Unmanned aerial vehicle path planning method based on multi-strategy dung beetle optimization algorithm

Abstract

The invention provides an unmanned aerial vehicle path planning method based on a multi-strategy dung beetle optimization algorithm, and aims to solve the flight safety problem that an unmanned aerial vehicle collides with ground pedestrians in an urban low-airspace, analyze threats to the pedestrians when the unmanned aerial vehicle runs in a city, construct an unmanned aerial vehicle running risk map, and meanwhile, in the population initialization stage of the dung beetle algorithm, determine the path of the unmanned aerial vehicle in the urban low-airspace. A fusion Logistic chaotic mapping and reverse learning strategy is adopted, the population diversity is increased, multiple improvement strategies such as an adaptive step length strategy, a variable spiral search strategy, a lens imaging reverse learning strategy and a level flight strategy are introduced, the population diversity is enhanced, and the global search ability and the local exploration ability are effectively coordinated. According to the unmanned aerial vehicle path planning method considering the urban operation safety risk of the unmanned aerial vehicle, the operation risk and the path distance of the unmanned aerial vehicle can be effectively reduced, the smooth performance of the path is improved, and the actual environment requirement is better met.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly relates to a method for unmanned aerial vehicle path planning based on a multi-strategy dung beetle optimization algorithm. Background Technique

[0002] In recent years, with the successful application of unmanned aerial vehicles in multiple fields such as logistics distribution, emergency rescue, and environmental monitoring, unmanned aerial vehicles play an irreplaceable role in disaster rescue operations in urban environments due to their advantages of flexibility, high efficiency, precision, and low cost. However, when operating in the urban low-altitude environment, unmanned aerial vehicles need to consider many restrictions and safety risks such as crowd density, unmanned aerial vehicle operating performance, and urban building obstacles, which greatly reduces the flight safety level and rescue efficiency. Therefore, in the complex urban environment, considering the operating risks of unmanned aerial vehicles and scientifically planning a safe, collision-free, and as short as possible flight path for unmanned aerial vehicles quickly and with high quality is of great significance for improving the rescue efficiency of unmanned aerial vehicles.

[0003] At present, many studies on the path planning of safe operation of unmanned aerial vehicles have been carried out at home and abroad. In terms of risk assessment, risk models are established from multiple perspectives such as environmental impact and pedestrian safety; in terms of path planning, a series of heuristic optimization algorithms are proposed and improved for their deficiencies. However, traditional unmanned aerial vehicle path planning algorithms mainly take "efficiency" as the optimization goal, focusing on aspects such as short paths and low energy consumption, and fail to comprehensively consider the trade-off between "safety" and "efficiency", so it is difficult to meet the requirements of safe operation of unmanned aerial vehicles in complex urban environments. Summary of the Invention

[0004] One of the purposes of the present invention is to provide a method for unmanned aerial vehicle path planning based on a multi-strategy dung beetle optimization algorithm. By establishing a risk distribution map of the operating environment, it provides a decision-making basis for unmanned aerial vehicle path planning, constructs an unmanned aerial vehicle path cost model with the goal of minimizing the total cost of executing rescue tasks, and designs a multi-strategy dung beetle optimization algorithm to improve the path planning efficiency of unmanned aerial vehicles in complex urban environments and increase the safety of unmanned aerial vehicle operation.

[0005] A method for unmanned aerial vehicle path planning based on a multi-strategy dung beetle optimization algorithm provided by an embodiment of the present invention includes:

[0006] S1: Establish the urban airspace environment of the unmanned aerial vehicle by using the three-dimensional grid method;

[0007] S2: Consider the collision risk of the unmanned aerial vehicle with the urban ground crowd and establish a safe operation risk map of the unmanned aerial vehicle in the city;

[0008] S3: Parameter initialization, including: determining the starting position and target position of the unmanned aerial vehicle, solving the urban space dimension D, the maximum number of iterations T, and the dung beetle population number N; based on the initialized parameters, execute steps S4 to S9;

[0009] S4: Initialize the positions of the dung beetle population using the fused Logistic map and reverse learning strategy;

[0010] S5: Update the positions of the dung beetle population using the multi-strategy dung beetle optimization algorithm;

[0011] S6: Calculate the fitness values of the dung beetle individuals and update the best positions of the dung beetle individuals;

[0012] S7: Determine whether the positions of the dung beetle individuals exceed the set boundary region and perform boundary processing;

[0013] S8: Iteratively execute the above steps S6 to S7 until the maximum number of iterations is reached, and output the optimal path result. Otherwise, continue to iterate;

[0014] S9: Based on the optimal path result, draw a 3D map of the UAV path planning in the urban environment on the final urban environment map integrated with the urban airspace environment and the UAV urban operation safety risk map.

