Parking apron resource joint scheduling method based on bilevel planning

Abstract

The invention discloses a parking apron resource joint scheduling method based on bilevel programming. The method comprises the steps that firstly, the parking apron resource scheduling cost is analyzed, and a cost quantification method comprising four factors of fuel oil, pollutants, carbon emission and time is put forward; then establishing a parking apron resource joint scheduling model based on double-layer programming, an upper layer target being minimization of parking space assignment cost, a lower layer target being minimization of ground support vehicle scheduling cost, taking an upper layer model result as a lower layer model constraint condition, and achieving overall optimization through iterative optimization; and finally, parking apron resource joint scheduling solving is carried out, an upper layer adopts a genetic algorithm, a lower layer adopts a brainstorm algorithm, and an improved hybrid genetic brainstorm algorithm is constructed to solve the bilayer planning model. Experimental results show that joint scheduling is superior to independent scheduling in economic cost and resource use, the cost of airlines and airports is effectively reduced, and decision basis is provided for parking apron scheduling of large airports in peak hours.

Description

Technical Field

[0001] The invention relates to the technical field of scene optimization, parking space assignment and ground support vehicle dispatching, and specifically designs a joint dispatching method of apron resources based on double-layer planning. Background Art

[0002] Apron resources include parking space resources and ground support vehicle resources. The parking space assignment plan is a prerequisite for ground support vehicle scheduling. The scheduling efficiency of vehicles will affect the length of time aircraft occupy the parking space. The two restrict each other and jointly affect the ground service quality and punctuality of flights. In the actual operation of the airport, in order to achieve the overall optimal scheduling, it is necessary to consider the joint scheduling of multiple airport surface resources.

[0003] In terms of research content, most of the current research on joint scheduling of apron resources is aimed at parking stands and taxiway resources. The few studies on joint scheduling of ground support vehicles and other resources only consider a single type of vehicle. In addition, the overall process of the research is to allocate parking stands to flights and then dispatch ground support vehicles, without considering the reverse impact of parking stands on the path arrangement of ground support vehicles. In terms of research methods, set partitioning models, mixed integer linear programming models, etc. are mostly used, and solutions are obtained through exact algorithms and heuristic algorithms. Among them, exact algorithms are inefficient when dealing with large-scale problems and are difficult to solve multi-objective optimization problems. Heuristic algorithms still have room for improvement in robustness and optimization accuracy. Therefore, it is necessary to conduct overall joint scheduling of parking stands and various ground support vehicles to further improve the efficiency of airport operations. Summary of the invention

[0004] The present invention discloses a method for joint scheduling of apron resources based on double-layer planning, aiming to perform overall joint scheduling of parking spaces and ground support vehicles, complete peak-period support tasks while meeting parking space and vehicle resource constraints, and improve apron operation efficiency.

[0005] To achieve the above purpose, the technical solution provided by the present invention is:

[0006] A method for joint scheduling of apron resources based on double-layer planning, characterized by comprising the following main steps:

[0007] Step 1: Divide the apron resource scheduling cost into the parking space assignment cost and the ground support vehicle scheduling cost, and propose a cost quantification method that includes four factors: fuel, pollutants, carbon emissions, and time;

[0008] Step 2: Establish a joint scheduling model for apron resources based on two-layer planning. The upper model is the parking space assignment model, and the lower model is the ground support vehicle scheduling model. The total cost of apron resource scheduling is minimized while meeting the apron resource capacity constraints.

[0009] Step 3: To solve the joint scheduling of apron resources, the upper layer uses an improved genetic algorithm and the lower layer uses a brainstorming algorithm to construct an improved hybrid genetic brainstorming algorithm (AHGBA) for solution.

[0010] In order to optimize the above technical solutions, the specific measures / limitations taken also include:

[0011] In step 1, the parking space assignment cost is divided into four parts: aircraft fuel cost, aircraft pollutant cost, aircraft carbon emission cost and aircraft time cost.

[0012] Aircraft fuel cost F incurred by flight i i aircraft The calculation formula is as follows. is the total taxiing time of flight i, Refers to the unit mass aviation fuel cost, r i is the fuel consumption rate of one engine at 7% thrust on flight i, n i Refers to the number of engines of flight i.

