insufficient power.

[0119] Step 5: Use the improved NSGA-II algorithm to solve the site selection model and obtain the optimal site selection layout scheme for the drone nest.

[0120] Population initialization: In the nest site selection optimization problem, each candidate nest point has only two states, i.e., selected to build a nest and not selected to build a nest. Each candidate nest point has only two states, using binary encoding. Each selected nest scheme corresponds to an individual in the population, and each gene value of the individual chromosome corresponds to the state of the candidate nest. The chromosome length is set to the number of candidate nest points, i.e., N. The chromosome genotype Xk = (x1, x2,..., x N ), where x i corresponds to the state of the i-th nest candidate. The population size is M, which is initialized by randomly generated individuals, each of which represents a nest site selection scheme.

[0121] Fitness evaluation: Fitness evaluation is performed on each individual. The fitness function is calculated based on the Pareto dominance relationship. That is, whether an individual is simultaneously dominated by other individuals on all objectives. If an individual is not worse than other individuals on all objectives and superior in at least one objective, it is called a Pareto optimal solution. According to this dominance relationship, the individuals in the population are layered by non-dominated sorting. Each layer contains mutually non-dominated individuals. The present invention sets the fitness evaluation index according to two objective functions: 1) the number of covered terrain points, i.e. maximizing coverage rate; 2) the number of selected nests, i.e. minimizing the number of nests. The nest site selection scheme is comprehensively evaluated by these two objective functions. In addition, in order to improve the rationality of site selection and cope with complex forest environments, the following penalty terms are added to the fitness function. The overlap rate between nests exceeding the set threshold is accumulated to avoid resource waste; the nests blocked by terrain are added to ensure that the inspection route is not affected by terrain obstruction; at least one standby nest needs to be included in the coverage range of each nest, otherwise the penalty is accumulated to enhance the fault tolerance and redundancy of the system. The fitness function modified by the comprehensive penalty term comprehensively evaluates the nest site selection scheme.

[0122] P = P overlap + P blocked + P effective + P redundancy (10)

[0123]

[0124]

[0125] In the formula: P is the sum of the penalty terms, P overlap , P blocked , P effective , P redundancy are overlap rate penalty, terrain blocking penalty, effective coverage penalty, and redundant coverage penalty, respectively. I{overlap i,j >β} is an indicator function, which takes 1 when the overlap rate of nests i and j is greater than β, otherwise 0. I{blocked i >γ} indicates that the nest is blocked by the terrain area greater than γ, otherwise 0. I{nearbycount >α} indicates that the effective coverage rate of the nest is less than i<1> indicates 1 if there is no other nest in the coverage of the nest, otherwise 0. F1 is the fitness value of the target 1), and F2 is the fitness value of the target 2). The larger the F1 value and the smaller the F2 value, the better the site selection scheme.

[0126] Environment selection: The Pareto optimal solution set is retained from the current population. This step first performs non-dominated sorting on the individuals, and selects the individuals with the lowest non-dominated level as the priority retention object. When the number of optimal solution set exceeds the preset external archive capacity, selection is made based on the crowding distance of the individuals to ensure diversity. In the selection process of the solution set, individuals with larger crowding distance will be preferentially retained to ensure that the solution set is evenly distributed on the Pareto front, and to avoid excessive concentration in local areas.

[0127] The individual i for the objective function f j Crowding distance:

[0128]

[0129] In the formula: D j (i) indicates the value of the jth fitness function of the ith individual.

[0130] The total crowding distance of the individual i:

[0131]

[0132] Two-point crossover and binary bit mutation: Through the crossover operation in the genetic algorithm, the gene information of the parent individual is recombined to generate a new individual. In this paper, the two-point crossover method is used, which randomly selects two crossover points of the parent chromosome, and exchanges the gene fragments between the two points. The crossover probability is set to 0.9. After the crossover operation, mutation operation is performed, and binary bit mutation operation is adopted. That is, one or more bits in the individual chromosome are randomly selected for mutation, and gene 0 is changed to 1, or gene 1 is changed to 0. The mutation probability is set to 0.1, and the mutation generates a new offspring individual.

