A Multi-Objective Unmanned Aerial Vehicle Control Method

By combining genetic algorithms, K-means clustering, and particle swarm optimization, a multi-objective hybrid genetic algorithm was developed to solve the problems of unstable clustering results and low efficiency in UAV formation, enabling rapid and accurate adjustment of UAV formations.

CN116088544BActive Publication Date: 2026-11-10NANTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211631382.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-12-19
Publication Date
2026-11-10
Estimated Expiration
2042-12-19

AI Technical Summary

Technical Problem

Existing UAV position adjustment algorithms suffer from unstable clustering results and low efficiency during formation flight. Genetic algorithms have poor local search capabilities, K-means clustering is sensitive to initial cluster centers and easily gets trapped in local optima, and particle swarm optimization is also prone to getting trapped in local optima, resulting in inaccurate UAV formation adjustment.

Method used

By combining genetic algorithms with K-means clustering and particle swarm optimization, the initial position of the UAV is determined through AOA passive localization. The position of the UAV at the receiving end is iteratively adjusted using a multi-objective hybrid genetic algorithm. By combining multiple algorithms to optimize the UAV formation model, the stability and efficiency of the UAV formation are improved.

Benefits of technology

It improves the stability and overall performance of clustering results for UAV formation adjustment, enhances local search capabilities, reduces problems of low computational efficiency and poor convergence accuracy, and enables rapid and accurate adjustment of UAV formations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116088544B_ABST
    Figure CN116088544B_ABST
Patent Text Reader

Abstract

This application discloses a multi-objective unmanned aerial vehicle (UAV) control method, comprising: determining the transmission coordinates of multiple transmitting UAVs in a first coordinate system; obtaining the initial coordinates of the receiving UAVs in the first coordinate system based on the angle between the line connecting the receiving UAV, the transmitting UAV, and the origin; obtaining a UAV formation optimization model based on the initial coordinates of the multiple receiving UAVs and multiple preset conditions; and determining the adjustment position of the receiving UAVs whose positions have shifted based on the UAV formation optimization model and a multi-objective hybrid genetic algorithm. This application determines the initial position of the UAVs using an azimuth-based passive positioning (AOA) method, sets a maximum number of iterations, and combines a genetic algorithm with multiple algorithms such as mean clustering and particle swarm optimization. The combined multi-objective hybrid genetic algorithm is used to iteratively adjust the receiving UAVs, enabling the UAV formation to reach a predetermined adjustment position.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application belongs to the field of unmanned aerial vehicle (UAV) technology, specifically relating to a multi-target UAV control method. Background Technology

[0002] Due to their small size, light weight, and ease of operation, drones have been widely used in both military and civilian fields in recent years. When drone swarms fly in formation, the positions of individual drones may deviate, necessitating positioning and adjustment. Drone positioning primarily employs two methods: active and passive positioning. Passive positioning, with its higher accuracy, better concealment, wider coverage, longer operating distance, and adaptability to complex environments, has gradually become the mainstream positioning system. Current passive positioning methods mainly utilize factors such as the signal transmission distance (RSS), the time of arrival (TOA) between the transmitter and receiver, the time difference of arrival (TDOA) of the signal at the receiver, the direction of arrival (DOA) of the radiated signal to estimate the location of the radiation source, the frequency difference of arrival (FDOA) of signals received at different receiving stations, and the lateral angle (AOA). Among these, AOA passive positioning has fewer errors, higher accuracy, and only requires angle measurement, avoiding the influence of the external environment on the drone, making it more advantageous among positioning methods.

[0003] The location of the receiving UAV is determined using AOA passive positioning. To maintain the UAV formation, adjustments need to be made for UAVs whose positions deviate. Researchers have attempted various algorithms to address this problem. Among existing technologies, genetic algorithms, K-means clustering, and particle swarm optimization (PSO) are prominent, but all three have certain limitations. Genetic algorithms can improve overall performance with their flexible fitness function and can perform global searches, but they have poor local search capabilities, low computational efficiency, and poor convergence accuracy. K-means clustering has strong local search capabilities, is efficient, and easy to implement, but the K value is difficult to determine, and the initial cluster centers are sensitive to getting trapped in local optima. PSO is computationally simple, efficient, and converges quickly, but it is also prone to getting trapped in local optima.

[0004] When adjusting the position of a drone, all three algorithms have certain limitations. This invention combines genetic algorithm with K-means clustering and particle swarm optimization, which can effectively improve the final clustering results, making the clustering results more stable, improving overall performance and enhancing the algorithm's local search capability. At the same time, it avoids low computational efficiency and poor convergence accuracy, further accelerating the algorithm's running speed. Summary of the Invention

[0005] This application provides a multi-objective-based UAV control method to solve the technical problems of unstable clustering results and low efficiency in traditional UAV position adjustment.