[0013]

[0014] The total pollutant cost P of flight i during the taxiing phase i aircraft The calculation formula is as follows. Where, P i CO is the carbon monoxide environmental pollution cost of flight i, P i HC is the hydrocarbon environmental pollution cost of flight i, is the environmental pollution cost of nitrogen oxides for flight i.

[0015]

[0016] Carbon emission cost of flight i The calculation formula is as follows. Where, c e is the CO2 emission cost per unit mass, is the CO2 emission index of aviation fuel.

[0017]

[0018] The total time cost T of flight i i aircraft The calculation formula is as follows. is the waiting time of flight i at the parking stand, is the unit time cost of the flight, is the delay time cost of flight i.

[0019]

[0020] The ground support vehicle dispatch cost is divided into four parts: vehicle fuel cost, vehicle pollutant cost, vehicle carbon emission cost and vehicle time cost.

[0021] The fuel consumption cost F of vehicle l l vehicle The calculation formula is as follows. Where, TD l is the total distance traveled by vehicle l, v vehicle is the vehicle speed, is the vehicle fuel cost per unit time, and u is the vehicle fuel consumption.

[0022]

[0023] The total pollution cost P of vehicle l l vehicle The calculation formula is as follows. are the carbon monoxide, hydrocarbon and nitrogen oxide emissions of vehicle l, c CO 、c HC , They are the emission costs per unit mass of carbon monoxide, hydrocarbons and nitrogen oxides respectively.

[0024]

[0025] Carbon emission cost of vehicle l The calculation formula is as follows. Where, c e is the CO2 emission cost per unit mass, It is the carbon dioxide emission index of vehicle fuel.

[0026]

[0027] The time cost of vehicle l is T l vehicle The calculation formula is as follows. is the unit time cost of ground support vehicles, RPT l The time when vehicle l returns to the parking lot, DPT l The time when vehicle l leaves the parking lot.

[0028]

[0029] In step 2, appropriate parking spaces are assigned to aircraft. The objective function is to minimize the parking space assignment cost, which includes aircraft fuel cost, pollutant cost, carbon emission cost and time cost. The upper-level planning model for parking space assignment is constructed. The formula is as follows. The decision variable X ikIndicates whether flight i is assigned to parking stand k.

[0030]

[0031] Arrange reasonable service routes for ground support vehicles to complete the services of four types of vehicles. The objective function is to minimize the vehicle scheduling cost, including vehicle fuel cost, pollutant cost, carbon emission cost and time cost, and build a lower-level planning model for ground support vehicle scheduling. The formula is as follows. The decision variables Indicates whether vehicle l in model q serves flight i.

[0032]

[0033] The parking space assignment plan obtained by the upper model is passed as input to the lower model, and the ground support vehicle scheduling plan and scheduling cost obtained by the lower model are returned to the upper model. The two are optimized through continuous iteration and finally reach the overall optimality.

[0034] In step 3, an improved genetic algorithm is designed for the upper-level parking lot assignment problem, using adaptive crossover and mutation probabilities, adding memory operations, and improving algorithm efficiency.

[0035] In genetic algorithms, the crossover operator affects the global search capability, and the mutation operator affects the local search capability. In the initial stage of iteration, in order to ensure that the population converges to the global in an orderly manner, it is necessary to set a larger crossover probability and a smaller mutation probability; in the middle and late stages of iteration, the population optimization turns to local search, and in order to maintain population diversity, it is necessary to set a smaller crossover probability and a larger mutation probability.

[0036] Aiming at the problem of lower-level ground support vehicle scheduling, a brainstorming algorithm is designed, and local search operations are used to improve the efficiency of the algorithm.

[0037] The local search operation draws on the idea of ​​"destruction" and "repair" in the neighborhood search algorithm. First, some flights are removed from the current plan to complete the destruction operation, and then the removed flights are reinserted into the destroyed plan to complete the repair operation.

[0038] Based on the idea of ​​evolutionary iteration, an improved hybrid genetic brainstorming algorithm (AHGBA) is constructed to solve the bi-level programming model. The basic steps are as follows.

