A method for intelligent airport site selection and task allocation for UAVs in urban management

Abstract

The present invention discloses a method for intelligent airport site selection and task allocation for UAVs oriented to urban management. The method comprehensively considers factors such as flight distance constraints, signal strength, energy consumption, and demand coverage, and uses a two-stage random 0-1 programming model to jointly optimize the intelligent airport site selection before random demand realization and the task allocation after random demand realization. At the same time, the present invention also discloses an optimal intelligent airport site selection and task allocation algorithm combining genetic algorithm and dynamic programming for the above two-stage random 0-1 programming model.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicles, and in particular is a method for intelligent airport site selection and task allocation for unmanned aerial vehicles oriented to urban management. Background Art

[0002] In urban management, different tasks often appear in different locations, such as renovation of dilapidated houses, sewage treatment, forest fires, sanitation and epidemic prevention, etc. The traditional manual management model not only takes up a lot of human resources, is time-consuming and labor-intensive, but also often leads to blind spots and delays in supervision. The use of drones equipped with various intelligent devices and smart airports with unmanned operation functions can achieve the goals of flexible deployment, unmanned operation, 360-degree monitoring without blind spots, remote control, and timely response in urban management.

[0003] At present, there have been some studies on the site selection of drone smart airports, but these studies usually do not comprehensively consider the smart airport setting cost, mission coverage, drone flight distance, signal strength, etc., and do not uniformly optimize the smart airport site selection and task allocation before and after the realization of random demand. Summary of the invention

[0004] The purpose of the present invention is to provide a method for site selection and task allocation of UAV intelligent airports for urban management, which optimizes the site selection of intelligent airports and the allocation of task requirements among different intelligent airports to achieve the best balance among the goals of intelligent airport setting cost, task coverage, total flight distance of UAVs, total signal strength, etc.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] Step (1): Gridding of the task area

[0007] The two-dimensional task area is divided into grids so that the locations of task requirements and the addresses of smart airports are both in grid units. is the set of all grids that will generate task requirements, It is the collection of grids of all candidate smart airport addresses.

[0008] Step (2), a method for allocating UAV tasks for urban management, when allocating tasks, the random task requirements have been realized, and the address of the smart airport has been given. The UAV smart airport is a ground device with modules such as UAV storage, smart standby, automatic replacement of UAV batteries, wireless communication, automatic battery maintenance, UPS power failure protection, fault self-test, and take-off condition detection. It includes the following processes:

[0009] (2.1) The address of the given smart airport:

[0010] (1)

[0011] In formula (1), Represents the set of addresses of all given smart airports;

[0012] (2.2) Calculate the distance between the smart airport and the mission location:

[0013] (2)

[0014] In formula (2) Smart Airport i Mission location j The distance between , , It is a smart airport i The two-dimensional coordinates of It is the mission location j The two-dimensional coordinates of

[0015] (2.3) Calculate the mission coverage, the total flight distance of the drone, and the total signal strength, and set the objective function to the maximum of their weighted functions:

[0016] (3)

[0017] In formula (3), , ,in, Indicates whether to set the task location j Assign tasks to smart airports i The decision variables, It indicates the location of the mission j Whether there is a task, the obedience parameter is A Bernoulli distributed random variable. Indicates the mission location j The importance of the task, Smart Airport i Mission location j The average reference signal received power between , and is the weight that represents the relative importance of mission coverage, total flight distance of the UAV, and total signal strength;

[0018] (2.4) Smart Airport i The total task intensity is constrained by:

[0019] (4)

[0020] In formula (4), Indicates the flight speed of the drone. Smart Airport i The total working hours limit;

[0021] (2.5) Smart Airport i Mission location j The flight distance between them is constrained:

[0022] (5)

[0023] In formula (5), Smart Airport i The maximum distance a drone can reach in a normal flight;

[0024] (2.6) Mission location j The task allocation is constrained by:

[0025] (6)

[0026] In formula (6), when When indicating the mission location j There is a task, when When indicating the mission location j No mission. When assigning drone missions, has been implemented as a definite value;

[0027] (2.7) Constrain the signal strength of the drone:

[0028] (7)

[0029] In formula (7), Indicates the minimum signal strength required for the drone to perform its mission. Smart Airport i Mission location j Minimum signal strength on the flight path between

[0030] (2.8) Constrain the task allocation decision variables:

[0031] (8)

[0032] In formula (8), when When the task location is j Assign tasks to smart airports i ,when When the task location is not j Assign tasks to smart airports i ;

[0033] (2.9), summarizing equations (3) to (8) into the following form:

[0034] (9);

[0035] (2.10) The above optimization problem belongs to the multidimensional 0-1 knapsack problem and is NP-complete. The present invention first uses equation (5) to narrow the range of feasible solutions, and then uses the dynamic programming algorithm to solve it. The framework of the algorithm is as follows: 1) For any Smart Airport i Mission location j , ; 2) Construct the state transfer equation; 3) Use for loop or memoized recursion to realize state transfer; 4) Determine the boundary problem, that is, the subproblem that can easily get the optimal solution. By solving (9), we can get the optimal task allocation when the random task demand has been realized and the address of the smart airport has been given. The algorithm has a pseudo-polynomial time complexity.