[0015] Optionally, in S2: Considering the risk of collision between the UAV and the urban ground population, establish a UAV urban operation safety risk map, including:

[0016] Obtain the two stages of the UAV-ground collision process, including: the UAV loses control and descends during flight and the UAV crashes and causes casualties to the ground personnel;

[0017] Considering the two stages of the UAV-ground collision process, establish a probability chain model R using the following causal chain:

[0018] R = P u ×P h

[0019]

[0020]

[0021] where P u is the probability of the UAV losing control and skidding, taking a fixed value of 10 -5 ; P h is the casualty rate of the ground personnel after being hit by the UAV; C p is the ground personnel density coefficient, and its value is related to the population density of the selected urban environment; C s is the shielding coefficient (the shielding coefficient refers to the degree of building density in the urban environment. When the value is 0, it means there is no environmental occlusion at the current position, and when the value is 1, it means the buildings are dense), and its value range is from 0 to 1; b is C sThe impact energy required for the casualty rate to reach 50% when it is 0.5, and d is C s The impact energy threshold required when it approaches 0. (Definition in the discipline of casualty limit. b is the impact energy when the casualty probability caused by the drone crash is 50%, and the value is 10 6 J. Similarly, the value of d here is 232 J); E is the collision kinetic energy of the drone operation; m is the mass of the drone; V op is the maximum flight speed of the drone;

[0022] Based on the probability chain model R, generate a safety risk map for the urban operation of drones:

[0023] By dividing the urban area into grids, the environment of the entire ground layer is represented by a cell matrix of size 40×40, that is, the probability chain model R. The side length of each cell is 25, and each matrix element corresponds to the risk probability R(i,j) of the cell location, where i∈[1,40] and j∈[1,40].

[0024] The risk map is a location-based two-dimensional map used to quantify the risk caused by the operation of drones to ground personnel in each cell.

[0025] Optionally, the step S4: Initialize the position of the dung beetle population using the fused Logistic mapping and reverse learning strategy, including:

[0026] Initialize the position of the dung beetle population using the fused Logistic mapping and reverse learning strategy through the following algorithm:

[0027] x n+1 =px n (1 - x n )(Logistic mapping formula)

[0028] ox n+1 =Lb + Ub - rand(1,D)×x n+1

[0029]

[0030] Among them, the value of x n is (0,1), p is the Logistic mapping control parameter, and x n+1 is the solution obtained by using the Logistic mapping; ox n is the reverse solution of x n+1 is the initial position of the dung beetle population; D is the exploration of the dung beetle population (both Lb and Ub are 1-row D-column vectors, and their values are generated by the random function rand. They jointly determine the position of the dung beetle population, that is, x n+1 n nThe value range of the vector is explored because they are random, so it is interpreted as exploration. The size of the urban spatial grid, i.e., the urban spatial dimension; rand(1,D) is a 1×D row vector; ones(1,D) is a 1×D row vector with all values being 1; Lb and Ub are the lower and upper limits of the population exploration area, which are 1×D row vectors all of 0 and 1×D row vectors all of 1 respectively.

[0031] Optionally, the step S5: updating the positions of the dung beetle population using a multi-strategy dung beetle optimization algorithm includes:

[0032] Updating the positions of the rolling dung beetle population using an adaptive step size strategy;

[0033] And / or,

[0034] Updating the positions of the breeding dung beetle population using a variable helix search strategy;

[0035] And / or,

[0036] Updating the positions of the foraging dung beetle population using a lens imaging reverse learning strategy;

[0037] And / or,

[0038] Updating the positions of the stealing dung beetle population through a levy flight strategy.

[0039] Optionally, the step of updating the positions of the rolling dung beetle population using an adaptive step size strategy includes:

[0040] Updating the positions of the rolling dung beetle population using an adaptive step size strategy through the following algorithm:

[0041]

[0042] Where, represents the position information of the i-th rolling dung beetle at the t-th iteration (when the dung beetle algorithm solves the UAV path planning problem, the total number of runs T of the algorithm is set in S3. Each time it runs, the position information of the dung beetle population will change. The t-th iteration refers to the number of runs of the dung beetle algorithm); represents the position information of the i-th rolling dung beetle at the (t + 1)-th iteration; α is an adaptive step size control factor; o 1 ∈(0, 0.2] and o 2 ∈(0, 1] are the deflection coefficient and the light coefficient respectively; γ represents the natural coefficient, taking a value of -1 or 1; is the worst position information in the current population (in S6, the calculated fitness value. According to the size of the fitness value, the larger the value, the worse the position information of the population); represents the change in light intensity; t represents the iteration number of the dung beetle algorithm; T is the maximum iteration number of the dung beetle algorithm; Levy is an intermediate variable.

[0043] Optionally, updating the position of the breeding dung beetle population using a variable helix search strategy includes:

[0044] Updating the position of the breeding dung beetle population using a variable helix search strategy through the following algorithm:

[0045]

[0046] ξ = e rl ·cos(2πl)

[0047] where Q 1 = 1 - t 1 / T, representing the dynamic selection factor of the breeding area; L b1 and U b1 are the lower and upper limits of the breeding area respectively, controlling the size of the breeding area of the breeding dung beetle; is the optimal position of the current breeding dung beetle; b 1 and b 2 are two independent random row vectors with values in [0, 1], is the position information of the i-th breeding dung beetle at the t 1 -th iteration, is the position information of the i-th breeding dung beetle at the t 1 +1-th iteration, r is a constant, usually set to 1, l is a random number between (-1, 1); e is the natural constant.