[0039] Step 1: First, initialize the upper-level planning decision variables and generate a random initial solution, i.e., the parking space assignment scheme for each aircraft;

[0040] Step 2: The lower-level planning generates the optimal vehicle and route scheduling for each individual in the upper-level planning population, and feeds back the optimal scheduling plan to the upper-level planning;

[0041] Step 3: The upper-level planning uses the fitness function to calculate the optimal objective function value of each individual;

[0042] Step 4: Perform improved genetic operations on the population and use iteration to approximate the optimal solution of the model.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention establishes a joint scheduling model for apron resources from a global perspective: currently, most airports conduct scheduling research on a single type of resource, and the scheduling process is highly independent. The present invention takes into account the reverse impact of parking spaces on the routing of ground support vehicles, while taking into account a variety of major vehicle types, and jointly schedules these two resources, breaking down the barriers between resource scheduling, effectively reducing the costs of airlines and airports, and providing a decision-making basis for apron scheduling during peak hours at large airports. In addition, an improved hybrid intelligent optimization algorithm is designed to solve the two-level programming model: the present invention uses an improved genetic algorithm to solve the upper model, and a brainstorming algorithm to solve the lower model, and uses methods such as adding a memory bank, introducing local search, and setting adaptive parameters to improve the traditional algorithm, constructing an improved hybrid genetic brainstorming algorithm to improve the algorithm's solving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Flowchart of the joint scheduling method of apron resources based on double-layer planning.

[0046] Figure 2 The idea of ​​joint scheduling of apron resources based on double-layer planning.

[0047] Figure 3 Parking stand configuration diagram.

[0048] Figure 4 Iteration graphs of upper and lower layer costs.

[0049] Figure 5 Gantt chart of stand assignments.

[0050] Figure 6 Guarantee vehicle service Gantt chart. DETAILED DESCRIPTION

[0051] The above contents of the present invention are further described in detail below in the form of embodiments, but this should not be understood as the scope of the above subject matter of the present invention being limited to the following embodiments, and all technologies realized based on the above contents of the present invention belong to the scope of the present invention.

[0052] The present invention proposes a joint scheduling method for apron resources based on double-layer planning, the flow chart is as follows: Figure 1 As shown, the following steps are included:

[0053] (1) The apron resource scheduling cost is divided into parking space assignment cost and ground support vehicle scheduling cost, and a cost quantification method that includes four factors: fuel, pollutants, carbon emissions, and time is proposed.

[0054] The specific steps in step (1) include:

[0055] 1.1, the parking space assignment cost is divided into four parts: aircraft fuel cost, aircraft pollutant cost, aircraft carbon emission cost and aircraft time cost.

[0056] Based on the ICAO database, the fuel flow rate at 7% of rated power is used as the fuel flow rate for aircraft taxiing. Aircraft fuel cost F generated by flight i i aircraft The calculation formula is as follows. is the total taxiing time for flight i, Refers to the unit mass aviation fuel cost, r i is the fuel consumption rate of one engine at 7% thrust on flight i, n i Refers to the number of engines of flight i.

[0057]

[0058] The present invention mainly considers the environmental costs of three pollutants: carbon monoxide, hydrocarbons and nitrogen oxides. The total pollutant cost P of flight i during the taxiing phase is i aircraft The calculation formula is as follows. Where, P i CO , P i HC , are the environmental pollution costs of carbon monoxide, hydrocarbons and nitrogen oxides of flight i respectively.

[0059]

[0060] CO2 emissions generated during the taxiing phase of an aircraft are one of the main sources of carbon emissions at an airport. Therefore, the carbon emission cost of flight i is The calculation formula is as follows. Where, c e is the CO2 emission cost per unit mass, is the CO2 emission index of aviation fuel.

[0061]

[0062] The aircraft time cost includes the airline cost and passenger delay time cost excluding fuel. The total time cost of flight i is T i aircraft The calculation formula is as follows. is the waiting time of flight i at the parking stand, is the unit time cost of the flight, is the delay time cost of flight i.