[0006] To solve the above-mentioned technical problems, this application adopts the following technical solution: a multi-target-based unmanned aerial vehicle (UAV) control method, comprising:

[0007] A first coordinate system is established with the position of the first UAV as the origin, and the transmission coordinates of multiple transmitting UAVs used to transmit signals are determined in the first coordinate system.

[0008] Based on the angle between the line connecting the receiving UAV and the transmitting UAV used to receive signals and the origin, the initial coordinates of the receiving UAV in the first coordinate system are obtained.

[0009] Based on the initial coordinates of multiple receiving drones and multiple preset conditions, an optimized drone formation model is obtained.

[0010] Based on the UAV formation optimization model and multi-objective hybrid genetic algorithm, the adjustment position of the receiving UAV whose position has shifted is determined.

[0011] Furthermore, a method for obtaining the initial coordinates of the receiving UAV in the first coordinate system based on the angle between the line connecting the receiving UAV used to receive the signal, the transmitting UAV, and the origin includes:

[0012] Based on the first / second / third lines connecting the receiving drone to the first / second transmitting coordinates and the origin, the first / second included angle between the third line and the first / second line, and the third included angle between the first line and the second line are obtained; wherein, the third included angle is the sum of the first included angle and the second included angle.

[0013] Based on the first / second included angle and the first / second connecting line, determine the first / second radius of the first / second circular curve where the first / second included angle and the first / second connecting line lie;

[0014] Based on the first / second radius, the first / second transmission coordinates, and the origin, determine the coordinates of the first / second center of the first / second circular curve;

[0015] Based on the first / second radius and the coordinates of the first / second circle center, determine the equation of the first / second circular curve;

[0016] Based on the first / second curve equations and the screening conditions, the initial coordinates of the receiving UAV are determined.

[0017] Furthermore, the method for determining the initial coordinates based on the first / second curve equations and screening conditions includes:

[0018] Based on the equations of the first and second curves, obtain all intersection points between the first and second circular curves; based on all intersection points and the filtering conditions, determine the initial coordinates of the receiving UAV.

[0019] Furthermore, a method for obtaining a drone formation optimization model based on the initial coordinates of multiple receiving drones and multiple preset conditions includes:

[0020] Based on the initial coordinates of the receiving drone and the first preset condition, the adjustment position of the receiving drone is determined to be within the third circular curve, and the formation model for the receiving drone to adjust its position is obtained; based on the third circular curve and the second preset condition, the preset radius of the third circular curve is determined; based on the formation model and the preset radius, the drone formation optimization model is obtained.

[0021] Furthermore, based on the UAV formation optimization model and multi-objective hybrid genetic algorithm, a method for determining the adjusted position of the receiving UAV that has experienced a position shift includes:

[0022] Based on the drone formation optimization model and the initial coordinates of the target to be optimized, the chromosome model is determined;

[0023] Determine the fitness function value for each chromosome in the chromosome model;

[0024] A population is randomly generated from the chromosome model, and the population is initialized. Initialization samples are obtained and the maximum number of iterations is preset.

[0025] In response to a chromosome not reaching the preset maximum number of iterations, determine whether each chromosome meets the fitness function value;

[0026] In response to chromosomes meeting fitness function values, crossover and mutation iterations based on a multi-objective hybrid genetic algorithm are performed on the first type of chromosomes that meet the fitness function values ​​to obtain the first type of chromosome population;

[0027] In response to chromosomes that do not conform to the fitness function value, K-means clustering is performed on the second type of chromosomes that do not conform to the fitness function value to obtain the second type of chromosome population;

[0028] Based on the first type of chromosome population and the second type of chromosome population, obtain a new population, and repeat the above steps until the chromosome reaches the preset maximum number of iterations.

[0029] Based on the chromosomes of the new population, the position of the receiving drone is iteratively adjusted until the receiving drone reaches the adjusted position.

[0030] Furthermore, in response to chromosomes meeting fitness function values, the method for obtaining the first type of chromosome population by performing crossover and mutation iterations based on a multi-objective hybrid genetic algorithm on the first type of chromosomes meeting fitness function values ​​includes:

[0031] Based on the first type of chromosomes, crossover and mutation substitutions are performed on the first type of chromosomes to obtain the offspring population. Based on the first type of chromosomes and the offspring population, the chromosomes in the offspring population are compared with the first type of chromosomes, and the optimal chromosome is selected to form the first type of chromosome population.

