Energy consumption and time joint optimization method for three-dimensional ground area perception of multiple unmanned aerial vehicles

Abstract

The invention discloses an energy consumption and time joint optimization method for multi-unmanned aerial vehicle three-dimensional ground area perception, and the method comprises the steps: carrying out the discretization of a target area based on the real topographic data of the target area, and generating a ground grid and a center coordinate of the ground grid; calculating an air waypoint based on the ground grid and the center coordinate of the ground grid; analyzing the air waypoint information and the flight state of the unmanned aerial vehicle, and determining an optimization target equation; formulating an optimization problem of minimizing energy consumption and time required by multiple unmanned aerial vehicles to access all air waypoints according to the optimization target equation; and based on an improved multi-target doliolaria algorithm and a multi-task learning mode, solving an optimization problem by using a swarm intelligence optimization algorithm, and outputting a multi-unmanned aerial vehicle area coverage scheme. According to the method, the timeliness and the stability of the visual coverage task can be improved on the premise of ensuring that the task is completed.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle flight optimization, and more specifically to a method for jointly optimizing energy consumption and time for three-dimensional ground area perception of multiple unmanned aerial vehicles. Background Art

[0002] In the past few years, drones, especially multi-rotor drones, have been widely used in various fields due to their advantages such as strong maneuverability and simple deployment. Among the relevant application scenarios, drone-based ground area perception, such as disaster relief and environmental monitoring tasks, has attracted increasing attention. These tasks require drones equipped with cameras to collect images of the entire area, so visual coverage becomes a key issue. However, due to limited onboard storage and the complex and large terrain of most target areas, visual coverage tasks are usually difficult to complete, especially using a single drone; in addition, time, as an important indicator to measure the efficiency of task execution, needs to be always considered as an optimization goal. However, existing studies usually only discuss minimizing drone flight time or flight distance, ignore energy consumption optimization, and mostly consider scenarios where a single drone covers a single target area, which limits the generality of related optimization methods. Therefore, there is an urgent need to develop a general and efficient optimization framework around drone energy consumption optimization, task completion time optimization, multi-drone cooperation, and multi-target areas.

[0003] As a mainstream optimization method, swarm intelligence optimization method belongs to bionic optimization algorithm, which is based on the behavior of groups of organisms in nature (such as bird flocks, fish schools, ant colonies), and solves complex optimization problems through collaboration and information sharing between individuals. Nevertheless, when faced with the complex optimization problems of multiple drones, multiple regions, and multiple objectives mentioned above, this method still shows insufficient global search ability and the ability to handle mixed or even high-dimensional variables.

[0004] Therefore, how to provide a swarm intelligence optimization method that aims to minimize the energy consumption and time of multiple drones visually covering multiple ground areas is an urgent problem that technicians in this field need to solve. Summary of the invention

[0005] In view of this, the present invention provides a method for jointly optimizing energy consumption and time for three-dimensional ground area perception of multiple UAVs, which can improve the timeliness and stability of visual coverage tasks while ensuring the completion of the tasks.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception includes:

[0008] S1: discretize the target area based on the real terrain data of the target area, generate the ground grid and the center coordinates of the ground grid;

[0009] S2: Calculate the air waypoint based on the ground grid and the center coordinates of the ground grid;

[0010] S3: Analyze the air waypoint information and the flight status of the UAV to determine the optimization target equation;

[0011] S4: Formulate the optimization problem of minimizing the energy consumption and time required for multiple UAVs to visit all air waypoints based on the optimization objective equation;

[0012] S5: Solve the optimization problem based on the swarm intelligence optimization framework of the improved multi-objective salp algorithm and multi-task learning model, and output the multi-UAV area coverage plan.

[0013] Preferably, the S1 specifically includes:

[0014] S101: Calculate the rectangular field of view of the drone based on the target area represented by the digital elevation model;

[0015] S102: On the basis of satisfying the rectangular field of view of the drone and the overlap rate of the aerial image, the horizontal projection of the target area is discretized through reciprocating scanning and rasterization methods to generate ground grids of equal size, where the center coordinates of the kth ground grid of the i-th target area are marked as k∈[1,N i ],N i represents the number of ground grids in the i-th target area, satisfying in and are the number of scan lines and the number of segments of the i-th target area respectively.