[0063]

[0064] 1.2. The dispatching cost of ground support vehicles is divided into four parts: vehicle fuel cost, vehicle pollutant cost, vehicle carbon emission cost and vehicle time cost.

[0065] When calculating the vehicle fuel cost, the driver's attributes, road conditions, and traffic conditions are not considered. The standard fuel consumption value is calibrated mainly through vehicle attributes and natural environmental factors. l vehicle The calculation formula is as follows. Where, TD l is the total distance traveled by vehicle l, v vehicle is the vehicle speed, is the vehicle fuel cost per unit time, and u is the vehicle fuel consumption.

[0066]

[0067] The calculation method of vehicle gas pollutant emission index refers to the Technical Guidelines for Compilation of Road Motor Vehicle Air Pollutant Emission Inventory. l vehicle The calculation formula is as follows. are the carbon monoxide, hydrocarbon and nitrogen oxide emissions of vehicle l, c CO 、c HC , They are the emission costs per unit mass of carbon monoxide, hydrocarbons and nitrogen oxides respectively.

[0068]

[0069] The consumption of fossil fuels by ground support vehicles during operation will increase the carbon emissions of airports. The calculation formula is as follows. Where, c e is the CO2 emission cost per unit mass, It is the carbon dioxide emission index of vehicle fuel.

[0070]

[0071] The vehicle usage time includes the vehicle driving time, waiting time and task execution time, that is, the total time from the vehicle leaving the parking lot to returning to the parking lot. The time cost of vehicle l is T l vehicle The calculation formula is as follows. is the unit time cost of ground support vehicles, RPT l The time when vehicle l returns to the parking lot, DPT l The time when vehicle l leaves the parking lot.

[0072]

[0073] (2) A joint scheduling model for apron resources is established based on two-layer planning. The upper model is the parking space assignment model, and the lower model is the ground support vehicle scheduling model. The total cost of apron resource scheduling is minimized while meeting the apron resource capacity constraints.

[0074] The specific steps in step (2) include:

[0075] 2.1, assign appropriate parking spaces to aircraft, take the minimum total cost of parking space assignment as the objective function, and build an upper-level planning model for parking space assignment. The formula is as follows. The decision variable X ik Indicates whether flight i is assigned to parking stand k.

[0076]

[0077] 2.2, arrange reasonable service routes for ground support vehicles to complete the services of four types of vehicles, take the minimum total cost of vehicle scheduling as the objective function, and build a lower-level planning model for ground support vehicle scheduling. The formula is as follows. Among them, the decision variables Indicates whether vehicle l in model q serves flight i.

[0078]

[0079] 2.3, the parking lot assignment scheme obtained by the upper model is passed as input to the lower model, and the ground support vehicle scheduling scheme and scheduling cost obtained by the lower model are returned to the upper model. The two are optimized through continuous iteration and finally reach the overall optimal. The idea of ​​the two-level planning model is as follows Figure 2 shown.

[0080] (3) To solve the joint scheduling of apron resources, the upper layer uses an improved genetic algorithm and the lower layer uses a brainstorming algorithm to construct an improved hybrid genetic brainstorming algorithm (AHGBA) for solution.

[0081] The specific steps in step (3) include:

[0082] 3.1, for the upper parking lot assignment problem, an improved genetic algorithm is designed, which uses adaptive crossover and mutation probability, increases memory operations, and improves algorithm efficiency.

[0083] The adaptive crossover probability Pc is calculated as follows:

[0084]

[0085] Where, Pc max Indicates the maximum crossover probability; Pc min represents the minimum crossover probability; fit' represents the individual fitness function value; fit avg Represents the average value of the population fitness function; fit max Represents the maximum value of the fitness function in the population.

[0086] The adaptive mutation probability Pm is calculated as follows:

[0087]

[0088] Where Pm max Indicates the maximum probability of mutation; Pm min Indicates the minimum mutation probability.

[0089] 3.2. For the problem of lower-level ground support vehicle scheduling, a brainstorming algorithm is designed, and the local search operation is used to improve the efficiency of the algorithm. The local search operation draws on the ideas of "destruction" and "repair" in the neighborhood search algorithm. First, several flights are removed from the current plan to complete the destruction operation, and then the removed flights are reinserted into the destroyed plan to complete the repair operation. The specific steps are as follows:

[0090] Step 1: Specify the number of flights to be removed.