[0032] Furthermore, methods for cross-substitution and mutation substitution include:

[0033] Two chromosomes are randomly selected and some genes of the two chromosomes are exchanged to obtain a type of replacement chromosome; the above steps are repeated multiple times to obtain a first generation population containing type one replacement chromosome; based on the first generation population, a gene locus of a chromosome is randomly selected and a random gene value in the gene locus is replaced with another gene value in the gene locus to obtain a type two replacement chromosome; the above steps are repeated multiple times to obtain a generation population containing type two replacement chromosomes.

[0034] Furthermore, methods for K-means clustering of second-class chromosomes that do not meet the fitness function value include:

[0035] Based on the second type of chromosomes that did not meet the fitness function value requirements, feature extraction was performed on the remaining second type of chromosomes that did not meet the fitness function value requirements, a feature matrix was constructed, and normalization was performed to obtain chromosome feature samples;

[0036] Based on chromosome feature samples, multiple chromosome feature samples were selected as initial cluster centers;

[0037] Determine the distance between each second-class chromosome and the initial cluster center;

[0038] The second type of chromosome is assigned to the nearest initial cluster center;

[0039] Determine the mean of each initial cluster center, and replace the initial cluster centers with the new mean to obtain new cluster centers;

[0040] In response to the change in cluster center, chromosomes form new clusters, and at least one chromosome is selected from the clusters as a particle for PSO restoration to obtain the second type of chromosome population.

[0041] Furthermore, a method for selecting at least one chromosome from the cluster partition as a particle for PSO reconstruction to obtain the second type of chromosome population includes:

[0042] Multiple particles selected from the clustering are used to form an initial particle swarm; the initial position and initial velocity of each particle in the initial particle swarm are randomly set; the fitness value of the particles is determined; the velocity and position of the particles are adjusted based on the historical best position and the global best position of the updated particles; the initial particle swarm is updated iteratively to obtain the second optimal chromosome; it is determined whether the number of update iterations has reached the preset maximum number of iterations; if the number of update iterations has not reached the preset maximum number of iterations, the above steps are repeated; if the number of update iterations has reached the preset maximum number of iterations, the second optimal chromosome is the optimized chromosome, forming the second type of chromosome population.

[0043] The beneficial effects of this application are as follows: This application determines the initial position of the UAV through the azimuth-based passive positioning method, sets a maximum number of iterations, and combines genetic algorithms with multiple algorithms such as mean clustering and particle swarm optimization. The combined multi-objective hybrid genetic algorithm is used to iteratively adjust the UAVs receiving the signal, enabling the UAV formation to reach the predetermined adjustment position. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating an embodiment of the multi-target-based unmanned aerial vehicle (UAV) control method of this application;

[0045] Figure 2 yes Figure 1 A schematic diagram of the structure of an embodiment of the first coordinate system in the diagram;

[0046] Figure 3 yes Figure 1 A flowchart illustrating step S2 of an embodiment;

[0047] Figure 4 yes Figure 1 A flowchart illustrating another embodiment of step S2 in the process;

[0048] Figure 5 yes Figure 3 A flowchart illustrating step S25 of an embodiment;

[0049] Figure 6 yes Figure 1 A flowchart illustrating step S3 of an embodiment;

[0050] Figure 7 yes Figure 1 A flowchart illustrating step S4 in one embodiment;

[0051] Figure 8 yes Figure 7 A flowchart illustrating step S45 of an embodiment;

[0052] Figure 9 yes Figure 7A flowchart of an embodiment of step S451;

[0053] Figure 10 yes Figure 9 A flowchart illustrating step S46 of an embodiment;

[0054] Figure 11 yes Figure 10 A flowchart illustrating step S466 of an embodiment;

[0055] Figure 12 It is a structural diagram of the multiple undetermined final intersection points of the application;

[0056] Figure 13 This is a structural diagram of the application for the position adjustment of multiple drones. Detailed Implementation

[0057] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0058] See Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the multi-objective unmanned aerial vehicle (UAV) control method of this application. The UAV control method includes:

[0059] Step S1. Establish a first coordinate system based on the position of the first UAV as the origin, and determine the transmission coordinates of multiple transmitting UAVs used to transmit signals in the first coordinate system.

[0060] See Figure 2 , Figure 2 yes Figure 1 A schematic diagram of the structure of an embodiment of the first coordinate system in the diagram.

[0061] Taking the location of the first UAV O at the center of the circle as the origin, Establish the first coordinate system with the x-axis as the positive direction.

[0062] Step S2. Based on the angle between the line connecting the receiving UAV and the transmitting UAV used to receive the signal and the origin, obtain the initial coordinates of the receiving UAV in the first coordinate system.

