Multi-vehicle-type ground support vehicle joint scheduling method and system for large airport

Abstract

The invention discloses a multi-vehicle-type ground support vehicle combined scheduling method and system for a large airport. The method comprises the following steps: establishing a multi-vehicle-type ground support vehicle combined scheduling model according to a ground support vehicle operation mode; a two-stage heuristic algorithm is designed to solve a multi-vehicle-type ground support vehicle joint scheduling problem, wherein in the first stage, a single-vehicle-type ground support vehicle scheduling model is solved; in the second stage, according to a single-vehicle-type scheduling result, combined cooperation between different vehicle types is considered, and a multi-vehicle-type combined airport ground support vehicle optimal scheduling scheme is obtained. According to the method, the actual scene of ground support vehicle scheduling is considered, a multi-target optimization model is established from single-vehicle-type ground support vehicle scheduling to multi-vehicle-type mutually-matched joint scheduling, and an improved NSGA-II algorithm and a tabu search algorithm are designed to solve the problem, so that the use cost of the ground support vehicle and the vehicle running distance are reduced, and the service life of the ground support vehicle is prolonged. And the efficiency of the airport ground guarantee link is improved, and practicability and applicability are achieved.

Description

Technical Field

[0001] The present invention relates to the technical fields of airport ground support, multi-objective optimization, vehicle path planning and scheduling, and in particular to a joint scheduling method and system for multi-type ground support vehicles for large airports. Background Art

[0002] Since airport ground support services have the characteristics of multiple subjects, time constraints and limited resources, unreasonable scheduling methods will not only increase the workload, but also lead to low support efficiency. The scheduling research for a single vehicle type has become saturated and can no longer meet the actual scheduling needs of airports. Therefore, in order to avoid the fragmentation of scheduling research, in-depth analysis of the operating characteristics of different types of vehicles and joint scheduling of different types of vehicles from an overall perspective have become urgent issues to be solved in the current airport operation. In recent years, domestic and foreign scholars have studied the research results on the scheduling of airport ground support vehicles, mainly focusing on single-vehicle scheduling. There are few studies on the joint scheduling of multi-vehicle ground support vehicles, and even fewer studies that consider the vehicle operation mode in the joint scheduling process. Most of the joint-related research is to add timing constraints between different vehicle types in the scheduling model to reflect the jointness, and use the operating time of a certain vehicle type as a constraint for the scheduling of another vehicle type. The independence between each vehicle type is high, and it is difficult to truly reflect the joint scheduling process of multi-vehicle ground support vehicles. Aiming at the problem that the continuously growing civil aviation transport volume has caused tremendous pressure on airport ground support services, from the perspective of ground operation efficiency and safety, a scheduling model that minimizes the number of vehicles used and the operating distance is proposed based on the joint services of ground support vehicles in different operating modes. This model provides airports with a more efficient ground support vehicle scheduling solution to cope with the increasingly tense ground support service demand, thus generating practical application value. Summary of the invention

[0003] The purpose of this invention is to propose a joint dispatching method for multi-type ground support vehicles for large airports, aiming to jointly dispatch multi-type vehicles at large airports, improve the efficiency of airport ground support links, and provide reference guidance for airports to make relevant decisions on ground support vehicle dispatching.

[0004] To achieve the above purpose, the technical solution provided by the present invention is:

[0005] A method for joint dispatching of multi-type ground support vehicles for large airports, characterized by comprising the steps of:

[0006] S1: Establish a joint dispatch model for multi-type ground support vehicles at large airports;

[0007] S2: Design an adaptive NSGA-Ⅱ algorithm integrating local search to solve the ground support vehicle scheduling model of a single vehicle type;

[0008] S3: Determine the weight of each objective function in decision making and then select the optimal solution;

[0009] S4: Design a taboo search (TS) algorithm to solve the scheduling scheme for multiple vehicle types

[0010] In step S1, the ground support vehicle operation modes are divided into three types: continuous operation, continuous operation with resource capacity constraints, and round-trip operation. Conveyor vehicles, water trucks, and baggage tractors are selected as key research models for the three operation modes. The total number of ground support vehicles in use and the total operating distance are taken as optimization objects, and a joint scheduling model for ground support vehicles of multiple models is established.