[0048] Optionally, updating the position of the foraging dung beetle population using a lens imaging backtracking learning strategy includes:

[0049] Updating the position of the foraging dung beetle population using a lens imaging backtracking learning strategy through the following algorithm:

[0050]

[0051] k = (1 + (t 2 / T) 0.5 ) 10

[0052] where Q 2 = 1 - t 2 / T, representing the dynamic selection factor of the foraging area; L b2 and U b2 are the lower and upper limits of the foraging area respectively, controlling the size of the foraging area of the foraging dung beetle; is the optimal position of the current foraging dung beetle; represents the i-th foraging dung beetle at the t 2The position information at the t-th iteration (when the dung beetle algorithm is used to solve the UAV path planning problem, the number of runs T of the algorithm is set in S3. Each time it runs, the position information of the dung beetle population changes. The t-th 2 iteration refers to the number of runs of the dung beetle algorithm); represents the position information of the i-th foraging dung beetle at the t-th 2 +1-th iteration; C 1 is a random number obeying the normal distribution; C 2 is a random row vector with a value range in [0, 1]; P k is the update method selection factor for the foraging dung beetle, which determines the update method of the foraging dung beetle; t 2 represents the number of runs of the dung beetle algorithm; T is the maximum number of iterations of the dung beetle algorithm.

[0053] Optionally, updating the position of the stealing dung beetle population through the levy flight strategy includes:

[0054] Updating the position of the stealing dung beetle population through the levy flight strategy by the following algorithm:

[0055]

[0056] where μ and ν are random numbers with values in [0, 1]; χ is a constant with a value of 1.5; Γ represents the standard Gamma function; is the optimal position of the current stealing dung beetle; represents the position information of the i-th stealing dung beetle at the t-th 3 iteration; represents the position information of the i-th stealing dung beetle at the t-th 3 +1-th iteration; g is a random row vector whose value range obeys the standard normal distribution; S is a constant value with a value of 0.5.

[0057] Optionally, S6: calculating the fitness value of the dung beetle individual and updating the best position of the dung beetle individual includes:

[0058] Calculating the fitness value of the dung beetle individual and updating the best position of the dung beetle individual by the following algorithm:

[0059] G = λ 1 (J path + J smooth + J rain ) + λ 2 J safe

[0060] where G represents the total cost of performing the rescue mission; J path represents the path distance cost; J smooth represents the path smoothness cost; J rain represents the threat cost in the harsh area; Jsafe represents the operation safety risk degree, that is, the sum of the risk probabilities R of the positions of all flight path nodes of the UAV; λ 1 is the cost coefficient comprehensively considering path distance, path smoothness and threat in harsh areas, and its value is 0.8; λ 2 is the safety coefficient, and its value is 0.2. (When solving the optimal position of the dung beetle individual, an evaluation criterion G is required, which here refers to the total cost of performing the rescue mission. The fitness value of the dung beetle individual is calculated through G. For example, if there are 30 dung beetles in the population, the fitness value of each of the 30 dung beetles is calculated, and then the smallest fitness value corresponds to the optimal position of the dung beetle individual).

[0061] Other features and advantages of the present invention will be described in the subsequent description, and in part will become obvious from the description, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the written description and the drawings.

[0062] The technical solution of the present invention will be further described in detail below through the drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings are used to provide a further understanding of the present invention, and constitute a part of the description. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0064] Figure 1 is a schematic diagram of a UAV path planning method based on a multi-strategy dung beetle optimization algorithm in an embodiment of the present invention;

[0065] Figure 2 is a schematic diagram of a UAV urban operation safety risk map established in an embodiment of the present invention;

[0066] Figure 3 is a comparison diagram of algorithm results for UAV path planning in an urban environment in an embodiment of the present invention;

[0067] Figure 4 is a result diagram of a multi-strategy dung beetle optimization algorithm based on a risk map in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0068] The preferred embodiments of the present invention are described below with reference to the drawings. It should be understood that the preferred embodiments described herein are only for explaining and understanding the present invention, and are not used to limit the present invention.