[0091] Step 2: According to the relevance principle, remove the specified number of flights from the original plan to form a set Move. The destroyed plan is S destroy .

[0092] Step 3: Initialize the repair solution, S repair =S destroy .

[0093] Step 4: If Move is not empty, proceed to the next step, otherwise jump to Step 8.

[0094] Step 5: Calculate all flights in Move and insert them into S repair Regret value.

[0095] Step 6: Find the sequence number of the largest item among all regret values, find the flight to be inserted back according to the sequence number, and finally insert the flight back to S repair The position where the regret cost is minimal.

[0096] Step7: Update Move and jump to Step4.

[0097] Step 8: Complete the repair and output the repaired solution S repair.

[0098] 3.3, based on the idea of ​​evolutionary iteration, an improved hybrid genetic brainstorming algorithm (AHGBA) is constructed to solve the bi-level programming model. The specific steps are as follows:

[0099] Step 1: First, initialize the upper-level planning decision variables and generate a random initial solution, i.e., the parking space assignment scheme for each aircraft;

[0100] Step 2: The lower-level planning generates the optimal vehicle and route scheduling for each individual in the upper-level planning population, and feeds back the optimal scheduling plan to the upper-level planning;

[0101] Step 3: The upper-level planning uses the fitness function to calculate the optimal objective function value of each individual;

[0102] Step 4: Perform improved genetic operations on the population and use iteration to approximate the optimal solution of the model.

[0103] The technical solution of the present invention is further illustrated by a specific embodiment below.

[0104] A large airport in my country is selected as the research object (hereinafter referred to as Airport A). Airport A currently has two terminals, three runways, and 224 passenger aircraft stands, including 142 close-in stands and 82 far-out stands. The domestic aircraft stands of Terminal T2 of Airport A are used as the research object for simulation verification. The research scope includes the West Concourse 5, West Concourse 6, West 4 overnight stands, and T2 frequently used far-out stands. The parking stand system has a total of 36 stands, including 29 close-in stands and 7 far-out stands. Its configuration is as follows: Figure 3 shown.

[0105] (1) Cost calculation related data

[0106] The fuel consumption rate during aircraft taxiing uses the fuel flow rate of each aircraft type at 7% thrust, and the data comes from the engine emission database compiled by ICAO; the aircraft surface taxiing speed is 45km / h; the aircraft fuel cost is 8RMB / L.

[0107] The data of ground support vehicles mainly include vehicle speed, vehicle unit fuel cost, tanker capacity and the number of different models available. The vehicle speed is 30km / h; the unit fuel cost of the vehicle is 8.43RMB / L; the maximum capacity of the tanker is 45000L. The number of four types of vehicles available is calculated based on the "Guidelines for the Configuration of Special Equipment for Flight Support at Civil Transport Airports" formulated by the Airport Department of the Civil Aviation Administration of China.

[0108] (2) Solution of apron resource scheduling based on bi-level programming model

[0109] The calculation results show that the final objective function value is 92,936 yuan when the operating cost is taken into account, the upper parking space assignment cost is 59,212 yuan, and the lower ground support vehicle dispatch cost is 33,724 yuan. Figure 4 shown.

[0110] Table 1 shows the parking space allocation plan.

[0111] Table 1 Parking stand assignment scheme

[0112]

[0113] A total of 19 ground support vehicles were used, including 8 tractors, 4 shuttles, 2 tanker trucks, and 5 pipeline trucks. The flight service plan for each vehicle type is shown in Table 2.

[0114] Table 2 Ground support vehicle dispatch plan

[0115]

[0116] Figure 5 Assign a Gantt chart to the parking stands, Figure 6 Gantt chart for vehicle service support.

[0117] The waiting time for tractors and shuttles is long. These two types of vehicles are subject to strict time constraints, so the main way to reduce costs is by driving distance. Tank refueling trucks are subject to strict vehicle resource capacity constraints, so the main way to reduce costs is by the number of vehicles used. Pipeline refueling trucks have relatively loose service start time constraints, so this type of vehicle has almost no waiting time. Since there are no resource constraints, they can return to the parking lot after completing all tasks. This type of vehicle mainly reduces costs by vehicle usage time.

