Unmanned aerial vehicle task allocation method based on normal cloud model artemisinin optimization algorithm

By introducing normal cloud model and artemisinin optimization algorithm into the drone task allocation algorithm, combining chaotic mapping and graph theory principles, the problem of local optimization and insufficient adaptability in drone task allocation is solved, and the more Efficient task planning and execution.

CN118963411APending Publication Date: 2024-11-15UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411020713.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

When facing complex and multi-peak optimization problems, existing drone task allocation algorithms are difficult to balance exploration and development, easily fall into local optimal solutions, and are not adaptable, making it difficult to find the optimal solutions under different constraints.

Method used

The artemisinin optimization algorithm based on the normal cloud model is adopted to initialize the population through a chaotic mapping strategy, and the normal cloud distribution is introduced to update the historical optimal position, enhancing the algorithm's ability to jump out of the local optimality, and construct the drone in combination with the principle of graph theory. Task allocation model, optimize task execution order.

Benefits of technology

It improves the global search capability of the drone task allocation algorithm, reduces computing costs, converges faster, is suitable for complex and large-scale problems, and achieves more efficient and stable task planning and execution.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118963411A_ABST
    Figure CN118963411A_ABST
Patent Text Reader

Abstract

The invention discloses an unmanned aerial vehicle task allocation method based on a normal cloud model artemisinin optimization algorithm. The method comprises the following steps: firstly, based on an actual flight distance and task requirements, considering the distribution condition of obstacles in a simple environment, and establishing an unmanned aerial vehicle task allocation model under voyage constraint; secondly, initializing an artemisinin population for a task allocation model and tent chaotic mapping based on an artemisinin optimization algorithm, and constructing an optimization model based on malaria artemisinin drug treatment process heuristic; and updating a global optimal position in the algorithm by using a normal cloud model, and adaptively adjusting an entropy value of the normal cloud model along with the increase of the number of iterations, thereby effectively improving the global development capability and solving speed, and jumping out of a local optimal solution. According to the method, while the voyage constraint of the unmanned aerial vehicle is considered, the obstacles possibly existing in the actual task process are also considered, solving is carried out based on the artemisinin optimization algorithm of the normal cloud model, and the task allocation problem of the unmanned aerial vehicle can be solved to a certain extent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicle task planning, and in particular relates to an artemisinin optimization algorithm unmanned aerial vehicle task allocation method based on a normal cloud model. Background Art

[0002] Unmanned aerial vehicles (UAVs) are high-tech devices that can carry a variety of monitoring devices. They fly autonomously through preset programming or remote control devices. Compared with traditional manned aircraft, drones have significant advantages such as low cost, strong maneuverability and the ability to avoid casualties, which makes them widely used in the field of military reconnaissance. With the rapid development of intelligent technology, the application fields of drones are constantly expanding and the types of tasks are becoming increasingly diverse. Especially in the field of post-disaster rescue, the application of drones has shown its unique value. They can quickly enter the disaster area, conduct real-time aerial monitoring, provide key disaster assessment information, and provide decision support for the planning and execution of rescue operations. The use of drones can not only improve rescue efficiency, but also maximize the safety of rescuers. However, with the continuous expansion of the application field of drones, the environment and mission types they face have become more complex and changeable. In this context, how to achieve effective mission planning of drones through efficient control strategies has become a hot topic in current research. A complete mission planning system is crucial for drones to accurately perform reconnaissance, monitoring or rescue missions and avoid various potential threats.

[0003] In the field of task allocation, methods based on optimization algorithms are mainly divided into two categories: deterministic optimization and non-deterministic optimization. Deterministic optimization methods rely on clear rules, mathematical models or iterative processes to find the optimal solution. Typical representatives of this type of method include Newton's method and gradient descent method, which use deterministic rules or gradient information to guide the update of parameters in the direction of the optimal solution. Although deterministic methods are more efficient in solving specific problems, when faced with complex problems, especially when there are multiple local optimal solutions, they may fall into local optimality and fail to find the global optimal solution. Non-deterministic optimization methods effectively deal with complex and multi-peak optimization problems by adding randomness to the search. Real-world problems are often difficult to solve with deterministic methods due to their complexity. At this time, metaheuristic algorithms, with their random search mechanism, become a powerful tool for finding the global optimum. However, swarm intelligence algorithms still have limitations in solving the problem of task allocation for drones:

[0004] (1) It is difficult for algorithms to balance exploration and exploitation. Every algorithm will encounter the problem of how to explore and exploit the solution space when searching for the optimal solution. Exploration refers to extensive exploration of the solution space, i.e., breadth search, and exploitation refers to searching for the optimal solution in a known area, i.e., depth search. In order to find the global optimal solution, the algorithm must balance exploration and exploitation to prevent premature convergence.