[0063] participate Figure 3-4 , Figure 3 yes Figure 1 A flowchart illustrating step S2 of an embodiment. Figure 4 yes Figure 1 A flowchart illustrating another embodiment of step S2 is provided. The specific method of step S2 described above is as follows:

[0064] Step S21. Based on the first / second / third lines connecting the receiving UAV to the first / second transmitting coordinates and the origin, obtain the first / second included angle between the third line and the first / second line, and the third included angle between the first line and the second line; wherein, the third included angle is the sum of the first included angle and the second included angle.

[0065] For details, please refer to Figure 2 Determine the sending drone The coordinates of the origin O and the receiving drone With the sending drone, The angle formed by the lines connecting the origins O is used to determine the location of all receiving drones based on the known coordinates and angles using the AOA passive positioning method. The initial coordinates of the S-type launcher UAV located on the circumference are determined. coordinates ; Receiver drone Each with the launching drone , and the origin The lines form three included angles, which are defined as the first included angle. The second included angle The third included angle ,in It is the largest angle and .

[0066] Step S22. Based on the first / second included angle and the first / second connecting line, determine the first / second radius of the first / second circular curve where the first / second included angle and the first / second connecting line are located.

[0067] Specifically, taking the first launch terminal UAV as an example The coordinates of the origin O and the first angle Determine the first circular curve , That is, the first included angle. First circular curve The length of the chord above corresponds to a central angle of 2. According to the first included angle and Calculate the first circular curve First radius The specific calculation formula (1) is as follows:

[0068] (1);

[0069] Similarly, using a second-launcher drone... The coordinates of the origin O and the second angle Determine the second circular curve According to the second included angle as well as Calculate the second circular curve Second radius .

[0070] Step S23. Based on the first / second radius, the first / second transmission coordinates, and the origin, determine the coordinates of the first / second center of the first / second circular curve.

[0071] Specifically, based on the first circular curve superior The coordinates of O and the first radius Establish the first circular curve The standard equation for is given by the following formula (2):

[0072] (2);

[0073] Similarly, establish a second circular curve. The standard equation;

[0074] Solve the system of equations in formula (2) to obtain the center of the circle. The relationship between the horizontal and vertical coordinates, i.e., formula (3):

[0075] (3);

[0076] Similarly, the center of the circle is obtained. The relationship between the horizontal and vertical coordinates.

[0077] Substituting the above relationship between the x and y coordinates into the original equation, we can solve it using the quadratic formula. Find the center of the circle From the coordinate expression, we can calculate:

[0078] The expression:

[0079] The expression: , ;

[0080] in, , c .

[0081] Through solving, it was found that There are two solutions, therefore the corresponding There are two interpretations, hence the center of the circle. There are two coordinates: the first center of the circle. and the first center of circle number two .

[0082] Similarly, based on the circle Known points on The coordinates and radius of O Determine the center of the circle , Coordinates: Center of the second circle (number 1) and the second center of circle number two .

[0083] Step S24. Based on the first / second radius and the coordinates of the first / second circle center, determine the equation of the first / second circular curve.

[0084] Specifically, based on the center of the circle , , , The coordinates and the corresponding first radius. Second radius Calculate all information about the circle Harmony The equation is as follows:

[0085] round The equation: ;

[0086] round The equation: ;

[0087] round The equation: ;

[0088] round The equation: .

[0089] Step S25. Determine the initial coordinates of the receiving UAV based on the first / second curve equations and the screening conditions.

[0090] Specifically, Figure 5 yes Figure 3 A flowchart illustrating an embodiment of step S25. The specific method of step S25 is as follows:

[0091] Step S251. Based on the equations of the first and second curves, obtain all intersection points between the first and second circular curves.

[0092] Specifically, based on the circle mentioned above ,round ,round ,round Solve the equations pairwise, and use the quadratic formula to find all possible intersection points. , Let o represent the number of intersection points, where Intersection Coordinate expression:

[0093] The expression:

[0094] ;

[0095] The expression: ;

[0096] in, ;

[0097] -2 ;

[0098] ;

[0099] According to the above equation, a total of eight intersection points are obtained.

[0100] Step S252. Determine the initial coordinates of the receiving UAV based on all intersections and filtering conditions.

[0101] Specifically, filter criterion 1: That is, the third included angle is the sum of the first included angle and the second included angle.

[0102] Exclude intersections that do not meet screening condition 1, and continue to investigate the points that meet the condition.

[0103] Selection criterion 2: Since the drone transmitting the signal will not deviate, i.e., the intersection point... Not with Since these points coincide, the second filtering condition is:

[0104] ;

[0105] Exclude intersections that do not meet condition 2, and continue to investigate the points that meet the condition.