[0036] Step (3), a method for selecting a site for a UAV smart airport for urban management, characterized in that the optimal smart airport address is determined before random demand realization and task allocation, including the following process:

[0037] (3.1) Calculate the optimal task allocation benefit function:

[0038] (10)

[0039] In formula (10), Indicates that the demand is In the case of, the optimal task allocation benefit function is obtained according to the task allocation method described in step (2);

[0040] (3.2) Calculate the setting cost of the smart airport, and set the objective function with the weighted maximum of the setting cost of the smart airport and the optimal task allocation benefit function:

[0041] (11)

[0042] In formula (11), Represented in the grid i The cost required to set up a smart airport (including equipment cost, installation cost, maintenance cost, etc.), Indicates whether it is in the grid i Set the decision variables of the smart airport, represents the expectation operation, and is the weight representing the relative importance of the setup cost of the smart airport and the benefit of optimal task allocation;

[0043] (3.3) Constrain the decision variables of smart airport site selection:

[0044] (12)

[0045] In formula (12), when When the grid i Setting up a smart airport When it is not in the grid i Setting up smart airports;

[0046] (3.4), summarizing equation (11) and equation (12) into the following form:

[0047] (13);

[0048] (3.5) The above optimization problem is combined with the task allocation represented by formula (9), which belongs to a two-stage stochastic programming problem. The intelligent airport site selection decision is made before the random demand is realized, and the task allocation decision is made after the random demand is realized. The present invention uses a genetic algorithm to solve, that is, randomly generate formula (13) N feasible solutions as the initial population; use the exponential sorting selection method to select individuals in the current population for replication; randomly pair the individuals generated by the selection-copy operation; perform a mutation operation on each individual generated by the crossover operation with a certain mutation probability; and iterate in this way until the termination condition is met.

[0049] The present invention proposes a method for intelligent airport site selection and task allocation for UAVs for urban management. The method comprehensively considers factors such as flight distance constraints, signal strength, energy consumption, and demand coverage, and uses a two-stage random 0-1 programming model to jointly optimize the intelligent airport site selection before the random demand is realized and the task allocation after the random demand is realized.

[0050] The present invention proposes an optimal intelligent airport site selection and task allocation algorithm that combines genetic algorithm and dynamic programming for the above two-stage random 0-1 programming model. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 Flow chart of intelligent airport site selection and task allocation for unmanned aerial vehicles of the present invention.

[0052] Figure 2 Grid diagram of the mission area.

[0053] Figure 3 Schematic diagram of task allocation after random requirements are realized.

[0054] Figure 4Schematic diagram of smart airport site selection before random demand is realized. DETAILED DESCRIPTION

[0055] The following is a description of the technical solution of the specific implementation of the present invention in conjunction with an example. This example is used to illustrate the implementation of the technical solution and does not represent the scale of solving actual problems.

[0056] In order to specifically describe the implementation of the technical solution of the present invention, the flight speed of the drone selected in this example is 7.2km / h, the maximum cruising range is 4km, and the drone is equipped with high-definition gimbal camera module, infrared thermal imaging module, intelligent shouting module, gas detector module and other equipment, as well as communication / navigation system, mission payload system, airborne computer system, etc., to meet the functional requirements of different business scenarios in urban management.

[0057] Smart Airport is a ground-mounted intelligent device that enables unmanned drone operations. It provides functions such as drone storage, intelligent standby, automatic replacement of drone batteries, automatic battery maintenance, UPS power failure protection, fault self-checking, and take-off condition detection. It can be flexibly deployed, unmanned 24 hours a day, extremely fast response, and cross-continuous execution of tasks without human control. The size is 1860x1860x1920mm, the drone battery charging time from 0 to 100% is 15min, and it can support drone landing at night.

[0058] The implementation process of the technical solution of the present invention is as follows Figure 1 shown.

[0059] Step 1: Gridding the Task Area

[0060] The two-dimensional task area is divided into grids so that the locations of task requirements and the addresses of smart airports are both in grid units. is the set of all grids that will generate task requirements, is the collection of grids of all candidate smart airport addresses. In this example, for Figure 2 Five grids with triangles in the middle, for Figure 2 Four grids with circles in them.