[0005] (2) The algorithm is not adaptable. According to the NFL theory, no algorithm can solve all optimization problems in the best way. When a drone performs a task, the specific constraints of the task, such as range, obstacles, etc., will directly affect the order in which the drone performs the task. Therefore, when faced with the problem of drone task allocation under different constraints, it is necessary to select the most appropriate optimization algorithm to solve it.

[0006] (3) The algorithm is prone to falling into local optimal solutions. In the later stages of iteration of many swarm intelligence optimization algorithms, the diversity of the population decreases, and the population gathers in the local optimal solution and cannot escape. For example, the Gray Wolf Algorithm, due to its relatively simple position update rules, lacks breadth search in the later stages of iteration, and has insufficient exploration space. Therefore, it is easy to fall into the local optimal solution when facing some complex optimization problems. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the prior art, provide a UAV task allocation method based on the normal cloud model artemisinin optimization algorithm, establish a UAV task allocation model under range constraints, use a chaotic mapping strategy to initialize the artemisinin population, and establish an optimization model by simulating the process of using artemisinin to eradicate malaria parasites in the human body. The normal cloud distribution is introduced to update the historical optimal position in the artemisinin optimization process, enhance the ability to jump out of the local optimum, and finally use it for the UAV task allocation problem to find the best solution.

[0008] The objective of the present invention is achieved through the following technical solution: a method for allocating unmanned aerial vehicle tasks based on a normal cloud model artemisinin optimization algorithm, comprising the following steps:

[0009] S1. Use graph theory to build a model for UAV task allocation, define the objective function and its constraints, and establish a mathematical model for UAV task allocation. The sub-steps are as follows:

[0010] S11. Assume that the task scenario N under consideration is an n×n square area, that is, an n×n matrix, and there are k tasks to be executed in N. The set of all tasks is set to C = {C1, C2, ..., C k}, the drone needs to execute k pending tasks without duplication;

[0011] S12, define that there are m obstacles on the way of the UAV to perform the task, and the set of obstacles is M = {Mi,j}, M contains the position and size of each obstacle in N. The drone needs to avoid these obstacles when performing the mission. In the mission area N, the area of ​​M is represented by 1, and the other areas are represented by 0, that is,

[0012] S13. Define the order in which the drone performs tasks as a matrix X, where X is a 1×k row matrix containing k non-repeating numbers from 1 to k, and X(1,x) and X(1,x+1) indicate that the drone will perform the task at X(1,x+1) after completing the task at X(1,x);

[0013] S14. Define the connection between two tasks as an edge, and set the edge set to E = {e ij},i,j=1,2,...,k,i≠j, where e ij is the line connecting task point i and task point j; edge e ij The length of is the distance between task point i and task point j;

[0014] S15. Define the range of the drone as L. Then the total mileage D required to execute the task corresponding to the matrix X must satisfy D≤L. The calculation formula for the total mileage D is: d mn Indicates the distance between task point m and task point n;

[0015] S16. According to the scenario constraints and the requirements of range minimization, the objective function and its constraints are established as follows:

[0016] minD

[0017]

[0018] S2. Setting the initial parameters of the artemisinin algorithm;

[0019] S3, execute the artemisinin algorithm: first set the parameters for solving the problem and the criteria for stopping iteration; then construct an optimization strategy by simulating three processes of using artemisinin to treat malaria, including three stages: comprehensive elimination stage strategy, local clearance stage strategy, and consolidation stage strategy; after completing the first three stages, use the normal cloud distribution to update the new position, and calculate the fitness value to determine whether it has been updated to the best position;

[0020] S4. Output the global optimal solution, that is, the UAV’s task execution sequence, calculate its fitness value, and draw a UAV task execution graph.