[0106] Screening criterion 3: The location of the drone receiving the signal is only slightly off, i.e., the intersection point. It is the point closest to the circumference of the circle, hence the definition It is the shortest distance from the intersection point to the circle O. If the distance from the intersection point to the center O of the circle is the shortest distance, then filter condition 3 is expressed as:

[0107] , ;

[0108] Calculate the distances from the remaining intersection points to the circle O, and determine the point with the closest distance; determine the final point that meets the above conditions. The coordinates of the final point. And so on, to obtain all the final points. The initial coordinates.

[0109] Step S3. Based on the initial coordinates of the receiving UAV and the first preset condition, determine that the adjustment position of the receiving UAV is located in the third circular curve, and obtain the formation model of the receiving UAV for position adjustment.

[0110] See Figure 6 , Figure 6 This is a flowchart illustrating one embodiment of step S3. The specific method of step S3 is as follows:

[0111] Step S31. Based on the initial coordinates of the receiving UAV and the first preset condition, determine that the adjustment position of the receiving UAV is located in the third circular curve, and obtain the formation model of the receiving UAV for position adjustment.

[0112] Specifically, the first preset condition in the optimal solution is that the UAV should ensure the shortest possible movement distance when adjusting its position. The initial position of the l-th UAV that needs to replan its trajectory is defined as... The final position after adjustment is Construct and adjust the distance target :

[0113] .

[0114] In the optimal solution, the first preset condition 2 is that the receiving UAV should tend to be circular during the iteration process when adjusting its position. Simultaneously, the coordinates of the l-th UAV that needs to replan its trajectory after the i-th iteration are defined as follows: ,in Therefore, an interpolation target is constructed. :

[0115] .

[0116] Step S32. Based on the third circular curve and the second preset condition, determine the preset radius of the third circular curve.

[0117] Specifically, when constructing the drone formation adjustment and optimization model, the optimal solution must also ensure that the drones are evenly distributed on the circumference, meaning that the angle formed by connecting any two adjacent drones to the center of the circle is 40°. The coordinates of any two adjacent drones are defined as follows: , The angle formed by connecting them to the coordinates of the circle's center is... Therefore, in the second preset condition for:

[0118] .

[0119] When constructing the drone formation adjustment and optimization model, in addition to considering the uniform distribution of drones on the circumference, it is also necessary to consider that the distance between all drones on the circumference and the drone at the center of the circle is consistent, that is, the final positioning coordinates of all drones fall on the same circle. The final positioning coordinates of the g-th drone are defined as: , The final location coordinates of the h-th drone are: , Then the second preset condition can be determined. for:

[0120] .

[0121] Step S33. Based on the formation model and the preset radius, obtain the UAV formation optimization model.

[0122] Specifically, based on the optimization objective and and preset radius and Based on the optimization objectives and constraints, a multi-objective optimization model M for UAV formation adjustment is established.

[0123] Step S4. Based on the UAV formation optimization model and the multi-objective hybrid genetic algorithm, determine the adjustment position of the receiving UAV whose position has shifted.

[0124] Specifically, a multi-objective hybrid genetic algorithm is used to iteratively adjust the positions of the receiver drones that have shifted, so that the ten receiver drones are in an ideal distribution state, and the ideal formation of the receiver drone formation and the adjusted positions of the drones are determined.

[0125] See Figure 7 , Figure 7 This is a flowchart illustrating one embodiment of step S4. The specific method of step S4 is as follows:

[0126] Step S41. Determine the chromosome model based on the UAV formation optimization model and the initial coordinates of the target to be optimized.

[0127] Specifically, the optimization target, i.e., any UAV in the formation that has deviated, and the target optimization model are transformed into a form like... Chromosomes. Among them, and This represents the coordinates of the constrained point in the search space, where m represents the constrained point in the search space. and This represents the fitness value corresponding to the objective optimization model.

[0128] Step S42. Determine the fitness function value for each chromosome in the chromosome model.

[0129] Specifically, based on the fitness function values ​​of each chromosome in the target optimization model... , .

[0130] Step S43. Randomly generate a population from the chromosome model, initialize the population, obtain initialization samples, and preset the maximum number of iterations.

[0131] Specifically, the transformed chromosomes are randomly used to generate a population. A completely random initialization method is used for the population. Perform initialization to obtain the initial sample. Set a maximum number of iterations n, and evaluate whether the preset maximum number of iterations n has been reached. If the preset maximum number of iterations n has been reached, stop iterating and obtain the final result. If the preset maximum number of iterations n has not been reached, calculate the chromosome fitness function value. , .

[0132] Step S44. In response to the chromosome not reaching the preset maximum number of iterations, determine whether each chromosome meets the fitness function value.

[0133] Specifically, when the number of chromosome iterations is less than the preset maximum number of iterations n, if the chromosome meets the fitness function value requirements, step S45 is executed; if the chromosome does not meet the fitness function value requirements, step S46 is executed.