[0016] Preferably, S3 specifically includes:

[0017] S301: Analyze the air waypoint information and the UAV flight status to determine the decision variables Decision variables Including the target area access order Number of drone air waypoints assigned Single-area air waypoint visit sequence Single area drone flight path And the speed distribution of drones in a single area

[0018] S302: Based on decision variables The minimum task completion time f1 is taken as the first optimization goal, expressed as:

[0019]

[0020] in, is the flight time of the g-th drone, g=1,2,...M, M represents the number of drones, satisfying

[0021]

[0022] Among them, M d is the number of discrete segments between two adjacent air waypoints, Indicates that a single drone visits a j bth region j The time required for segment p on the route is Is a single drone in a j bth region j The time spent on the fth turn on the route, Represents the number of air waypoints that the g-th drone needs to visit;

[0023] Based on decision variables The minimum total energy consumption f2 of the UAV is taken as the second optimization goal, which is expressed as:

[0024]

[0025] in, is the flight energy consumption of the g-th UAV, satisfying

[0026]

[0027] in, and Respectively represent the drone visiting the a j bth region j The average speed and average acceleration of the pth segment on the knot line, and The drones are in the a j bth region j The average speed and average acceleration of the lth segment on the fth turning trajectory on the knot line, τ is the number of line segments corresponding to the turning trajectory of the UAV, P mu Power consumption of the drone.

[0028] Preferably, the optimization problem is defined as:

[0029]

[0030] Among them, F is defined as the overall objective function, C1-C5 are constraints, and v max 、a max , H max and E maxrespectively represent the maximum flight speed, maximum flight acceleration rate, maximum flight altitude and maximum energy consumption of the UAV, e z = [0, 0, 1] is a unit vector on the Z-axis, v (·) and p (·) respectively represent the three-dimensional velocity and three-dimensional path point of the UAV, v (·) is the corresponding UAV speed, and are respectively the set of the number of target areas and the set of the number of UAVs.

[0031] Preferably, the S5 specifically includes:

[0032] Constructing an unconstrained form of the multi-task learning mode optimization problem by integrating the optimal population ratio calculation method;

[0033] Constructing a multi-objective salp swarm algorithm for multi-task learning mode by integrating the optimal population ratio calculation method, initializing the solution of the multi-objective salp swarm algorithm by using the random number generation method based on the Weierstrass equation, and updating the solution of the multi-objective salp swarm algorithm by integrating Levy flight and dynamic learning mechanism to improve the original multi-objective salp swarm algorithm for multi-task;

[0034] Based on the improved multi-objective salp swarm algorithm for multi-task, forming a swarm intelligence optimization framework and simultaneously solving the optimization problems in the constrained and unconstrained forms, and outputting a multi-UAV area coverage scheme.

[0035] Preferably, the random number generation method based on the Weierstrass equation, the specific calculation formula is:

[0036]

[0037] where, represents the initial continuous solution, and respectively represent the lower limit and upper limit of the initial continuous solution, F w. is generated by the Weierstrass function and satisfies 0 < a1 < 1, and b1 is an odd number.

[0038] Preferably, for the leader in the d-th dimension in Levy flight, it is defined as Its update method satisfies:

[0039]

[0040] where, X best (d) (r) represents the optimal solution of the d-th dimension group or salp chain in the r-th iteration process, ub d and lb dThey represent the upper and lower limits of the d-dimensional population, respectively. c1 is an important parameter for balancing exploration and utilization. is a random step size that follows the Levy distribution, and the calculation method satisfies Γ(·) is the gamma equation, satisfying

[0041] Preferably, in the dynamic learning mechanism, for the followers in the dth dimension in the rth iteration, it is defined as The update method of its qth item satisfies:

[0042]

[0043] in, is the weight operator corresponding to the rth iteration, satisfying The weight operator, Set to 2.595, represents the fitness function, is a dominating operator if means a is better than b, otherwise, b is the solution.

[0044] Preferably, the optimal population ratio calculation method is:

[0045]

[0046] Among them, R(p a ) is the population p a The corresponding optimal ratio is and Represents population P α The number of selected individuals and the total number of individuals.