[0021] The beneficial effects of the present invention are as follows: the present invention mainly targets the specific physical object of drones, and abstracts the problem of drones performing multiple tasks in scenarios such as post-disaster searches or material transportation into a task allocation problem. First, a mathematical model of the task allocation problem is established based on knowledge such as graph theory, and then an artemisinin optimization algorithm based on the normal cloud model is designed to solve it. The algorithm combines tent chaotic mapping and the normal cloud model to optimize the randomly generated population, increase randomness and unpredictability, better simulate the complex behavior of populations in natural systems, and improve global search capabilities. At the same time, the algorithm can converge faster and reduce computational costs on large-scale problems. Therefore, the proposed drone task allocation method based on the normal cloud model artemisinin optimization algorithm has obvious advantages when facing complex and large-scale problems. Compared with the prior art, the improvements of the present invention are mainly reflected in the following three aspects:

[0022] (1) A novel UAV task allocation method based on the normal cloud model artemisinin optimization algorithm is proposed. Compared with the ant colony algorithm, particle swarm algorithm and other methods, this invention introduces a newer artemisinin algorithm with better performance, and expands the application scenarios of the artemisinin algorithm;

[0023] (2) The present invention introduces a chaotic mapping strategy in the initialization phase. Compared with some swarm intelligence algorithms that use uniform distribution to randomly generate the initial position of the population, the present invention performs chaotic mapping on the generated initial population position. The chaotic mapping guides the algorithm to optimize the initial search path in the solution space in a nonlinear manner, and the unpredictability of the chaotic mapping can increase the diversity of individuals in the population and avoid the convergence curve from converging prematurely.

[0024] (3) The feasibility of the present invention proves that the combination of artemisinin algorithm with chaotic mapping and normal cloud model can be generalized and used in other swarm intelligence algorithms. In addition, the normal cloud model can balance global and local search, improve the adaptability and multi-objective optimization capability of the algorithm, and thus achieve more efficient and stable UAV swarm task planning and execution in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flow chart of the method for allocating UAV tasks based on the normal cloud model artemisinin optimization algorithm of the present invention;

[0026] Figure 2 is a flow chart of the iterative portion of the present invention;

[0027] Figure 3 It is a task scenario diagram of an embodiment;

[0028] Figure 4 The optimal task execution sequence scheme of the embodiment;

[0029] Figure 5It is a graph of algorithm convergence of the embodiment. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0031] like Figure 1 As shown, a method for allocating unmanned aerial vehicle tasks based on a normal cloud model artemisinin optimization algorithm of the present invention comprises the following steps:

[0032] S1. Use graph theory to build a model for UAV task allocation, define the objective function and its constraints, and establish a mathematical model for UAV task allocation. The sub-steps are as follows:

[0033] S11. Assume that the task scenario N under consideration is an n×n square area, that is, an n×n matrix, and there are k tasks to be executed in N. The set of all tasks is set to C = {C1, C2, ..., C k}, the drone needs to execute k pending tasks without duplication;

[0034] S12, define that there are m obstacles on the way of the drone to perform the mission, simulate the location and size of the post-disaster urban ruins, and the set of obstacles is M = {M i,j}, M contains the position and size of each obstacle in N. The drone needs to avoid these obstacles when performing the mission. In the mission area N, the area of ​​M is represented by 1, and the other areas are represented by 0, that is,

[0035]

[0036] S13. Define the order in which the drone performs tasks as a matrix X, where X is a 1×k row matrix containing k non-repeating numbers from 1 to k, and X(1,x) and X(1,x+1) indicate that the drone will perform the task at X(1,x+1) after completing the task at X(1,x);

[0037] S14. Define the connection between two tasks as an edge, and set the edge set to E = {e mn},m,n=1,2,...,k,m≠n, where e mn is the line connecting task point m and task point n; edge e mn The length of is the distance between task point m and task point n, expressed as d mn ;

[0038] S15. Define the range of the drone as L. Then the total mileage D required to execute the task corresponding to the matrix X must satisfy D≤L. The calculation formula for the total mileage D is: d mn Indicates the distance between task point m and task point n;

[0039] S16. According to the scenario constraints and the requirements of range minimization, the objective function and its constraints are established as follows:

[0040] minD

[0041]

[0042] S2, setting the initial parameters of the artemisinin algorithm; there are the following sub-steps:

[0043] S21, define the initial population size popsize, tent chaos mapping parameter α, maximum number of iterations MaxFEs, and problem dimension dim;

[0044] S22, Initialize population position: conceptualize drug particles as search agents of the algorithm, and the entire set of these search agents constitutes the solution set of the algorithm;

[0045]

[0046] Where T and B are the boundaries of the solution space, R is a set of random number sequences, and its value range is [0,1]; A popsize,dim An element in is a task execution order, and the optimal individual is found through the artemisinin algorithm as the final task execution order.