[0069] An embodiment of the present invention provides a UAV path planning method based on a multi-strategy dung beetle optimization algorithm, as Figure 1 shown, including:

[0070] S1: Establish the urban airspace environment of the UAV using the three-dimensional grid method; the urban airspace environment refers to urban buildings, and in this application, multiple cuboids are established using the three-dimensional grid method for representation;

[0071] S2: Consider the collision risk of the UAV with the ground population in the city and establish a safety risk map for the urban operation of the UAV;

[0072] S3: Parameter initialization, including: determining the starting position and target position of the UAV, solving the urban space dimension D, the maximum number of iterations T, and the number of dung beetle populations N; execute steps S4 to S9 based on the initialized parameters;

[0073] S4: Initialize the positions of the dung beetle population using the fused Logistic mapping and reverse learning strategy; for N dung beetles, adopt the fused Logistic mapping and reverse learning strategy so that each position of these N dung beetles stores 1 row of D-dimensional vectors (this vector group stores the position information of the dung beetle population), and after the information in the vectors is processed, the path result of the UAV from the starting position to the target position can be obtained;

[0074] S5: Update the positions of the dung beetle population using the multi-strategy dung beetle optimization algorithm;

[0075] S6: Calculate the fitness value of the dung beetle individuals and update the best positions of the dung beetle individuals; assume that there are 30 dung beetle populations, and each population has its corresponding position, that is, the dung beetle population position; calculate the fitness value of each dung beetle individual, sort the fitness values of these 30 dung beetle individuals from smallest to largest, and the smallest fitness value corresponding to the dung beetle individual is the corresponding best position of the dung beetle individual;

[0076] S7: Judge whether the positions of the dung beetle individuals exceed the set boundary area and perform boundary processing; this is the content in the dung beetle algorithm. For example, for the first dung beetle individual in the algorithm, through the initialization of the dung beetle population in step S4, it is a 1-row D-column mathematical vector in mathematics. It is required that each number in this vector has a value range between 0 and 1. Then, for the numbers outside this range, boundary processing is required. Numbers greater than 1 are taken as 1, and numbers less than 0 are taken as 0;

[0077] S8: Iteratively execute the above steps S6 to S7 until the maximum number of iterations is reached, and output the optimal path result; otherwise, continue to iteratively execute; in each iteration process, the optimal path result will be output once, and here it stops when the value reaches the maximum number of iterations;

[0078] S9: Based on the optimal path result, draw a 3D map of the UAV path planning in the urban environment on the final urban environment map after integrating the urban airspace environment and the UAV urban operation safety risk map; it is the map after considering the integration of two aspects (the urban airspace environment in step S1 and the UAV urban operation safety risk map in step S2). When the UAV is flying, it is required to avoid the buildings in the city (represented by a series of cuboids). In addition, the safety of the UAV's urban operation needs to be considered. Considering these two issues, the final urban environment map is obtained, and a 3D map of the UAV path planning is drawn on the final urban environment map.

[0079] Urban obstacles refer to urban buildings, which are represented by cuboids. The operation risk of the UAV in the urban environment refers to the risk of causing harm to ground personnel. When the UAV performs rescue tasks in the urban environment, the urban obstacles are established through the 3D grid map method. The urban airspace environment is divided into several grid cells by the grid method, and each grid cell is a cube. Whether the UAV can pass through the urban area is represented by marking the cube. In addition to the above obstacle threats, the safety of the UAV's path in the urban environment is also considered. Therefore, the safety probability R in the risk map is used as one of the evaluation criteria for the UAV to select a path, and this content is reflected in the total cost G of the rescue UAV.

[0080] First, obtain environmental information such as the urban environmental terrain, threat areas, starting point, and ending point. Then, considering the total cost of the UAV performing rescue tasks, construct a mathematical model for UAV path planning. Use the multi-strategy dung beetle optimization algorithm to encode the population individuals and solve the minimum value of the UAV path planning mathematical model to obtain the path planning result.

[0081] Establish a UAV urban operation safety risk map as Figure 2 shown; a comparison diagram of the UAV path planning result algorithms in the urban environment is as Figure 3 shown; a result diagram of the multi-strategy dung beetle optimization algorithm based on the risk map is as Figure 4 shown.

[0082] This application provides a decision-making basis for UAV path planning by establishing a risk distribution map of the operating environment, constructs a UAV path cost model with the goal of minimizing the total operating cost of performing rescue tasks, designs a multi-strategy dung beetle optimization algorithm, improves the UAV path planning efficiency in complex urban environments, and increases the safety of UAV operation.

[0083] In one embodiment, S2: Considering the risk of the UAV colliding with the urban ground population, establish a UAV urban operation safety risk map, including:

[0084] The two stages of obtaining the process of the drone colliding with the ground include: the drone descending out of control during flight and the drone crashing and causing casualties to ground personnel;

[0085] Considering the two stages of the process of the drone colliding with the ground, a probability chain model R is established using the following causal chain:

[0086] R = P u ×P h

[0087]