[0118] (3) Results comparison

[0119] Using the same data and parameters, the results of the apron resource joint scheduling model proposed in the present invention are compared with the results of the independent scheduling model. The comparison of the total cost and the cost of each layer of the two scheduling models is shown in Table 3. While ensuring that the upper-level parking space assignment cost remains basically unchanged, the joint scheduling reduces the lower-level vehicle scheduling cost by 18.74%, the number of vehicles by 17.39%, and the vehicle driving distance by 9.22%. This shows that the joint scheduling takes into account the convenience provided by the parking space assignment scheme for vehicle scheduling from the perspective of global optimization, and ultimately reduces the total scheduling cost and the number of vehicle resources used, thereby achieving the purpose of improving the efficiency of apron services.

[0120] Table 3 Comparison of cost values ​​of each layer between independent scheduling and joint scheduling

[0121]

[0122] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any technician familiar with the profession, without departing from the scope of the technical solution of the present invention, according to the technical essence of the present invention, any simple modification, equivalent replacement and improvement made to the above embodiment still falls within the protection scope of the technical solution of the present invention.

Claims

1. A joint scheduling method for apron resources based on double-layer planning, characterized in that: The following steps are involved: Step 1: Divide the apron resource scheduling cost into the parking space assignment cost and the ground support vehicle scheduling cost, and propose a cost quantification method that includes four factors: fuel, pollutants, carbon emissions, and time; Step 2: Establish a joint scheduling model for apron resources based on two-layer planning. The upper model is the parking space assignment model, and the lower model is the ground support vehicle scheduling model. The total cost of apron resource scheduling is minimized while meeting the apron resource capacity constraints. Step 3: For the joint scheduling stage of apron resources, the upper layer uses an improved genetic algorithm and the lower layer uses a brainstorming algorithm to construct an improved hybrid genetic brainstorming algorithm for solution.

2. The method for joint scheduling of apron resources based on double-layer planning according to claim 1 is characterized in that: The specific process of step 1 is as follows: Step 1.1: Divide the parking slot assignment cost into four parts: aircraft fuel cost, aircraft pollutant cost, aircraft carbon emission cost and aircraft time cost; Step 1.2: Divide the ground support vehicle dispatch cost into four parts: vehicle fuel cost, vehicle pollutant cost, vehicle carbon emission cost and vehicle time cost.

3. The method for joint scheduling of apron resources based on double-layer planning according to claim 2 is characterized in that: The parking stand assignment cost is divided into: Aircraft fuel cost F incurred by flight i i aircraft The calculation formula is as follows: in, is the total taxiing time for flight i, Refers to the unit mass aviation fuel cost, r i is the fuel consumption rate of one engine at 7% thrust on flight i, n i refers to the number of engines of flight i; The total pollutant cost P of flight i during the taxiing phase i aircraft The calculation formula is as follows: Among them, P i CO is the carbon monoxide environmental pollution cost of flight i, P i HC is the hydrocarbon environmental pollution cost of flight i, is the environmental pollution cost of nitrogen oxides for flight i; Carbon emission cost of flight i The calculation formula is as follows: Among them, c e is the CO2 emission cost per unit mass, is the aviation fuel CO2 emission index; The total time cost T of flight i i aircraft The calculation formula is as follows: in, is the waiting time of flight i at the parking stand, is the unit time cost of the flight, is the delay time cost of flight i; In the ground support vehicle dispatch cost, The fuel consumption cost F of vehicle l l vehicle The calculation formula is as follows: Among them, TD l is the total distance traveled by vehicle l, v vehicle is the vehicle speed, is the vehicle fuel cost per unit time, u is the vehicle fuel consumption; The total pollution cost P of vehicle l l vehicle The calculation formula is as follows: in, are the carbon monoxide, hydrocarbon and nitrogen oxide emissions of vehicle l, c CO 、c HC , They are the emission costs per unit mass of carbon monoxide, hydrocarbons, and nitrogen oxides; Carbon emission cost of vehicle l The calculation formula is as follows: Among them, c e is the CO2 emission cost per unit mass, is the CO2 emission index of vehicle fuel; The time cost of vehicle l is T l vehicle The calculation formula is as follows: in, is the unit time cost of ground support vehicles, RPT l The time when vehicle l returns to the parking lot, DPT l The time when vehicle l leaves the parking lot.