[0133] Update population: The newly generated offspring is selected together with the parent individuals to construct the next generation population. First, all individuals are non-dominated sorted to determine the dominance level of each individual. For individuals in the same dominance level, the elite retention tournament selection method (etour) is used to retain individuals with greater diversity in the same dominance level to ensure the diversity of the solution set, while preferentially retaining individuals with higher non-dominated levels.

[0134] Termination of iteration: Two termination conditions are adopted: when the preset maximum number of iterations is reached, the algorithm terminates; when the Pareto optimal solution set does not change significantly within a certain number of generations, the algorithm terminates. The final output solution set is the Pareto optimal solution set, which is provided for the decision maker to select the optimal scheme.

[0135] Step 6: Calculate compromise solutions using fuzzy set theory to provide decision support. For the Pareto optimal solution set obtained in Step 5, calculate the optimization index u for each solution. i The solutions are then sorted, and the solution with the largest optimal value is the compromise solution for the multi-objective problem.

[0136]

[0137] u(i)=u1(i)+u2(i)(21)

[0138] In the formula: u j (i) is the optimal value of the j-th objective function value of the i-th solution, f j (i) is the objective function value of the j-th objective function value of the i-th solution, f jmin and f jmax These are the minimum and maximum values ​​among the j-th objective function values.

[0139] For example, taking a typical mountainous terrain in Jiangsu Province, China, as an example, a site selection scheme for drone nests is designed. FCER and TOPRRANK are used as evaluation indicators, and the calculation process is as follows:

[0140]

[0141]

[0142] Among them, Cover i N represents the number of points required for coverage in the terrain for site selection scheme i. i This indicates the required number of nests, and T represents the total number of required points in the terrain. ij Let A represent the value of the j-th attribute in the i-th scheme. + and A - These represent the positive ideal solution and the negative ideal solution, respectively.

[0143] Table 1 compares the experimental results of the proposed method with those of mainstream location problem methods. As can be seen from Table 1, the location scheme of this invention has a higher cost-performance ratio and is more economical and efficient than the location schemes obtained by greedy algorithms and genetic algorithms. The proposed method requires fewer nests than other algorithms, achieves higher coverage, and has a higher FCER than other algorithms, indicating that the scheme provided by this invention can achieve higher coverage with less infrastructure investment, thus providing higher cost-performance. The TOPRANK of this invention is lower than that of other algorithms, indicating that the scheme provided by this invention can maintain high coverage efficiency while approaching the ideal solution.

[0144] Table 1 Comparison of results from different methods

[0145]

[0146] The above merely describes the preferred embodiments of the present application, and it should be pointed out that those skilled in the art can make several improvements and refinements without departing from the principles of the present application, and these improvements and refinements should also be considered as falling within the protection scope of the present application.