[0011] In step S2, an integer coding form is used to construct chromosomes with the flight service order of each ground support vehicle as the gene value; in the initialization stage, the mileage saving (CW) algorithm is used to generate the initial solution of the ground support vehicle scheduling model that meets the requirements of a single vehicle type; the design of the adaptive NSGA-Ⅱ algorithm integrating local search includes: calculation of congestion, selection, adaptive crossover and adaptive mutation operations, adaptive local search operations, and elite selection strategy;

[0012] In step S3, for multi-objective optimization problems, it is necessary to evaluate the importance of the optimization objectives, and select the objective assignment method based on the Pareto optimal solution set to introduce the C-OWA operator to determine the objective function weight and select the most suitable solution;

[0013] In step S4, the chromosomes of the optimal scheduling solutions of the three types of vehicles are merged into a matrix-coded chromosome, and the generated chromosome is corrected by a destruction and repair method, so as to obtain a feasible initial solution for the joint scheduling of ground support vehicles of multiple types of vehicles; the design content of the taboo search algorithm for the joint scheduling of ground support vehicles of multiple types of vehicles includes: neighborhood generation, evaluation function and neighbor selection strategy, taboo table and special criteria;

[0014] The multi-model ground support vehicle joint scheduling method for large airports adopts the adaptive NSGA-Ⅱ algorithm integrated with local search to solve the single-model ground support vehicle scheduling model. The specific steps are as follows:

[0015] Step (1): Generate the initial parent population;

[0016] Step (2): Calculate the crowding degree of the parent population. The essence of crowding degree is to reflect the similarity between individual f and other individuals. The greater the crowding degree, the smaller the similarity. Therefore, when performing selection operations on the same frontier, individuals with greater crowding degree are more likely to be selected to maintain the diversity of the population.

[0017] Step (3): Initialize the offspring population; the selection operation adopts the binary tournament principle. Randomly select two individuals from the parent population and select the one with the smaller non-dominated rank; if the non-dominated ranks are the same, select the one with the larger crowding degree;

[0018] Step (4): Execute adaptive crossover, mutation and local search operations. The adaptive crossover operator is used for the crossover operation, and the adaptive mutation operator is used for the mutation operation. The core of the local search operation comes from the neighborhood search algorithm, which removes the flights that need to be adjusted from the individuals through the destruction operator, and then reinserts these flights into the vehicle path according to the repair operator to obtain a new vehicle path. There are two types of destruction operators used in the present invention, namely the worst removal destruction operator and the random destruction removal operator; there are two types of repair operators used, namely the minimum cost insertion repair operator and the maximum regret value insertion repair operator;

[0019] Step (5): Merge the parent population and the child population, and perform non-dominated sorting, crowding calculation and offset calculation on the merged population.

[0020] Step (6): Apply the elite selection strategy to the merged population. Sort the parent population in the order of non-dominated level, crowding degree and deviation degree, and select a specified number of individuals as the offspring population.

[0021] Step (7): Update the number of iterations, repeat steps (3) to (6) until the maximum number of iterations is reached, return to find the final population and stop the program.

[0022] The specific steps of the adaptive NSGA-Ⅱ algorithm integrating local search and the elite selection strategy are as follows:

[0023] Step (1): Sort the individuals according to the non-dominated level from small to large, starting from the non-dominated level 1, and put the individuals in the entire frontier into the next generation parent population, followed by the individuals of the next non-dominated level, until all the individuals of a certain level cannot be put into the next generation parent population, then go to the next step;

[0024] Step (2): Arrange the individuals of this level in descending order of crowding, and place the individuals with higher crowding degrees into the next generation parent population in turn, until all individuals with the same crowding degree cannot be placed into the next generation parent population, and then proceed to the next step;

[0025] Step (3): Sort the remaining individuals with the same crowding degree from small to large according to the degree of deviation, and prioritize individuals with small deviations into the next generation parent population until the next generation parent population is filled.