[0047] Preferably, the coordinate calculation formula of the kth air waypoint in the i-th target area is:

[0048]

[0049] in, represents the coordinates of the kth aerial waypoint of the ith target area, represents the unit normal vector operator, k∈[1,N i ],h u is the visual coverage height of the UAV, and the number of waypoints is consistent with the number of grids.

[0050] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method for joint optimization of energy consumption and time for multi-UAV three-dimensional ground area perception, which has the following advantages:

[0051] 1) Full exploration of energy consumption and time optimization: Different from the current research on multi-UAV ground area visual coverage that ignores the problem of joint optimization of energy consumption and time, the present invention formulates a multi-objective optimization problem to explore the correlation between the two, focusing on improving the search capability of the optimization algorithm, determining the task plan that minimizes both energy consumption and time under constraints, and improving the efficiency of the coverage task.

[0052] 2) Significant improvement in search capability: Innovatively introduce multi-task learning mode, design simple tasks (i.e., optimization problems with constraints) and complex tasks (i.e., optimization problems without constraints), combine swarm intelligence optimization algorithms, realize knowledge transfer between different tasks, and enhance population diversity to avoid falling into local optimality. In addition, in order to handle high-dimensional variables, a multi-mechanism solution update operator for large-scale search space is proposed. This operator utilizes information exchange between populations and the introduction of new information based on population variation, which also enhances the algorithm's search capability.

[0053] 3) Effective processing of complex variables: In order to cope with the challenges brought by the difficulty or even inability of traditional optimization algorithms to handle complex variables (mixed, high-dimensional), the present invention designs two new operators, namely, a variable feature-guided hybrid solution initialization operator and a multi-mechanism solution update operator for large-scale search space. Based on a full analysis of variable characteristics, a variety of processing mechanisms are introduced and customized to ensure the randomness and uniformity of the initialization solution distribution, and the algorithm can efficiently complete the solution update.

[0054] In summary, the comprehensive technical innovation of the present invention not only realizes the joint optimization of drone energy consumption and mission time in the multi-drone ground area visual coverage scenario for the first time, but also provides a novel multi-objective optimization framework, which can well serve the current research field and also presents development potential in other fields. The above advantages promote the present invention to become an important technical support for drone ground perception tasks, significantly improving the service level and scene adaptability of drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0056] Figure 1 A schematic diagram of the energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception provided by the present invention.

[0057] Figure 2A specific flow chart of the energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception provided by the present invention.

[0058] Figure 3 A specific flow chart of solving optimization problems using a swarm intelligence optimization framework based on an improved multi-objective salp algorithm and a multi-task learning model provided by the present invention.

[0059] Figure 4 A schematic diagram of a three-dimensional ground area perception scene for multiple UAVs provided by the present invention.

[0060] Figure 5 This is a schematic diagram of the rectangular field of view of the drone provided by the present invention.

[0061] Figure 6 A schematic diagram of generating aerial waypoints provided by the present invention. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] The embodiment of the present invention discloses a method for jointly optimizing energy consumption and time for three-dimensional ground area perception of multiple unmanned aerial vehicles, such as Figure 1 As shown, including:

[0064] S1: discretize the target area based on the real terrain data of the target area, generate the ground grid and the center coordinates of the ground grid;

[0065] S2: Calculate the air waypoint based on the ground grid and the center coordinates of the ground grid;

[0066] S3: Analyze the air waypoint information and the flight status of the UAV to determine the optimization target equation;

[0067] S4: Formulate the optimization problem of minimizing the energy consumption and time required for multiple UAVs to visit all air waypoints based on the optimization objective equation;

[0068] S5: Based on the improved multi-objective salp algorithm and multi-task learning model and using the swarm intelligence optimization algorithm to solve the optimization problem, a multi-UAV area coverage solution is output.

[0069] This embodiment provides a specific implementation process, such as Figure 2 As shown, including:

[0070] The first step is to divide the ground area.

[0071] The rectangular field of view of the drone is calculated based on the target area represented by the digital elevation model, such as Figure 4 and Figure 5 As shown, the calculation formula is as follows:

[0072]

[0073] Among them, h u is the visual coverage height of the UAV, β h and β v They represent the horizontal and vertical viewing angles of the camera respectively, and l and w are the length and width of the rectangular field of view of the drone respectively.