[0134] Step S45. In response to chromosomes meeting fitness function values, perform crossover and mutation iterations based on a multi-objective hybrid genetic algorithm on the first type of chromosomes that meet the fitness function values ​​to obtain the first type of chromosome population.

[0135] For details, please refer to Figure 8 , Figure 8 This is a flowchart illustrating one embodiment of step S45. The specific method of step S45 is as follows:

[0136] Step S451. Based on the first type of chromosome, perform crossover substitution and mutation substitution on the first type of chromosome to obtain the offspring population.

[0137] Continue reading Figure 7 The aforementioned methods of cross-substitution and mutation substitution include:

[0138] Step S4511. Randomly select two chromosomes and exchange some genes between the two chromosomes to obtain a type of replacement chromosome.

[0139] Specifically, two chromosomes are randomly selected. and Two crossover points were randomly generated in these two chromosomes. , ,in The two chromosomes located at arrive genes and The chromosomes are swapped, forming two new types of replacement chromosomes. and .

[0140] Step S4512. Repeat step S4511 multiple times to obtain the first offspring population containing a type of substituted chromosome.

[0141] Specifically, following the above method, the chromosomes are crossed n times, and the 2n chromosomes generated after the crossing are used to form the first offspring population.

[0142] Step S4513. Based on the first generation population, randomly select a gene locus of a chromosome, and replace one random gene value in the gene locus with another gene value in the gene locus to obtain a second-replacement chromosome.

[0143] Specifically, a chromosome is randomly selected. A gene locus is randomly selected on the chromosome, and the gene value at that locus is... , The change occurred, resulting in a new type II replacement chromosome. .

[0144] Step S4514. Repeat step S4513 multiple times to obtain a progeny population containing two types of replacement chromosomes.

[0145] Specifically, following the above method, the offspring population is repeatedly mutated to generate a new offspring population.

[0146] Step S452. Based on the first type of chromosomes and the offspring population, compare the chromosomes in the offspring population with the first type of chromosomes, select the optimal chromosome, and form the first type of chromosome population.

[0147] Specifically, the original chromosomes and The offspring population generated after crossover is selected, and the two best chromosomes are chosen from the newly formed population as the next generation chromosomes.

[0148] Mutation and replacement: replacing the mutated chromosome The chromosome is placed into a set and compared with the worst type 1 chromosome. If the chromosome is not better than the worst chromosome, it is not replaced; otherwise, the mutated chromosome is used to replace the worst chromosome.

[0149] The newly formed population is iteratively replaced repeatedly until the optimal chromosome is obtained, forming the first type of chromosome population.

[0150] Step S46. In response to chromosomes not conforming to the fitness function value, perform K-means clustering on the second type of chromosomes that do not conform to the fitness function value to obtain the second type of chromosome population.

[0151] For details, please refer to Figure 10 , Figure 10 This is a flowchart illustrating one embodiment of step S46. The specific method of step S46 is as follows:

[0152] Step S461. Based on the second type of chromosomes that did not meet the fitness function value requirements, extract features from the second type of chromosomes, construct a feature matrix, perform normalization processing, and obtain chromosome feature samples;

[0153] Step S462. Based on chromosome feature samples, select multiple chromosome feature samples as initial cluster centers.

[0154] Step S463. Determine the distance between each second-class chromosome and the initial cluster center.

[0155] Step S464. Assign the second type of chromosome to the nearest initial cluster center.

[0156] Step S465. Determine the mean of each initial cluster center, and replace the initial cluster centers with the new mean to obtain new cluster centers.

[0157] Step S466. In response to the change in cluster center, chromosomes form new clusters, and at least one chromosome is selected from the clusters as a particle for PSO restoration to obtain the second type of chromosome population.

[0158] For details, please refer to Figure 11 , Figure 11 for Figure 10 A flowchart illustrating an embodiment of step S466. The specific method of step S466 is as follows:

[0159] Step S4661. Form an initial particle swarm from multiple particles selected from the clustering partition. .

[0160] Step S4662. Randomly set the initial position and initial velocity of each particle in the initial particle swarm.

[0161] Specifically, initialize the particle swarm. That is, randomly setting the initial position of the particles. and initial velocity .

[0162] Step S4663. Determine the fitness value of the particle. .

[0163] Step S4664. Based on the updated best historical position of the particle. and the best position in the world Adjust the speed and position of the particles.

[0164] Specifically, the best historical position of the updated particle. and the best position in the world Furthermore, the velocity and position of the particles are adjusted, and for particle d, its fitness function value is used. and individual optimal value Comparison, if Then use replace Furthermore, the fitness function value of each particle is... With global optimum Compare, if Then use replace .