[0061] Step 2: Task allocation after random requirements are realized

[0062] 1. The address of the given smart airport:

[0063] (14)

[0064] In formula (14), represents the set of addresses of all given smart airports. In this example, for Figure 3 Three grids with five-pointed stars in the middle;

[0065] 2. Calculate the distance between the smart airport and the mission location:

[0066] (15)

[0067] In formula (15) Smart Airport i Mission location j The distance between , , It is a smart airport i The two-dimensional coordinates of It is the mission location j The two-dimensional coordinates of

[0068] 3. Calculate the mission coverage, the total flight distance of the drone, and the total signal strength, and set the objective function to the maximum of their weighted functions:

[0069] (16)

[0070] In formula (16), , ,in, Indicates whether to set the task location j Assign tasks to smart airports i The decision variables, It indicates the location of the mission j Whether there is a task, the obedience parameter is A Bernoulli distributed random variable. Indicates the mission location j The importance of the task, Smart Airport i Mission location j The average reference signal received power between , and is the weight representing the relative importance of mission coverage, total flight distance of the drone, and total signal strength. , , , , , , , , , , , , , , , , , , , , , , ;

[0071] 4. Smart Airport i The total task intensity is constrained by:

[0072] (17)

[0073] In formula (17), Indicates the flight speed of the drone. Smart Airport i The total working time limit. In this example, , ;

[0074] 5. Smart Airport i Mission location j The flight distance between them is constrained:

[0075] (18)

[0076] In formula (18), Smart Airport i The maximum distance that a drone can reach in a normal flight. ;

[0077] 6. Mission location j The task allocation is constrained by:

[0078] (19)

[0079] In formula (19), when When indicating the mission location j There is a task, when When indicating the mission location j No mission. When assigning drone missions, has been realized as a definite value. In this case, all task locations have tasks, i.e. ;

[0080] 7. Limit the signal strength of the drone:

[0081] (20)

[0082] In formula (20), Indicates the minimum signal strength required for the drone to perform its mission. Smart Airport i Mission location j The lowest signal strength on the flight path between , , , , , , , , , , , , , , , ;

[0083] 8. Constrain the task allocation decision variables:

[0084] (twenty one),

[0085] In formula (21), when When the task location is j Assign tasks to smart airports i ,when When the task location is not j Assign tasks to smart airports i ;

[0086] 9. Summarize equations (16) to (21) and write them in the following form:

[0087] (twenty two);

[0088] 10. The above optimization problem belongs to the multidimensional 0-1 knapsack problem and is NP-complete. The present invention first uses equation (18) to narrow the range of feasible solutions, and then uses the dynamic programming algorithm to solve it. The framework of the algorithm is as follows: 1) For any satisfying Smart Airport i Mission location j , ; 2) Construct the state transfer equation; 3) Use for loop or memoized recursion to realize state transfer; 4) Determine the boundary problem, that is, the subproblem that can easily get the optimal solution. By solving (22), we can get the optimal task allocation when the random task demand has been realized and the address of the smart airport has been given. The algorithm has a pseudo-polynomial time complexity. In this case, if Figure 3 As shown, it can be solved , , , , , , , , , , , , , , .

[0089] Step 3: Random demand implementation of previous smart airport site selection

[0090] 1. Calculate the optimal task allocation benefit function:

[0091] (twenty three),

[0092] In formula (23), Indicates that the demand is In the case of, the optimal task allocation benefit function is obtained according to the task allocation method described in step (2);

[0093] 2. Calculate the setup cost of the smart airport, and use the setup cost of the smart airport and the optimal task allocation benefit function

[0094] Weighted maximum setting objective function:

[0095] (twenty four),

[0096] In formula (24), Represented in the grid i The cost required to set up a smart airport (including equipment cost, installation cost, maintenance cost, etc.), Indicates whether it is in the grid i Set the decision variables of the smart airport, represents the expectation operation, and is a weight that represents the relative importance of the smart airport setup cost and the optimal task allocation benefit. In this example, Ten thousand yuan, Ten thousand yuan, Ten thousand yuan, Ten thousand yuan, , , , , , , ;

[0097] 3. Constrain the smart airport site selection decision variables:

[0098] (25),

[0099] In formula (25), when When the grid i Setting up a smart airport When it is not in the grid i Setting up smart airports;

[0100] 4. Summarize equation (24) and equation (25) and write them in the following form:

[0101] (26);

[0102] 5. The above optimization problem combined with the task allocation expressed by formula (22) belongs to a two-stage stochastic programming problem. The intelligent airport site selection decision is made before the random demand is realized, while the task allocation decision is made after the random demand is realized. Use the genetic algorithm to solve the maximum value of the objective function, that is, randomly generate 100 feasible solutions of formula (26) as the initial population; use the exponential sorting selection method to select individuals in the current population for replication; randomly pair the individuals generated by the selection-replication operation; perform a mutation operation on each individual generated by the crossover operation with a mutation probability of 0.1; and iterate in this way until the 20th generation is iterated. In this example, Figure 4 As shown, it can be solved that , , , .