[0047] S23, use chaos mapping to re-initialize the population position: for the population A initialized in S22 popsize,dim , use the tent chaotic mapping strategy to initialize the population:

[0048]

[0049] Then let the initialized population a i,j =a' i,j , that is, the initial population A after chaotic mapping is obtained popsize,dim .

[0050] S3. Execute the artemisinin algorithm: During the execution of the algorithm, periodic iterative updates are performed. First, set the parameters for solving the problem and the criteria for stopping iterations; then construct an optimization strategy by simulating the three processes of using artemisinin to treat malaria, including three stages: comprehensive elimination stage strategy, using higher doses of artemisinin in the early stage of treatment to eliminate a large number of malarial parasites in the human body; local clearance stage strategy, gradually reducing the drug dose in the middle and late stages of treatment to eliminate the small amount of malarial parasites remaining in the human body; consolidation stage strategy, reducing the possibility of malaria recurrence in the late stage of treatment. And introduce the normal cloud distribution as a new update strategy. After completing the first three stages, use the normal cloud distribution to update the new position, and calculate the fitness value to determine whether it has been updated to the best position; after finding a new solution in each iteration, calculate the fitness value of the position and compare it with the fitness value of the previous position, obtain and save the historical optimal fitness value and the corresponding position, and then update the position to the global optimal solution. This process continues until the preset iteration stop condition is reached.

[0051] Artemisinin algorithm Figure 2 As shown, there are the following sub-steps:

[0052] S31. Comprehensive elimination phase strategy: In the initial stage of malaria treatment, patients are given larger doses of drugs to quickly control the progression of the disease; once artemisinin is absorbed, it is transported to various parts of the human body through the blood and diffuses throughout the body; considering that the drug concentration decreases over time, c is used to represent the attenuation index of the drug concentration in the human body, and a single-compartment model is used to describe it:

[0053]

[0054] C(t)=C0×e (-k×t)

[0055] The variable concentration is the concentration of the drug, and k is the rate constant. Solving the above differential equations yields the drug concentration c. C(t) represents the drug concentration at time t. In this model, the drug concentration C(t) decays exponentially over time. Therefore, the exponent c of the artemisinin drug concentration is calculated as:

[0056]

[0057] In this strategy, assuming that the initial drug concentration is 1 and the drug decay rate is 4, the evaluation process of the algorithm is used to simulate the time course in the model; where f and MaxFEs represent the current iteration number and the maximum iteration number of the algorithm, respectively;

[0058] Since the distribution of drugs in the body is affected by factors such as blood flow, vascular permeability, and the binding affinity between drugs and proteins, as well as the complex structure of the human body, a search model is introduced to simulate the drug diffusion process:

[0059]

[0060] Taking into account the differences in the severity of the patient's condition and the differences in physiological factors, which lead to different medication dosages and durations, patients may spend different durations in this phase. To account for this inherent variability, a probability coefficient K is introduced:

[0061]

[0062] K is used as a probability coefficient, combined with the evaluation progress of the algorithm to simulate the objective scenario. Patients show different reactions and durations according to their individual conditions at this stage. It can be seen that K gradually decreases with the increase of the number of iterations.

[0063] Before the position is updated, a random number r1 is randomly generated, with a value range of [0,1]. When r1<K, the comprehensive elimination phase strategy is executed. In the early stage of iteration, K is large, so the probability of conducting a comprehensive elimination phase is greater, that is, global exploration. In the comprehensive elimination phase, a random number rand is randomly generated again, with a value between [0,1]. When rand<0.5, the first position update strategy is performed:

[0064]

[0065] In the formula represents the jth dimension of the i-th artemisinin particle at time t, It indicates the next time position of the particle, that is, the updated position; when rand ≥ 0.5, the second position update strategy is performed:

[0066]

[0067] in represents the position of the jth dimension of the global optimal individual in the artemisinin particle at time t;