[0074] On this basis, the requirements of aerial image overlap are met, and the horizontal projection of the target area is discretized using back and forth scanning (BFS) and rasterization methods to generate a series of ground grids of equal size. The center coordinates of the kth ground grid in the ith area are marked as k∈[1,N i ],N i represents the number of grids in the ith region, satisfying in and are the number of scan lines and the number of segments in the i-th region respectively.

[0075] The second step is to generate air waypoints.

[0076] Calculate the air waypoint based on the ground grid and the center coordinates of the ground grid, such as Figure 6 The coordinates of the kth air waypoint in the ith target area are recorded as The calculation formula is as follows:

[0077]

[0078] in, represents the unit normal vector operator, k∈[1,N i ].

[0079] The third step is to establish the optimization target equation.

[0080] Analyze air waypoint information and drone flight status to determine decision variables (include Respectively represent the regional visit order, the number of UAV waypoints allocated, the single regional waypoint visit order, the single regional UAV flight path and speed distribution), based on the theoretical UAV energy consumption model, as shown in formula (3), the flight energy consumption and flight time of a single UAV are obtained.

[0081]

[0082] in, and Respectively represent T u The average speed and average acceleration of the drone during the time period, P mu is the power consumption of the drone, E u Represents the energy consumption of the drone.

[0083] On this basis, two optimization objectives are constructed in a closed form, namely, the total energy consumption of the UAV and the mission completion time, which are described in formulas (4) and (6), respectively.

[0084] The minimum task completion time is taken as the first optimization goal, denoted as f1, and expressed as

[0085]

[0086] in, is the flight time of the g-th drone, satisfying

[0087]

[0088] Among them, M d is the number of discrete segments between two adjacent air waypoints, Indicates that a single drone visits a j bth region j The time required for segment p on the route is Is a single drone in a j bth region j The time spent on the fth turn on the route, Represents the number of air waypoints that the g-th drone needs to visit.

[0089] The minimum total energy consumption of the UAV is taken as the second optimization objective, denoted as f2, and expressed as

[0090]

[0091] in, is the flight energy consumption of the g-th UAV, satisfying

[0092]

[0093] in, and Respectively represent the drone visiting the a j bth region j The average speed and average acceleration of the pth segment on the knot line, and The drones are in the a jbth region j The average speed and average acceleration of the lth segment on the fth turning trajectory on the knot route, τ is defined as the number of line segments corresponding to the turning trajectory of the UAV, P mu Power consumption of the drone.

[0094] The fourth step is to optimize the problem formulation.

[0095] According to the optimization objective equation, the constraints are analyzed and a closed multi-objective optimization problem is formulated to minimize the energy consumption and time required for multiple UAVs to visit all air waypoints, which is expressed as follows:

[0096]

[0097] Among them, F is defined as the overall objective function, C1-C5 are constraints, and v max 、a max , H max and E max They represent the maximum flight speed, maximum flight acceleration rate, maximum flight altitude and maximum energy consumption of the UAV respectively. z =[0,0,1] is a unit vector on the Z axis, v (·) and p (·) Represent the 3D speed and 3D path point of the UAV, v (·) is the corresponding UAV speed, and They are the ground area quantity set and the drone quantity set respectively.

[0098] The fifth step is to use MTMSSA to solve the optimization problem, such as Figure 3 shown.

[0099] In the face of challenges exposed by optimization problems, including NP-hardness, mixed variables, and high-dimensional variables, the present invention focuses on solving them by introducing a multi-task learning model and two improved operators.

[0100] Specifically, first of all, around the closed UAV visual area coverage optimization problem (constrained multi-objective optimization task), the present invention designs a corresponding multi-objective salp swarm algorithm (MSSA), utilizes the collaborative search ability between individuals in the population, dynamically adjusts the search direction, and explores the optimal solution that satisfies the conditions of minimizing the flight energy consumption of the UAV and the task completion time. Secondly, the present invention introduces a multi-task learning mode into MSSA (called MTMSSA), carries out knowledge sharing between tasks, utilizes the useful knowledge provided by the unconstrained multi-objective optimization task, reduces the redundant calculation of the single task search, enhances the population diversity, and accelerates the global search process while avoiding falling into the local optimal solution. Multi-task learning is reflected in the need for MTMSSA to process multiple tasks simultaneously. In the present invention, two tasks are involved, one complex task is a constrained optimization problem formulated, and the other simple task is an optimization problem after the unconstrained processing of the former. In addition, the task processing process of MTMSSA mainly involves two stages, namely, the early stage and the late stage. In the early stage, MTMSSA selects offspring as better individuals to assist in solution updating. In the late stage, due to the complex relationship between simple tasks and complex tasks, it is difficult to distinguish whether the parent and offspring are better individuals. Therefore, the present invention proposes an ingenious optimal population ratio calculation method for judgment, as shown in formula (9):