Claims

1. A method for allocating unmanned aerial vehicle tasks for urban management, characterized in that: Includes task area meshing: The two-dimensional task area is divided into grids so that the location of the task requirement and the address of the smart airport are both in grid units; is the collection of all grids that generate task requirements, It is the collection of grids of all candidate smart airport addresses; When performing task allocation, the random task requirements have been realized and the address of the smart airport has been given; the UAV smart airport is a ground device composed of a UAV storage module, an intelligent waiting module, an automatic replacement UAV battery module, a wireless communication module, an automatic battery maintenance module, a UPS power failure protection module, a fault self-checking module, and a take-off condition detection module; it includes the following processes: (2.1) The address of the given smart airport: (1)， In formula (1), Represents the set of addresses of all given smart airports; (2.2) Calculate the distance between the smart airport and the mission location: (2)， In formula (2) Smart Airport i Mission location j The distance between , , It is a smart airport i The two-dimensional coordinates of It is the mission location j The two-dimensional coordinates of (2.3) Calculate the mission coverage, the total flight distance of the drone, and the total signal strength, and set the objective function to the maximum of their weighted functions: (3)， In formula (3), , ,in, Indicates whether to set the task location j Assign tasks to smart airports i The decision variables, It indicates the location of the mission j Whether there is a task, the obedience parameter is A Bernoulli distributed random variable; Indicates the mission location j The importance of the task, Smart Airport i Mission location j The average reference signal received power between , and is the weight that represents the relative importance of mission coverage, total flight distance of the UAV, and total signal strength; (2.4) Smart Airport i The total task intensity is constrained by: (4)， In formula (4), Indicates the flight speed of the drone. Smart Airport i The total working hours limit; (2.5) Smart Airport i Mission location j The flight distance between them is constrained: (5)， In formula (5), Smart Airport i The maximum distance a drone can reach in a normal flight; (2.6) Mission location j The task allocation is constrained by: (6)， In formula (6), when When indicating the mission location j There is a task, when When indicating the mission location j No mission; when assigning drone missions, has been implemented as a definite value; (2.7) Constrain the signal strength of the drone: (7)， In formula (7), Indicates the minimum signal strength required for the drone to perform its mission. Smart Airport i Mission location j Minimum signal strength on the flight path between (2.8) Constrain the task allocation decision variables: (8)， In formula (8), when When the task location is j Assign tasks to smart airports i ,when When the task location is not j Assign tasks to smart airports i ; (2.9), summarizing equations (3) to (8) into the following form: (9)； (2.10) The above optimization problem belongs to the multidimensional 0-1 knapsack problem, which is NP-complete. First, use equation (5) to narrow the range of feasible solutions, and then use the dynamic programming algorithm to solve it. The framework of the algorithm is as follows: 1) For any Smart Airport i Mission location j , ; 2) Construct the state transfer equation; 3) Use for loop or memoized recursion to realize state transfer; 4) Determine the boundary problem, that is, the subproblem that can easily obtain the optimal solution; By solving (9), the optimal task allocation can be obtained when the random task demand has been realized and the address of the smart airport has been given. The algorithm has a pseudo-polynomial time complexity.

2. A method for selecting a site for a UAV smart airport for urban management based on the task allocation method of claim 1, characterized in that: Determining the optimal smart airport location before random demand realization and task allocation includes the following processes: (3.1) Calculate the optimal task allocation benefit function: (10)， In formula (10), Indicates that the demand is , the optimal task allocation benefit function obtained according to the task allocation method; (3.2) Calculate the setting cost of the smart airport, which is the sum of the setting cost of the smart airport and the optimal task allocation benefit function. The objective function is set to the maximum weight: (11)， In formula (11), Represented in the grid i The cost required to set up a smart airport, including equipment cost, installation cost, and maintenance cost, Indicates whether it is in the grid i Set the decision variables of the smart airport, represents the expectation operation, and is the weight representing the relative importance of the setup cost of the smart airport and the benefit of optimal task allocation; (3.3) Constrain the decision variables of smart airport site selection: (12)， In formula (12), when When the grid i Setting up a smart airport When it is not in the grid i Setting up smart airports; (3.4), summarizing equation (11) and equation (12) into the following form: (13)； (3.5) The above optimization problem is combined with the task allocation represented by formula (9), which belongs to a two-stage stochastic planning problem; the intelligent airport site selection decision is made before the random demand is realized, and the task allocation decision is made after the random demand is realized; the genetic algorithm is used to solve it, that is, the random generation formula (13) N feasible solutions as the initial population; use the exponential sorting selection method to select individuals in the current population for replication; randomly pair the individuals generated by the selection-copy operation; perform a mutation operation on each individual generated by the crossover operation with a certain mutation probability; and iterate in this way until the termination condition is met.

Images