Claims

1. A method for multi-UAV nest site selection for forest inspection, characterized in that, The method comprises the following steps: (1) obtaining geographic information data and performing data preprocessing; (2) extracting main ridge lines of a forest area, and preliminarily selecting multiple unmanned aerial vehicle nest candidate points by comprehensively considering three factors of geographic environment, infrastructure and safety guarantee; (3) constructing a multi-objective optimization site selection model by comprehensively considering an unmanned aerial vehicle electric quantity, coverage rate, anti-interference performance and terrain influence, and determining a target function and a constraint condition of the model; (4) solving the multi-objective optimization site selection model by using an improved NSGA-II algorithm, and obtaining an optimal site selection layout scheme of the unmanned aerial vehicle nest; (5) calculating a Pareto optimal solution set solved in the step (4) based on a fuzzy set theory, calculating optimal indexes of each solution and sorting, and the solution with the maximum optimal value being a compromise solution of the multi-objective problem; The target function of the multi-objective optimization site selection model in the step (3) comprises: a minimum total number of actual nests: where x i is a 0-1 variable indicating whether to build a nest at candidate point i (1) or not (0). a maximum total coverage rate of the selected nests to the terrain: where Cover i represents the coverage of candidate nest i on the terrain; The constraint condition of the multi-objective optimization site selection model in the step (3) comprises: an overlap rate between any two selected nests being less than β: wherein O ij represents the overlap ratio between candidate nests i and j; a proportion of an area blocked by a significant terrain to any selected nest being less than γ: wherein D k (2) represents the proportion of the area of the candidate nest that is blocked by the terrain line. The proportion of the effective coverage area of any selected nest is greater than In the formula, D k (3) a ratio representing an effective area of a candidate nest; each selected nest covering at least one other selected nest: The drone starts from any point in the terrain to ηd max There is at least one nest within the range of In the formula, d max is the endurance mileage of the unmanned aerial vehicle, I{d kp ≤η·d max} is an indicator function, which takes the value 1 when d kp ≤η·d max , otherwise 0, and p is any point in the terrain. The step (4) is implemented as follows: (41) Population initialization: each candidate nest site has only two states, selected to build nests and not selected to build nests, using binary coding; each selected nest scheme corresponds to an individual in the population, and each gene value of the individual chromosome corresponds to the state of the candidate nest; the chromosome length is set to the number of candidate nest sites, i.e. N; the chromosome genotype X k = (x1, x2, …, x N ), x i corresponds to the state of the i-th nest candidate site; the population size is M, which is initialized by randomly generated, each individual represents a nest site selection scheme; (42) fitness evaluation: fitness evaluation indexes are set according to the two target functions of the maximum coverage rate and the minimum nest number, and a penalty term is added in the fitness function, the overlap rate between nests exceeding a set threshold is accumulated for penalty, the nest blocked by the terrain is increased for penalty, the nest with low effective coverage is accumulated for penalty, and at least one standby nest needs to be included in the coverage range of each nest, otherwise the penalty is accumulated: P = P overlap + P blocked + P effective + P redundancy (10) In the formula, P is the total sum of penalty terms, P overlap , P blocked , P effective , and P redundancy are respectively the overlap rate penalty, the terrain blocking penalty, the effective coverage penalty, and the redundant coverage penalty; I{overlap i,j >β} is an indicator function, which takes 1 when the overlap rate of nests i and j is greater than β, and 0 otherwise; I{blocked i >γ} indicates that the nest is blocked by a terrain blocking area greater than γ, and 0 otherwise; I{efficover >α} indicates that the effective coverage rate of the nest is less than α, and 0 otherwise; I{nearbycount i <1} indicates that there is no other nest in the coverage range of the nest, and 0 otherwise; F1 is the fitness value of the target maximum coverage rate, and F2 is the fitness value of the target minimized nest number; the greater the F1 value and the smaller the F2 value, the better the site selection scheme. (43) environment selection: first, the individuals are non-dominated sorted, and the individuals with the lowest non-dominated level are selected as the priority reservation objects; when the number of optimal solutions exceeds the preset external archive capacity, the individuals are selected based on the crowding distance to ensure diversity; in the selection process of the solution set, the individuals with larger crowding distances are preferentially reserved to ensure that the solution set is uniformly distributed on the Pareto frontier, and over-concentration in a local area is avoided; Individual i for objective function f j Crowding distance: where F j (i) represents the value of the jth fitness function of the ith individual; a total crowding distance of the individual i: (44) double-point crossover and binary bit mutation: a double-point crossover method is adopted, the gene fragments between two randomly selected crossover points of parent chromosomes are exchanged, and after the crossover operation, a binary bit mutation operation is performed; that is, one or more bits in the individual chromosome are randomly selected for mutation, and the gene 0 is changed to 1 or the gene 1 is changed to 0; (45) updating the population: the new offspring and the parent individuals are selected together to construct the next generation population; first, all individuals are non-dominated sorted to determine the dominance level of each individual; for the individuals in the same dominance level, an elite reservation tournament selection method is adopted, and the individuals with greater diversity are reserved in the same dominance level to ensure the diversity of the solution set, and the individuals with higher non-dominated levels are preferentially reserved. (46) Termination of iteration: the algorithm terminates when a preset maximum number of iterations is reached or when the Pareto optimal solution set no longer changes significantly over successive generations; the final output solution set is the Pareto optimal solution set for the decision maker to select the optimal scheme. 2.The method of claim 1, wherein, The geographic information data in step (1) includes high-precision DEM data, annual meteorological information, power supply network, and signal coverage network.