[0088] where P u is the probability of the drone experiencing an out-of-control skid; P h is the casualty rate of ground personnel after being hit by the drone; C p is the ground personnel density coefficient, and its value is related to the population density of the selected urban environment; C s is the shielding coefficient, and its value range is from 0 to 1; b is the impact energy required for the casualty rate to reach 50% when C s = 0.5; d is the impact energy threshold required when C s approaches 0; E is the collision kinetic energy of the drone's operation; m is the mass of the drone; V op is the maximum flight speed of the drone. Among them, the shielding coefficient refers to the degree of building density in the urban environment. When the value of C s is 0, it means there is no environmental occlusion at the current location, and when the value is 1, it means the buildings are dense; the impact energy is defined in the discipline of casualty limit. b is the impact energy when the probability of casualties caused by the drone falling is 50%, and its value is 10 6 J. Similarly, the value of d here is 232J;

[0089] Based on the probability chain model R, a safety risk map for the drone's urban operation is generated:

[0090] The risk map is a location-based two-dimensional map used to quantify the risk caused by the drone's operation to ground personnel in each unit. By dividing the urban area into grids, the entire ground layer environment is represented by a cell matrix of size 40×40, that is, the probability chain model R. The side length of each cell is 25, and each matrix element corresponds to the risk probability R(i,j) of the cell's location, where i ∈ [1,40], j ∈ [1,40].

[0091] In one embodiment, the step S4: initializing the position of the dung beetle population using the fused Logistic mapping and reverse learning strategy includes:

[0092] Initializing the position of the dung beetle population using the fused Logistic mapping and reverse learning strategy through the following algorithm:

[0093] x n+1 = px n (1 - x n )

[0094] ox n+1 = Lb + Ub - rand(1, dim) × x n+1

[0095]

[0096] where x n takes values in (0, 1), p is the control parameter of the Logistic map, and x n+1 is the solution obtained using the Logistic map; ox n is the reverse solution of x n+1 i.e., the initial position of the dung beetle population; D is the size of the urban space grid explored by the dung beetle population, i.e., the urban space dimension; rand(1, D) is a 1-by-D vector; ones(1, D) is a 1-by-D vector with all elements equal to 1; Lb and Ub are the lower and upper bounds of the population exploration region, which are 1-by-D vectors of all 0s and 1-by-D vectors of all 1s, respectively. Exploration means that: Lb and Ub are both 1-by-D vectors, and their values are generated by the random function rand. Together, they determine the position of the dung beetle population, which is the value range of the x n+1 vector. Exploration is due to their randomness, so it is interpreted as exploration. n The initial population of DBO is generated randomly. This method is prone to problems such as uneven population distribution, resulting in reduced population diversity, low population solution quality, and affecting the convergence speed of the algorithm. The Logistic chaotic map has characteristics such as randomness and chaotic ergodicity, and can make the population distribution uniform compared with random probability generation. At the same time, when using the Logistic chaotic map to initialize the population, a reverse learning strategy is introduced to improve the population solution quality.

[0097] In one embodiment, step S5: updating the position of the dung beetle population using a multi-strategy dung beetle optimization algorithm includes:

[0098] Updating the position of the rolling dung beetle population using an adaptive step size strategy;

[0099] And / or,

[0100] Updating the position of the breeding dung beetle population using a variable helix search strategy;

[0101] And / or,

[0102] Updating the position of the foraging dung beetle population using a lens imaging reverse learning strategy;

[0103] And / or,

[0104] and / or

[0105] Update the position of the stealing dung beetle population through the Levy flight strategy.

[0106] Adopt an adaptive step size strategy to update the position of the ball-rolling dung beetle, expanding the global search ability of the ball-rolling dung beetle; adopt a variable helix search strategy, utilize the position information of the breeding area to improve the algorithm convergence speed; adopt a lens imaging reverse learning strategy, by adjusting the size of k, the dung beetle can be perturbed, which can not only jump out of the current position but also expand the search range of the foraging dung beetle, improving the diversity of the dung beetle population; update the position of the stealing dung beetle through the Levy flight strategy, and when a safer and more efficient path is solved, it can quickly transfer to a path with good quality.

[0107] In one embodiment, the step of updating the position of the ball-rolling dung beetle population by adopting the adaptive step size strategy includes:

[0108] Update the position of the ball-rolling dung beetle population by adopting the adaptive step size strategy through the following algorithm:

[0109]

[0110] wherein, represents the position information of the i-th ball-rolling dung beetle at the t-th iteration (when the dung beetle algorithm solves the UAV path planning problem, the total number of runs T of the algorithm is set in S3. Each time it runs, the position information of the dung beetle population will change. The t-th iteration refers to the number of runs of the dung beetle algorithm); represents the position information of the i-th ball-rolling dung beetle at the (t + 1)-th iteration; α is the adaptive step size control factor; o 1 ∈(0, 0.2] and o 2 ∈(0, 1] are the deflection coefficient and the light coefficient respectively; γ represents the natural coefficient, taking a value of -1 or 1; is the worst position information in the current population (in S6, the calculated fitness value. According to the size of the fitness value, the larger the value, the worse the position information of the population); represents the change in light intensity; t represents the iteration number of the dung beetle algorithm; T is the maximum iteration number of the dung beetle algorithm.