[0068] If r1≥K, the position of the artemisinin particle remains unchanged, that is:

[0069]

[0070] After the initial treatment phase, as the disease is controlled, treatment transitions to the maintenance phase to ensure complete cure of malaria, entering S32;

[0071] S32, local elimination phase strategy: After early treatment, most malaria parasites are eliminated, but a small number of malaria parasites may still exist, so low-dose artemisinin is needed for treatment to ensure eradication of the parasites. Calculate the fitness fit(i) of each artemisinin particle and normalize it:

[0072]

[0073] Fit norm (i) is the normalized fitness value of the ith artemisinin particle, min(fit) is the minimum fitness in the artemisinin population, and max(fit) is the maximum fitness in the artemisinin population. This ensures that individuals with high fitness have a lower probability of updating their positions, retaining excellent individuals to a certain extent, while also having a greater probability of updating the positions of individuals with low fitness, thereby adaptively adjusting the overall quality of the population. In this strategy, artemisinin particles will exchange local information with each other. By understanding the position status of other particles, artemisinin particles can better find and eliminate malaria parasites. In this stage, a random number rand1 is randomly generated, with a value between [0,1]; when rand1<Fit norm (i) When the location is updated (the three location updates are performed one after another. If the second location update is performed, the location obtained in the first place will be overwritten. Similarly, if the third location update is performed, the first two locations will be overwritten. Therefore, if this condition is not met, the first location update will still be maintained):

[0074]

[0075] b1,b2,b3~U(1,popsize),b1≠b2≠b3

[0076] Where d is the coefficient, which takes a random value between [0.1, 0.6]; b1, b2, b3 are random integers between [1, popsize] and are not equal to each other; It represents the position of the jth dimension of the b1th artemisinin particle at time t. This process is mainly The location is updated based on and The relative position of , multiplied by a random weight d, the new position is expressed as

[0077] S33, Consolidation Phase Strategy: In the later stages of treatment, the patient's condition improves, and they may be less vigilant against malaria. A small number of parasites may gradually develop resistance to artemisinin, or enter a dormant stage, after which they begin to attack the human body, causing the disease to relapse. In this stage, artemisinin particles need to simulate this specific situation, assuming that inactive parasites still exist in the human body, to search and remove the parasites. Randomly generate two random values, rand2 and rand3, between [0,1]; when rand2 < 0.05, update the position:

[0078]

[0079] This strategy can make the position update of artemisinin more flexible and avoid falling into local optimality in the later stage.

[0080] When rand3 < 0.2, the position of the artemisinin particle is updated to the best historical position:

[0081]

[0082] This update will The steps are covered, so the convergence accuracy of the algorithm can be improved to a certain extent.

[0083] S34. Introduce the normal cloud model; among the many uncertainties, randomness and fuzziness are undoubtedly the most common attributes. In order to overcome the shortcomings of dealing with uncertainty, the cloud model is proposed to achieve the uncertainty conversion between quantitative description and qualitative concepts. The cloud model is characterized by three mathematical parameters: expectation (E x ), entropy (En) and super entropy (He). En represents E x The entropy of is used to measure the randomness of a qualitative concept, while He is used to measure En. The normal distribution is a common and important probability distribution in nature, so it is more suitable for describing the randomness of search particles. The definition of the normal cloud model is as follows:

[0084] Let V be a qualitative concept on the quantitative domain W. If is a random realization of W, then the certainty of v to V is a random number with a stable tendency. The distribution of cloud droplets v in W is called normal cloud distribution. The forward normal cloud generator generates a cloud droplet when it runs once according to the set digital characteristics until the desired number of cloud droplets is generated.

[0085] In order to improve the convergence accuracy of the AO algorithm in solving this problem, after completing the three stages, the optimal position is taken as the expected value E of the normal cloud model. x Explore. Inspired by the process of normal cloud droplet generation, the position update formula of the new position can be set as:

[0086]

[0087] Where Gnc() is the normal cloud distribution function, Nd is the expected number of cloud droplets. In this update strategy, the range of the updated position of artemisinin particles from the optimal individual is adjusted by the value of En, and the dispersion of the updated position of artemisinin particles is controlled by He. En and He are determined by the following formula:

[0088]

[0089] He=En×10-ξ

[0090] Among them, ω is a random number between [0,1], τ and ξ are random integers between [1,dim]. It can be seen that with the increase of the number of iterations, En gradually shows a decreasing trend, and He is positively correlated with En. Therefore, with the increase of iterations, the step size of artemisinin particle search gradually decreases, the dispersion of the position also gradually decreases, and the convergence accuracy is improved.