[0165] Step S4665. Update the initial particle swarm. Obtain the second optimal chromosome; determine whether the number of update iterations has reached the preset maximum number of iterations n.

[0166] Step S4666. In response to the fact that the number of update iterations has not reached the preset maximum number of iterations n, repeat the above steps S4663-S4665.

[0167] Step S4667. In response to the number of update iterations reaching the preset maximum number of iterations, the second optimal chromosome is the optimized chromosome, forming the second type of chromosome population.

[0168] Specifically, if the maximum number of iterations n has been reached, the returned second-best restored chromosome is used as the optimized chromosome to form the second type of chromosome population.

[0169] Step S47. Obtain a new population based on the first type of chromosome population and the second type of chromosome population. Repeat the above steps until the chromosome reaches the preset maximum number of iterations.

[0170] For details, please refer to [link / reference]. Figure 6The chromosomes resulting from crossover mutations and those restored by K-means clustering and PSO were combined to form a new population. To create a new population Return to step S44 and repeat the iteration until the maximum number of iterations n is reached.

[0171] Step S48. Based on the chromosomes of the new population, iteratively adjust the position of the receiving drone until the receiving drone reaches the adjusted position.

[0172] Example 1

[0173] See Figures 12-13 , Figure 12 This is a structural diagram of multiple pending final intersection point locations. In this embodiment, the radius of the pre-set circular formation of UAVs is 100 meters. A rectangular coordinate system is established with the central UAV (i.e., the center of the formation O) as the origin and OB as the X-axis. The coordinates of the signal transmitting UAVs O, A, and B are (0,0), (17.365, 98.481), and (100, 0), respectively. The direction information received by the signal receiving UAV is... , , .

[0174] Assume that after the UAVs are positioned in a circular formation, the polar coordinates of each UAV are (0,0), (100,0), (98,40.10), (112,80.21), (105,119.75), (98,159.86), (112,199.96), (105,240.07), (98,280.17), and (112,320.28).

[0175] Figure 12 The positions of the transmitting UAV and the eight intersection points obtained using the circle equation in this embodiment are: (-44.9999, 59.9999), (3.45e-13, -2.03e-13), (38.0832, 12.9756), (-2.57e-13, -1.56e-13), (-35.6089, 36.5183), (5.61e13, -3.30e-13), (49.3945, 13.7001), and (-1.98e-13, -1.19e-13). It can be seen that intersection points 2, 4, 6, and 8 have minimal errors relative to the center O (0, 0), therefore these four intersection points can be considered to coincide with the center O. Intersection points 3 and 7 do not meet the selection criteria. Therefore, it cannot be used as an optional intersection point. For intersection point 1 and intersection point 5, the distances to the center O of the circle are calculated using the two-point distance formula as 74.9999 and 51.0057 respectively. Therefore, intersection point 1 is closer to the circumference of the circle, thus determining the position of the passive signal receiving UAV as intersection point 1, with coordinates (-49.9999, 59.9999).

[0176] Figure 13 The iterative adjustment path diagram for the 9 drones in this embodiment is shown, and the final adjustment of the circular formation radius of the drones is 98.2 meters.

[0177] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A multi-target-based unmanned aerial vehicle (UAV) control method, characterized in that, include: A first coordinate system is established with the position of the first UAV as the origin, and the transmission coordinates of multiple transmitting UAVs used for transmitting signals are determined in the first coordinate system. Based on the angle between the line connecting the receiving drone used to receive the signal, the transmitting drone, and the origin, the initial coordinates of the receiving drone in the first coordinate system are obtained. Based on the initial coordinates of multiple receiving drones and multiple preset conditions, a drone formation optimization model is obtained; Based on the aforementioned UAV formation optimization model and multi-objective hybrid genetic algorithm, the adjusted position of the receiving UAV whose position has shifted is determined; The step of determining the adjusted position of the receiving drone that has experienced a position shift, based on the drone formation optimization model and the multi-objective hybrid genetic algorithm, includes: Based on the aforementioned UAV formation optimization model and the initial coordinates of the target to be optimized, the chromosome model is determined; Determine the fitness function value for each chromosome in the chromosome model; A population is randomly generated from the chromosome model, and the population is initialized. Initialization samples are obtained and the maximum number of iterations is preset. In response to the fact that the chromosome has not reached the preset maximum number of iterations, it is determined whether each chromosome meets the fitness function value; In response to the chromosomes conforming to the fitness function value, crossover and mutation iterations based on a multi-objective hybrid genetic algorithm are performed on the first type of chromosomes conforming to the fitness function value to obtain a first type of chromosome population, including: Based on the first type of chromosomes, crossover and mutation substitutions are performed on the first type of chromosomes to obtain the offspring population; Based on the first type of chromosomes and the offspring population, the chromosomes in the offspring population are compared with the first type of chromosomes, and the optimal chromosomes are selected to form the first type of chromosome population. In response to the chromosome not conforming to the fitness function value, K-means clustering is performed on the second type of chromosomes that do not conform to the fitness function value, and at least one chromosome is selected from the cluster division as a particle for PSO restoration to obtain the second type of chromosome population; Based on the first type of chromosome population and the second type of chromosome population, a new population is obtained, and the above steps are repeated until the chromosome reaches the preset maximum number of iterations; Based on the chromosomes of the new population, the position of the receiving drone is iteratively adjusted until the receiving drone reaches the adjusted position.