[0111] The global search of the dung beetle algorithm is mainly realized through the ball-rolling behavior of the dung beetle. The ball-rolling behavior is divided into two modes: with obstacles and without obstacles. During the position update process of the ball-rolling dung beetle in the obstacle-free mode, the lack of connection with other individuals in the population leads to a decrease in the global search ability. The adaptive step size strategy can effectively expand the global search ability of the ball-rolling dung beetle through the adaptive step control factor α.

[0112] In one embodiment, updating the positions of the breeding dung beetle population by using a variable spiral search strategy includes:

[0113] Update the positions of the breeding dung beetle population by using a variable spiral search strategy through the following algorithm:

[0114]

[0115] ξ = e rl ·cos(2πl)

[0116] where Q 1 = 1 - t 1 / T, representing the dynamic selection factor of the breeding area; L b1 and U b1 are the lower and upper limits of the breeding area respectively, controlling the size of the breeding area of the breeding dung beetles; is the optimal position of the current breeding dung beetle; b 1 and b 2 are two independent random row vectors with values in [0, 1], is the position information of the i-th breeding dung beetle at the t 1 -th iteration, is the position information of the i-th breeding dung beetle at the t 1 +1-th iteration, r is a constant, usually set to 1, l is a random number between (-1, 1); e is the natural constant.

[0117] The variable spiral search strategy can adjust the moving distance of each position update of the population in a spiral shape according to the optimal position in the dung beetle population and the current self-position. Using the variable spiral search strategy to update the positions of the breeding dung beetles can make great use of the position information of the breeding area and improve the convergence speed of the algorithm.

[0118] In one embodiment, updating the positions of the foraging dung beetle population by using a lens imaging reverse learning strategy includes:

[0119] Update the positions of the foraging dung beetle population by using a lens imaging reverse learning strategy through the following algorithm:

[0120]

[0121] k = (1 + (t 2 / T) 0.5 ) 10

[0122] where Q 2 = 1 - t 2 / T, representing the dynamic selection factor of the foraging area; L b2 and U b2 are the lower and upper limits of the foraging area respectively, controlling the size of the foraging area of the foraging dung beetles; is the optimal position of the current foraging dung beetle; represents the position information of the i-th foraging dung beetle at the t 2 -th iteration (when the dung beetle algorithm is used to solve the UAV path planning problem, the number of runs T of the algorithm is set in S3. Each time it runs, the position information of the dung beetle population will change. The t 2 -th iteration refers to the number of runs of the dung beetle algorithm); represents the position information of the i-th foraging dung beetle at the t 2 +1-th iteration; C 1 is a random number following a normal distribution; C 2 is a random row vector with a value range in [0,1]; P k is the update method selection factor for foraging dung beetles, which determines the update method of foraging dung beetles; t 2 represents the number of runs of the dung beetle algorithm; T is the maximum number of iterations of the dung beetle algorithm.

[0123] When foraging dung beetles update their positions in the foraging area, there are problems of slow population convergence speed and low diversity, which easily cause the algorithm to fall into local optimum. By adopting the lens imaging reverse learning strategy and adjusting the size of k, the dung beetles can be perturbed, which can not only jump out of the current position but also expand the search range of foraging dung beetles, improving the diversity of the dung beetle population.

[0124] In one embodiment, updating the position of the stealing dung beetle population through the levy flight strategy includes:

[0125] Updating the position of the stealing dung beetle population through the levy flight strategy by the following algorithm:

[0126]

[0127] where μ and ν are random numbers with values in [0,1]; χ is a constant with a value of 1.5; Γ represents the standard Gamma function; is the optimal position of the current stealing dung beetle; represents the position information of the i-th stealing dung beetle at the t 3 -th iteration; represents the position information of the i-th stealing dung beetle at the t 3 +1-th iteration; g is a random row vector whose value range follows a standard normal distribution; S is a constant value with a value of 0.5; Levy is an intermediate variable.

[0128] The kleptoparasitic dung beetle tends to perform local position updates near the best food sources. This update method often causes the dung beetles to quickly gather in the area near the optimal position during the iteration process, easily leading to the population falling into a local optimum and causing the search to stagnate. Introducing Levy flight into the position update of the kleptoparasitic dung beetle can expand the search range of the dung beetle, increase population diversity, and enable the kleptoparasitic dung beetle to perform more accurate optimization near the optimal solution, improving the global search ability of the algorithm.

[0129] In one embodiment, the step S6: calculating the fitness value of the dung beetle individual and updating the best position of the dung beetle individual includes:

[0130] Calculating the fitness value of the dung beetle individual and updating the best position of the dung beetle individual through the following algorithm:

[0131] G = λ 1 (J path + J smooth + J rain ) + λ 2 J safe

[0132] where G represents the total cost of performing the rescue mission; J path represents the path distance cost; J smooth represents the path smoothness cost; J rain represents the threat cost in the harsh area; J safe represents the operation safety risk degree, that is, the sum of the risk probabilities R of all flight path nodes of the UAV; λ 1 is the cost coefficient comprehensively considering the path distance, path smoothness and threat in the harsh area, with a value of 0.8; λ 2 is the safety coefficient, with a value of 0.2. (When finding the best position of the dung beetle individual, an evaluation criterion G is needed, that is, the total cost of performing the rescue mission. The fitness value of the dung beetle individual is calculated through G. For example, if there are 30 dung beetles in the population, calculate the fitness value of each of the 30 dung beetles, then the smallest fitness value corresponds to the best position of the dung beetle individual).