4. The method for joint scheduling of apron resources based on double-layer planning according to claim 3 is characterized in that: The specific process of step 2 is as follows: Step 2.1: Assign appropriate parking stands to aircraft, take the minimum total cost of parking stand assignment as the objective function, and build an upper-level planning model for parking stand assignment; Step 2.2: Arrange reasonable service routes for ground support vehicles to complete the services of the four types of vehicles, take the minimum total cost of vehicle scheduling as the objective function, and build a lower-level planning model for ground support vehicle scheduling; Step 2.3: The parking space assignment solution obtained by the upper model is passed as input to the lower model, and the ground support vehicle scheduling solution and scheduling cost obtained by the lower model are returned to the upper model. The two are optimized through continuous iteration and finally reach the overall optimal solution.

5. The method for joint scheduling of apron resources based on double-layer planning according to claim 4 is characterized in that: The formula of the upper-level planning model is as follows: Among them, the decision variable X ik Indicates whether flight i is assigned to parking stand k; The formula of the lower-level planning model is as follows: Among them, the decision variables Indicates whether vehicle l in model q serves flight i.

6. The method for joint scheduling of apron resources based on double-layer planning according to claim 1 is characterized in that: The specific process of step 3 is as follows: Step 3.1: Design an improved genetic algorithm for the upper-level parking lot assignment problem, use adaptive crossover and mutation probabilities, add memory operations, and improve algorithm efficiency; Step 3.2: Design a brainstorming algorithm for the lower-level ground support vehicle scheduling problem and use local search operations to improve algorithm efficiency; Step 3.3: Based on the idea of ​​evolutionary iteration, an improved hybrid genetic brainstorming algorithm is constructed to solve the bi-level programming model.

7. The method for joint scheduling of apron resources based on double-layer planning according to claim 6 is characterized in that: In step 3.1, The adaptive crossover probability Pc is calculated as follows: Where, Pc max Indicates the maximum crossover probability; Pc min represents the minimum crossover probability; fit' represents the individual fitness function value; fit avg Represents the average value of the population fitness function; fit max Represents the maximum value of the fitness function in the population; The adaptive mutation probability Pm is calculated as follows: Where Pm max Indicates the maximum probability of mutation; Pm min Indicates the minimum mutation probability.

8. The method for joint scheduling of apron resources based on double-layer planning according to claim 6 is characterized in that: The specific process of step 3.2 is as follows: Step 1: Specify the number of flights to be removed; Step 2: According to the relevance principle, remove the specified number of flights from the original plan to form a set Move. The destroyed plan is S destroy ; Step 3: Initialize the repair solution, S repair =S destroy ; Step 4: If Move is not empty, proceed to the next step, otherwise jump to Step 8; Step 5: Calculate all flights in Move and insert them into S repair Regret value; Step 6: Find the sequence number of the largest item among all regret values, find the flight to be inserted back according to the sequence number, and finally insert the flight back to S repair The position with the minimum regret cost; Step7: Update Move and jump to Step4; Step 8: Complete the repair and output the repaired solution S repair .

9. The method for joint scheduling of apron resources based on double-layer planning according to claim 6 is characterized in that: The specific process of step 3.3 is as follows: Step 1: First, initialize the upper-level planning decision variables and generate a random initial solution, i.e., the parking space assignment scheme for each aircraft; Step 2: The lower-level planning generates the optimal vehicle and route scheduling for each individual in the upper-level planning population, and feeds back the optimal scheduling plan to the upper-level planning; Step 3: The upper-level planning uses the fitness function to calculate the optimal objective function value of each individual; Step 4: Perform improved genetic operations on the population and use iteration to approximate the optimal solution of the model.

Images