[0091] S4. Output the global optimal solution, that is, the UAV’s task execution sequence, calculate its fitness value, and draw a UAV task execution graph.

[0092] In this embodiment, the task scene is a 50×50 square area N, and there are 6 obstacles N={N m},m=1,2,...,6,N m ,m=1,2,...,6 are all rectangular areas, such as Figure 3 As shown. The four vertex coordinates of N1 are {(10,10), (10,25), (15,10), (15,25)}, which is a rectangular area with a length of 15 and a width of 5. The four vertex coordinates of N2 are {(10,40), (12,40), (10,42), (12,12)}, which is a square area with a side length of 2. The four vertex coordinates of N3 are {(20,10), (20,12), (22,10), (22,12)}, which is a square area with a side length of 2. The four vertex coordinates of N4 are {(20,30), 20,45), (20,30), (30,45)}, which is a rectangular area with a length of 15 and a width of 10. The four vertex coordinates of N5 are {(40,15), (40,23), (48,15), (48,23)}, which is a square area with a side length of 8. The four vertex coordinates of N6 are {(40,35), (40,37), (42,35), (42,35)}, which is a square area with a side length of 8. There are 20 tasks to be performed in N, and each task location corresponds to a two-dimensional coordinate. The task scene is as follows Figure 3 As shown, the UAV range constraint L is 250.

[0093] In this example, the initial population size of the optimization model popsize=10000, the tent chaos mapping control parameter α=0.5, the maximum number of iterations MaxFEs=10000, and the problem dimension dim=20.

[0094] According to the above task scenarios and parameter settings, the optimal task allocation sequence is calculated as follows: Figure 4 As shown, the algorithm converges as follows Figure 5As shown in the figure, it can be seen that the allocation sequence converges to the global optimum. The simulation results confirm that the UAV task allocation method based on the artemisinin optimization algorithm of the normal cloud model can achieve the optimal allocation of UAV tasks.

[0095] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A UAV task allocation method based on normal cloud model artemisinin optimization algorithm, characterized in that: The following steps are involved: S1. Use graph theory to build a model for UAV task allocation, define the objective function and its constraints, and establish a mathematical model for UAV task allocation. The sub-steps are as follows: S11. Assume that the task scenario N under consideration is an n×n square area, that is, an n×n matrix, and there are k tasks to be executed in N. The set of all tasks is set to C = {C1, C2, ..., C k }, the drone needs to execute k pending tasks without duplication; S12, define that there are m obstacles on the way of the UAV to perform the task, and the set of obstacles is M = {M i,j }, M contains the position and size of each obstacle in N. The drone needs to avoid these obstacles when performing the mission. In the mission area N, the area of ​​M is represented by 1, and the other areas are represented by 0, that is, S13. Define the order in which the drone performs tasks as a matrix X, where X is a 1×k row matrix containing k non-repeating numbers from 1 to k, and X(1,x) and X(1,x+1) indicate that the drone will perform the task at X(1,x+1) after completing the task at X(1,x); S14. Define the connection between two tasks as an edge, and set the edge set to E = {e ij },i,j=1,2,...,k,i≠j, where e ij is the line connecting task point i and task point j; edge e ij The length of is the distance between task point i and task point j; S15. Define the range of the drone as L. Then the total mileage D required to execute the task corresponding to the matrix X must satisfy D≤L. The calculation formula for the total mileage D is: m=X(1,x),n=X(1,x+1),d mn Indicates the distance between task point m and task point n; S16. According to the scenario constraints and the requirements of range minimization, the objective function and its constraints are established as follows: S2. Setting the initial parameters of the artemisinin algorithm; S3, execute the artemisinin algorithm: first set the parameters for solving the problem and the criteria for stopping iteration; then construct an optimization strategy by simulating three processes of using artemisinin to treat malaria, including three stages: comprehensive elimination stage strategy, local clearance stage strategy, and consolidation stage strategy; after completing the first three stages, use the normal cloud distribution to update the new position, and calculate the fitness value to determine whether it has been updated to the best position; S4. Output the global optimal solution, that is, the UAV’s task execution sequence, calculate its fitness value, and draw a UAV task execution graph.