[0133] Table 1 below shows the comparison of the path planning results of the method of the present invention (MSDBO) with the dung beetle optimization algorithm (DBO), the adaptive dung beetle optimization algorithm (AMDBO), the particle swarm optimization algorithm (PSO), and the grey wolf optimization algorithm (GWO) in the simulation.

[0134] Table 1 Path planning result data

[0135]

[0136]

[0137] From the path planning result data in Table 1, it can be seen that the three evaluation indicators of the average path, the worst path, and the standard deviation solved by the MSDBO algorithm are better than those of the other four comparison algorithms. The variances of MSDBO, DBO, AMDBO, GWO, and PSO are 20.09, 69.41, 26.79, 32.81, and 28.80 respectively. From the above analysis, it can be seen that the MSDBO algorithm can provide a more efficient, safe, and stable flight path compared to other heuristic algorithms.

[0138] In the population initialization stage of this application, a strategy that combines Logistic chaotic mapping and reverse learning is adopted to increase population diversity. Multiple improvement strategies such as an adaptive step size strategy, a variable helix search strategy, a lens imaging reverse learning strategy, and a levy flight strategy are introduced to improve the global optimization ability and convergence speed of the algorithm. At the same time, considering the operation risks of unmanned aerial vehicles in urban environments, a urban risk map is constructed, which improves the operation safety of unmanned aerial vehicles. The experimental results show that the algorithm proposed in this invention is an effective and feasible method, which can improve the operation efficiency of the algorithm and the safety performance of unmanned aerial vehicles.

[0139] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A UAV path planning method based on a multi-strategy dung beetle optimization algorithm, characterized in that: include: S1: Use the three-dimensional grid method to establish the urban airspace environment for UAVs; S2: Consider the risk of drones colliding with people on the ground in the city and establish a safety risk map for drone operations in the city; S3: parameter initialization, including: determining the starting position and target position of the UAV, solving the urban space dimension D, the maximum number of iterations T, and the number of dung beetle populations N; Execute steps S4 to S9 based on the initialization parameters; S4: Initialize the position of dung beetle population using fusion logistic mapping and reverse learning strategy; S5: Use multi-strategy dung beetle optimization algorithm to update the dung beetle population position; S6: Calculate the fitness value of the dung beetle individual and update the optimal position of the dung beetle individual; S7: determining whether the position of the dung beetle individual exceeds the set boundary area and performing boundary processing; S8: iteratively execute the above steps S6 to S7 until the maximum number of iterations is reached and the optimal path result is output; otherwise, continue iterative execution; S9: Based on the optimal path result, a three-dimensional map of the drone path planning in the urban environment is drawn on the final urban environment map after the urban airspace environment and the drone urban operation safety risk map are integrated.

2. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm according to claim 1, characterized in that: S2: Considering the risk of UAVs colliding with people on the ground in the city, a UAV city operation safety risk map is established, including: The two stages of obtaining the UAV-to-ground collision process include: the UAV falls out of control during the flight and the UAV falls and hits the ground personnel causing casualties; Considering the two stages of the UAV-to-ground collision process, the probability chain model R is established using the following causal chain: R=P u ×P h Among them, P u The probability of the drone gliding out of control is set to a fixed value of 10. -5 ;P h C is the casualty rate of ground personnel after being hit by drones; p is the ground population density coefficient, and its value is related to the population density of the selected urban environment; C s is the shielding coefficient, ranging from 0 to 1; b is C s = 0.5, the impact energy required to reach 50% casualty rate; d is C s The required impact energy threshold when approaching 0; E is the collision kinetic energy of the UAV; m is the mass of the UAV; V op is the maximum flight speed of the drone; Based on the probability chain model R, a safety risk map of drone operation in cities is generated: By dividing the urban area into grids, the environment of the entire ground layer is represented by a 40×40 cell matrix, namely the probability chain model R. The side length of each cell is 25, and each matrix element corresponds to the risk probability R(i,j) of the cell location, where i∈[1,40], j∈[1,40].

3. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 1, characterized in that: S4: using fusion logistic mapping and reverse learning strategy to initialize the position of dung beetle population, including: The position of the dung beetle population is initialized by fusion logistic mapping and reverse learning strategy through the following algorithm: x n+1 =px n (1-x n ) ox n+1 =Lb+Ub-rand(1,D)×x n+1 Among them, x n The value of is (0,1), p is the control parameter of Logistic mapping, x n+1 For x n The solution obtained using Logistic mapping; ox n+1 For x n+1 The reverse solution of is the initial position of the dung beetle population; D is the size of the urban space grid explored by the dung beetle population, that is, the dimension of the urban space; rand(1,D) is a 1-row and 1-column vector; ones(1,D) is a 1-row and 1-column vector with all values ​​1; the lower and upper limits of the exploration area of ​​the Lb and Ub populations are a 1-row and 1-column vector with all values ​​0 and a 1-row and 1-column vector with all values ​​1, respectively.

4. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 1, characterized in that: S5: using a multi-strategy dung beetle optimization algorithm to update the dung beetle population position, including: Adopting adaptive step-size strategy to update the population position of rolling dung beetles; and / or, The variable spiral search strategy is used to update the location of the breeding dung beetle population; and / or, The lens imaging reverse learning strategy is used to update the location of foraging dung beetle populations; and / or, Update the location of the stealing dung beetle population through the levy flight strategy.

5. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 4, characterized in that: The method of updating the position of the rolling dung beetle population by using an adaptive step length strategy includes: The position of the rolling dung beetle population is updated using the following algorithm using an adaptive step size strategy: in, represents the position information of the i-th ball-rolling dung beetle at the t-th iteration; represents the position information of the i-th rolling ball dung beetle at the t+1th iteration; α is the adaptive step size control factor; o1∈(0,0.2] and o2∈(0,1] are the deflection coefficient and illumination coefficient respectively; γ represents the natural coefficient, which takes the value of -1 or 1; is the worst position information in the current population; Represents the change of light intensity; t represents the number of iterations of the dung beetle algorithm; T is the maximum number of iterations of the dung beetle algorithm.

6. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 4, characterized in that: The method of updating the breeding dung beetle population position by adopting a variable spiral search strategy comprises: The position of the breeding dung beetle population is updated using the variable spiral search strategy through the following algorithm: ξ=e rl ·cos(2πl) Among them, Q1 = 1-t1 / T, which represents the dynamic selection factor of the breeding area; L b1 and U b1 They are the lower and upper limits of the breeding area, respectively, and control the size of the breeding area of ​​dung beetles; is the optimal position for the current breeding of dung beetles; b1 and b2 are two independent random row vectors with values ​​of [0,1]. is the position information of the i-th breeding dung beetle at the t1th iteration, is the position information of the i-th breeding dung beetle at the t1+1th generation selection, r is a constant, usually set to 1, l is a random number between (-1,1); e is a natural constant.

7. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 4, characterized in that: The method of updating the position of the foraging dung beetle population by adopting the lens imaging reverse learning strategy comprises: The position of the foraging dung beetle population is updated using the lens imaging reverse learning strategy through the following algorithm: k=(1+(t2 / T) 0.5 ) 10 Among them, Q2 = 1-t2 / T, which represents the dynamic selection factor of the foraging area; L b2 and U b2 are the lower and upper limits of the foraging area, respectively, controlling the size of the foraging area of ​​the foraging dung beetle; The optimal position for the current foraging dung beetle; represents the position information of the i-th foraging dung beetle at the t2-th iteration; represents the position information of the i-th foraging dung beetle at the t2+1th iteration; C1 is a random number that obeys the normal distribution; C2 is a random row vector with a value range of [0,1]; P k It is the factor for selecting the update mode of the foraging dung beetle, which determines the update mode of the foraging dung beetle; t2 represents the number of times the dung beetle algorithm runs; T is the maximum number of iterations of the dung beetle algorithm.

8. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 4, characterized in that: The updating of the population position of the stealing dung beetle through the Levy flight strategy includes: The position of the stealing dung beetle population is updated through the levy flight strategy using the following algorithm: Among them, μ and ν are random numbers with values ​​in [0,1]; χ is a constant with a value of 1.5; Γ represents the standard Gamma function; The optimal position for the current stealing dung beetle; represents the position information of the i-th stealing dung beetle at the t3th iteration; represents the position information of the i-th thieving dung beetle at the t3+1th iteration; g is a random row vector whose value range obeys the standard normal distribution; S is a constant value, which is 0.5; Levy is an intermediate variable.

9. The UAV path planning method based on the multi-strategy dung beetle optimization algorithm as claimed in claim 1, characterized in that: S6: calculating the fitness value of the dung beetle individual and updating the optimal position of the dung beetle individual, including: The fitness value of the dung beetle individual is calculated by the following algorithm, and the optimal position of the dung beetle individual is updated: G=λ1(J path +J smooth +J rain )+λ2J safe Where G represents the total cost of performing the rescue mission; J path represents the path distance cost; J smooth represents the path smoothing cost; J rain represents the cost of severe regional threats; J safe It represents the operational safety risk, that is, the sum of the risk probabilities R of all flight path nodes of the UAV; λ1 is the cost coefficient that comprehensively considers the path distance, path smoothness and threats in harsh areas, and its value is 0.8; λ2 is the safety factor, and its value is 0.2.

Images