[0101]

[0102] Among them, Opl pα and Represents population p α The number of excellent (selected) individuals and the total number of individuals, R(p a ) is the population p a Therefore, the one with a larger optimal ratio between the parent and the offspring, the more useful its knowledge is.

[0103] Then, in the process of initializing and updating the solutions of the two tasks, MTMSSA adopted improved operators. In the initialization stage, for continuous variables, a random number generation method based on the Weierstrass equation was adopted, as shown in formula (10), to ensure the uniformity and randomness of the solution distribution; in addition, for discrete variables, the sequence characteristics were analyzed, and a sequence random initialization scheme was customized. In the update stage, for continuous variables, the Levy flight mechanism was used to process the head of the ascidian chain (i.e., the leader), as shown in formula (11), to enhance the population diversity and avoid falling into local optima; in order to effectively increase the information update between other parts of the ascidian chain (i.e., the followers), a dynamic learning mechanism was proposed, as shown in formula (12), to adaptively adjust the influence proportion of adjacent followers on the solution. For discrete variables, a sequence update method based on crossover and mutation was constructed to improve the search efficiency of the solution.

[0104] Regarding the random number generation method based on the Weierstrass equation, it satisfies

[0105]

[0106] where, represents the initial continuous solution, and respectively represent the lower and upper bounds of the initial continuous solution, F w. is generated by the Weierstrass function and satisfies 0 < a1 < 1, and b1 is an odd number.

[0107] For the leader in the d-th dimension, it is defined as and its update method satisfies

[0108]

[0109] where, X best (d) (r) represents the optimal solution of the d-th dimension population (or ascidian chain) in the r-th iteration process. In addition, ub d and lb d respectively represent the upper and lower bounds corresponding to the d-th dimension population. c1 is an important parameter for balancing exploration and exploitation, is a random step size that follows the Levy distribution, and its calculation method satisfies Γ(·) is the gamma equation and satisfies

[0110] For the followers in the g-th dimension, it is defined as and its update method satisfies

[0111]

[0112] in, is the weight operator corresponding to the rth iteration, satisfying The weight operator of . represents the fitness function, is the dominating operator. If means a is better than b, otherwise, b is a better solution.

[0113] The present invention integrates Levy flight, dynamic learning and genetic mechanism, focuses on enhancing population diversity to improve search performance, alleviates the computational burden brought by high-dimensional data, and realizes efficient solution of complex problems by algorithms.

[0114] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0115] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A joint optimization method of energy consumption and time for multi-UAV three-dimensional ground area perception, characterized in that: include: S1: discretize the target area based on the real terrain data of the target area, generate the ground grid and the center coordinates of the ground grid; S2: Calculate the air waypoint based on the ground grid and the center coordinates of the ground grid; S3: Analyze the air waypoint information and the UAV flight status to determine the optimization target equation; S4: Formulate the optimization problem of minimizing the energy consumption and time required for multiple UAVs to visit all air waypoints based on the optimization objective equation; S5: Solve the optimization problem based on the swarm intelligence optimization framework of the improved multi-objective salp algorithm and multi-task learning model, and output the multi-UAV area coverage plan.

2. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 1 is characterized in that: The S1 specifically includes: S101: Calculate the rectangular field of view of the drone based on the target area represented by the digital elevation model; S102: On the basis of satisfying the rectangular field of view of the drone and the overlap rate of the aerial image, the horizontal projection of the target area is discretized through reciprocating scanning and rasterization methods to generate ground grids of equal size, where the center coordinates of the kth ground grid of the i-th target area are marked as k∈[1,N i ],N i represents the number of ground grids in the i-th target area, satisfying in and are the number of scan lines and the number of segments of the i-th target area respectively.

3. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 1 is characterized in that: The S3 specifically includes: S301: Analyze the air waypoint information and the UAV flight status to determine the decision variables Decision variables Including the target area access order Number of drone air waypoints assigned Single-area air waypoint visit sequence Single area drone flight path And the speed distribution of drones in a single area S302: Based on decision variables The minimum task completion time f1 is taken as the first optimization goal, expressed as: in, is the flight time of the g-th drone, g=1,2,...M, M represents the number of drones, satisfying Among them, M d is the number of discrete segments between two adjacent air waypoints, Indicates that a single drone visits a j bth region j The time required for segment p on the route is Is a single drone in a j bth region j The time spent on the fth turn on the route, Represents the number of air waypoints that the g-th drone needs to visit; Based on decision variables The minimum total energy consumption f2 of the UAV is taken as the second optimization goal, which is expressed as: in, is the flight energy consumption of the g-th UAV, satisfying in, and Respectively represent the drone visiting the a j bth region j The average speed and average acceleration of the pth segment on the knot line, and The drones are in the a j bth region j The average speed and average acceleration of the lth segment on the fth turning trajectory on the knot line, τ is the number of line segments corresponding to the turning trajectory of the UAV, P mu Power consumption of the drone.

4. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 3 is characterized in that: The optimization problem is defined as: Among them, F is defined as the overall objective function, C1-C5 are constraints, and v max 、a max , H max and E max They represent the maximum flight speed, maximum flight acceleration rate, maximum flight altitude and maximum energy consumption of the UAV respectively. z =[0,0,1] is a unit vector on the Z axis, v (·) and p (·) Represent the 3D speed and 3D path point of the UAV, v (·) is the corresponding UAV speed, and They are the target area quantity set and the drone quantity set respectively.

5. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 4 is characterized in that: The S5 specifically includes: The multi-task learning model that integrates the optimal population ratio calculation method constructs an unconstrained form of the optimization problem; A multi-task learning model integrating the optimal population ratio calculation method is used to construct a multi-task oriented multi-objective salp algorithm, and a random number generation method based on the Weierstrass equation is used to initialize the solution of the multi-objective salp algorithm. The solution of the multi-objective salp algorithm is updated by integrating the Levy flight and dynamic learning mechanism, thus improving the original multi-task oriented multi-objective salp algorithm. Based on the improved multi-task oriented multi-objective salp algorithm, a swarm intelligence optimization framework is formed to simultaneously solve the optimization problems in constrained and unconstrained forms, and output a multi-UAV area coverage plan.

6. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 5 is characterized in that: The random number generation method based on the Weierstrass equation, the specific calculation formula is: Among them, represents the initial continuous solution, and represent the lower and upper limits of the initial continuous solution respectively, F w· is generated by the Weierstrass function and satisfies 0 < a1 < 1, and b1 is an odd number.

7. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 5 is characterized in that: The leader in the dth dimension in Levi's flight is defined as Its update method meets the following requirements: Among them, X best (d) (r) represents the optimal solution of the d-th dimension group or ascidian chain in the r-th iteration, ub d and lb d They represent the upper and lower limits of the d-dimensional population, respectively. c1 is an important parameter for balancing exploration and utilization. is a random step size that follows the Levy distribution, and the calculation method satisfies Γ(·) is the gamma equation, satisfying 8. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 7 is characterized in that: In the dynamic learning mechanism, for the followers on the dth dimension in the rth iteration, it is defined as The update method of its qth item satisfies: in, is the weight operator corresponding to the rth iteration, satisfying Set to 2.595, represents the fitness function, is a dominating operator if means a is better than b, otherwise, b is the solution.

9. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 5 is characterized in that: Optimal population ratio calculation method: Among them, R(p a ) is the population p a The corresponding optimal ratio is and Represents population p α The number of selected individuals and the total number of individuals.

10. The energy consumption and time joint optimization method for multi-UAV three-dimensional ground area perception according to claim 2 is characterized in that: The coordinate calculation formula of the kth air waypoint in the i-th target area is: in, represents the coordinates of the kth aerial waypoint of the ith target area, represents the unit normal vector operator, k∈[1,N i ],h u is the visual coverage height of the UAV, and the number of waypoints is consistent with the number of grids.

Images