3. The method of claim 2, wherein, The data preprocessing in step (1) is implemented as follows: Calculate the Z-Score and IQR of the DEM data elevation values, which measure the deviation of the data from the average value of the sample set, and the IQR is the difference between the upper quartile and the lower quartile; the Z-Score calculation formula is as follows: In the formula, h is a single observation elevation value, μ is the sample elevation mean, and σ is the sample elevation standard deviation; For small-range outliers and missing values, a bilinear interpolation method is used, i.e., the value is replaced or estimated according to the surrounding four known pixel values: In the formula, (lon, lat) represents the horizontal and vertical coordinates of the pixel point, and h(lon, lat) represents the height value of the pixel point (lon, lat); For large-range outliers and missing values, a UAV is used to take aerial photographs in the range, and new high-precision DEM data is generated using GIS software to replace the original error data; The latitude and longitude coordinate points of the forest boundary are extracted from the open street map data to form an initial boundary point set P = {P1, P2, …, P n}, wherein P i =(lon i ,lat i ), a convex hull point set C = {C1, C2, …, C m} of the boundary point set is obtained by using a system function, wherein C i =(lon i ,lat i ), and the point set relationship A regular convex polygon region formed by the convex hull point set is taken as the forest range, and the original DEM data is cut according to the forest range; the minimum circumscribed rectangle of the convex hull region is calculated to obtain the coordinates of the four vertices The forest area is gridded, and the side length of the square grid is defined as L according to the accuracy requirement of the inspection task, to ensure that each grid can reasonably represent the geographical area within the range. The DEM data is divided into a series of square grids with a side length of L, and the elevation values within each grid are aggregated to represent the elevation value of the grid.

4. The multi-UAV nest cooperative site selection method for forest inspection of claim 1, wherein, The implementation process of step (2) for extracting the main ridge line of the forest area is as follows: The features of the key grid are defined as opposite topographic trends in front and back and the start and end points of flat terrain; the terrain grid is scanned in sections in multiple directions to identify the grid that meets the key grid features; the section scanning is carried out along the main direction and the diagonal direction is considered to capture more details; all grid points that meet the key grid features are marked and extracted; the minimum spanning tree is calculated for the extracted key grid point set using Kruskal algorithm to connect all key points while ensuring the minimum total weight of the tree; the branches in the minimum spanning tree are analyzed to remove the branches that do not meet the main ridge line features and the main path is reserved; the point set of the main ridge line after the minimum spanning tree processing is reserved MST={M1,M2,…M t}. Wherein M i =(lon i ,lat i ).

5. The multi-UAV nest cooperative site selection method for forest inspection according to claim 1, characterized in that, The implementation process of step (2) for preliminarily selecting multiple UAV nest candidate points is as follows: Process the digital elevation data based on geographic environmental factors to obtain the height and slope information of each point to evaluate the suitability of the terrain; at the same time, analyze the local annual wind direction and speed to simulate the surface wind speed and ensure that the climate conditions of the selected area are suitable for UAV flight; Obtain power supply data to ensure that the selected area is within the power supply reachable area to meet the energy demand of the nest; at the same time, obtain signal coverage data to consider the signal coverage area to ensure smooth communication between the UAV and the control center; Obtain no-fly zone data to avoid setting the nest in the no-fly zone to ensure flight safety and compliance with relevant regulations.

6. The multi-UAV nest cooperative site selection method for forest inspection of claim 1, wherein, The implementation process of step (5) is as follows Calculate the preferred index u(i) of each solution and sort them, where the solution with the largest preferred value is the compromise solution of the multi-objective problem: u(i)=u1(i)+u2(i)(21) where u j (i) is the preferred value of the jth objective function value of the ith solution, f j (i) is the objective function value of the jth objective function value of the ith solution, f jmin and f jmax is the minimum and maximum of the jth objective function value.

7. The multi-UAV nest site selection method for forest inspection of claim 3, wherein, The small-range outliers and missing values are data whose number of affected pixels is less than 5% of the total number of pixels.

Images