2. The method for allocating unmanned aerial vehicle tasks based on the normal cloud model artemisinin optimization algorithm according to claim 1 is characterized in that: The S2 has the following sub-steps: S21, define the initial population size popsize, tent chaos mapping parameter α, maximum number of iterations MaxFEs, and problem dimension dim; S22, Initialize population position: conceptualize drug particles as search agents of the algorithm, and the entire set of these search agents constitutes the solution set of the algorithm; Where T and B are the boundaries of the solution space, R is a set of random number sequences, and its value range is [0,1]; A popsize,dim An element in is a task execution order; S23, use chaos mapping to re-initialize the population position: for the population A initialized in S22 popsize,dim , use the tent chaotic mapping strategy to initialize the population: Then let the initialized population a i,j =a' i,j .

3. The method for allocating unmanned aerial vehicle tasks based on the normal cloud model artemisinin optimization algorithm according to claim 1 is characterized in that: The S3 has the following sub-steps: S31, comprehensive elimination phase strategy: first randomly generate a random number r1, the value range is [0,1]; when r1 < K, execute the comprehensive elimination phase strategy; in the comprehensive elimination phase, randomly generate another random number rand, the value is between [0,1], when rand < 0.5, execute the first position update strategy: In the formula represents the jth dimension of the i-th artemisinin particle at time t, It indicates the next time position of the particle, that is, the updated position; when rand ≥ 0.5, the second position update strategy is performed: in represents the position of the jth dimension of the global optimal individual in the artemisinin particle at time t; If r1≥K, the position of the artemisinin particle remains unchanged, that is: After the initial treatment phase, as the disease is controlled, treatment transitions to the maintenance phase to ensure complete cure of malaria, entering S32; S32, local elimination phase strategy: After early treatment, most malaria parasites are eliminated, but a small number of malaria parasites may still exist, so low-dose artemisinin is needed for treatment to ensure eradication of the parasites. Calculate the fitness fit(i) of each artemisinin particle and normalize it: Fit norm (i) is the normalized fitness value of the i-th artemisinin particle, min(fit) is the minimum fitness in the artemisinin population, and max(fit) is the maximum fitness in the artemisinin population; In this stage, a random number rand1 is randomly generated, with a value between [0,1]; when rand1<Fit norm (i) When the location is updated: b1,b2,b3~U(1,popsize),b1≠b2≠b3 Where d is the coefficient, which takes a random value between [0.1, 0.6]; b1, b2, b3 are random integers between [1, popsize] and are not equal to each other; It represents the position of the jth dimension of the b1th artemisinin particle at time t. This process is mainly The location is updated based on and The relative position of , multiplied by a random weight d, the new position is expressed as S33, consolidation phase strategy: randomly generate two random values ​​rand2 and rand3, with values ​​between [0,1]; when rand2 < 0.05, update the position: When rand3 < 0.2, the position of the artemisinin particle is updated to the best historical position: This update will Steps covered; S34. Introduce the normal cloud model. The definition of the normal cloud model is as follows: Let V be a qualitative concept on the quantitative domain W. If is a random realization of W, then the certainty of v to V is a random number with a stable tendency; the distribution of cloud droplets v in W is called normal cloud distribution, and the positive normal cloud generator generates a cloud droplet when it runs once according to the set digital characteristics until the expected number of cloud droplets is generated; After completing the three stages, the optimal position is the expected value E of the normal cloud model. x Explore; inspired by the process of normal cloud droplet generation, the position update formula of the new position is set as: Where Gnc() is the normal cloud distribution function, Nd is the expected number of cloud droplets. In this update strategy, the range of the updated position of artemisinin particles from the optimal individual is adjusted by the value of En, and the dispersion of the updated position of artemisinin particles is controlled by He. En and He are determined by the following formula: He=En×10 -ξ Among them, ω is a random number between [0,1], τ and ξ are random integers between [1,dim].

Citation Information

Cited By

  • Medical image segmentation method and system for chronic obstructive pulmonary disease

    CN119904480A

  • Medium and long term wind speed prediction method and system

    CN120409303A

  • A medium- and long-term wind speed prediction method and system

    CN120409303B