2. The control method according to claim 1, characterized in that, The method for obtaining the initial coordinates of the receiving drone in the first coordinate system based on the angle between the line connecting the receiving drone for receiving signals, the transmitting drone, and the origin includes: Based on the first / second / third lines connecting the receiving drone to the first / second transmitting coordinates and the origin, the first / second included angle between the third line and the first / second line, and the third included angle between the first line and the second line are obtained; Based on the first / second included angle and the first / second connecting line, determine the first / second radius of the first / second circular curve where the first / second included angle and the first / second connecting line are located; Based on the first / second radius, the first / second transmission coordinates, and the origin, determine the coordinates of the first / second center of the first / second circular curve; Based on the first / second radius and the coordinates of the first / second circle center, determine the first / second curve equation of the first / second circular curve; Based on the first / second curve equations and the screening conditions, the initial coordinates of the receiving UAV are determined.

3. The control method according to claim 2, characterized in that, The method for determining the initial coordinates based on the first / second curve equation and the screening conditions includes: Based on the first curve equation and the second curve equation, obtain all intersection points between the first circular curve and the second circular curve; Based on all the intersection points and the filtering conditions, the initial coordinates of the receiving UAV are determined.

4. The control method according to claim 3, characterized in that, The method for obtaining a drone formation optimization model based on the initial coordinates of multiple receiving drones and multiple preset conditions includes: Based on the initial coordinates of the receiving drone and the first preset condition, it is determined that the adjusted position of the receiving drone is located in the third circular curve, and the formation model of the receiving drone for position adjustment is obtained. Based on the third circular curve and the second preset condition, the preset radius of the third circular curve is determined; Based on the formation model and the preset radius, the UAV formation optimization model is obtained.

5. The control method according to claim 4, characterized in that, The methods for cross-substitution and mutation substitution include: Two chromosomes are randomly selected and some genes of the two chromosomes are exchanged to obtain a class of replacement chromosomes; Repeat the above steps multiple times to obtain a first-generation population containing the aforementioned type of replacement chromosome; Based on the first offspring population, a gene locus on a chromosome is randomly selected, and a random gene value in the gene locus is replaced with another gene value in the gene locus to obtain a second type of replacement chromosome. Repeat the above steps multiple times to obtain a progeny population containing the two types of replacement chromosomes.

6. The control method according to claim 1, characterized in that, The method for K-means clustering of the second type of chromosomes that do not meet the fitness function value includes: Based on the second type of chromosomes that did not meet the fitness function value requirements, feature extraction was performed on the second type of chromosomes, a feature matrix was constructed, and normalization was performed to obtain chromosome feature samples; Based on chromosome feature samples, multiple chromosome feature samples are selected as initial cluster centers; Determine the distance between each of the second type of chromosomes and the initial cluster center; The second type of chromosome is assigned to the nearest initial cluster center; Determine the mean of each initial cluster center, and replace the initial cluster centers with the new mean to obtain new cluster centers; In response to the change of the cluster center, the chromosomes form a new cluster division, and at least one chromosome is selected from the cluster division as a particle for PSO restoration to obtain the second type of chromosome population.

7. The control method according to claim 1, characterized in that, The method for selecting at least one chromosome as a particle from the clustering partition for PSO reconstruction to obtain the second type of chromosome population includes: Multiple particles selected from the clustering partition will form an initial particle swarm; The initial position and initial velocity of each particle in the initial particle swarm are randomly set; Determine the fitness value of the particle; Based on the updated historical best position and global best position of the particle, adjust the particle's velocity and position; The initial particle swarm is updated iteratively to obtain the second optimal chromosome; Determine whether the number of update iterations has reached the preset maximum number of iterations; If the number of update iterations does not reach the preset maximum number of iterations, then repeat the above steps; When the number of update iterations reaches the preset maximum number of iterations, the second optimal chromosome becomes the optimized chromosome and forms the second type of chromosome population.