[0026] The method for joint dispatching of multi-type ground support vehicles for large airports and the taboo search algorithm for joint dispatching of multi-type ground support vehicles have the following specific steps:

[0027] Step (1): Based on the optimal solution for the dispatch of single-model support vehicles, the chromosomes of the optimal dispatch solutions of the three models are merged into a matrix-coded chromosome, and the generated chromosome is corrected by the destruction and repair method to obtain a feasible initial solution for the joint dispatch of multi-model ground support vehicles. Clear the taboo table and set the taboo length.

[0028] Step (2): For the current local optimal solution X * Iterative learning is performed to obtain the overall neighborhood movement set and the overall neighborhood candidate solution set according to the neighborhood generation strategy, and the corresponding rating function value of each candidate solution is calculated.

[0029] Step (3): Select the candidate solution X in the neighborhood with the smallest evaluation function value min , if Z(X min )<Z(X * ), then X min Become the new local optimal solution X * , and its corresponding neighborhood moves are included in the taboo table. Then, best For comparison, if Z(X min )<Z(X best ), then X min Become the new global optimal solution. If the neighborhood move that can generate the global optimal solution or the local optimal solution is in the taboo table, then the neighborhood move is unbanned. After the comparison is completed, the taboo table is updated to determine whether the current number of iterations has reached the maximum number of iterations. If not, return to step (2); if so, go to step (4).

[0030] Step (4): Stop searching after reaching the maximum number of iterations and output the current global optimal solution X best , the solution is an airport ground support vehicle scheduling scheme for multiple vehicle types.

[0031] The taboo search algorithm for joint dispatch of multi-type ground support vehicles has the following steps for generating a neighborhood:

[0032] Step (1): First, perform neighborhood generation operation on the path of the water truck in the continuous operation mode, and optimize its path separately; because the path planning of vehicles in the continuous operation mode is not affected by the coordination constraint, their optimized paths are determined first; the corresponding chromosome X wv Neighborhood generation strategy using the “break and repair” philosophy;

[0033] Step (2): Generate separate neighborhood movement sets for the baggage tractor and conveyor truck, and combine them with the optimal neighborhood movement operation of the water truck to obtain the overall neighborhood movement set; the "destruction and repair" neighborhood generation strategy is also used for the baggage tractor and conveyor truck.

[0034] The present invention also proposes a joint dispatching method and system for multi-type ground support vehicles for large airports, including:

[0035] The model building module is used to establish a joint scheduling model for ground support vehicles of multiple models. The model takes the total number of ground support vehicles in use and the total operating distance as the optimization objects, combines the unique operating characteristics of different models, and takes into account that some support links require different models to complete jointly. A joint scheduling model for ground support vehicles of multiple models is established.

[0036] The solution module is used to design an adaptive NSGA-Ⅱ algorithm integrating local search to solve the single-model ground support vehicle scheduling model, and to design a taboo search algorithm for joint scheduling to obtain a solution for the joint scheduling of multiple-model ground support vehicles;

[0037] The present invention also protects an electronic device comprising:

[0038] A memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for joint dispatching of multi-type ground support vehicles for large airports is implemented.

[0039] The present invention also protects a computer-readable storage medium comprising:

[0040] A computer program is stored, and the computer program enables the computer to execute the joint dispatching method of multi-type ground support vehicles for large airports.

[0041] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0042] The present invention considers the actual scenario of multi-model vehicle dispatching at large airports, and divides the ground support vehicle operation mode into three types according to the operation characteristics of the field vehicles, namely, continuous operation, continuous operation with resource capacity limitation, and round-trip operation. In the three operation modes, conveyor belt vehicles, water trucks, and baggage tractors are selected as key research models, and a multi-model ground support vehicle joint dispatching model is established. Firstly, an adaptive NSGA-Ⅱ algorithm integrating local search is proposed to solve the single-model ground support vehicle dispatching model. The advantage of this improved algorithm is that it uses the mileage saving algorithm to generate high-quality initial solutions, and integrates the adaptive local search into the NSGA-Ⅱ algorithm framework, which improves the search ability of the algorithm and makes it more suitable for vehicle dispatching scenarios. Subsequently, a taboo search algorithm for multi-model dispatching is adopted to merge the chromosomes of the optimal dispatching schemes of the three models into matrix-coded chromosomes, and the generated chromosomes are corrected by the destruction and repair method to obtain a feasible initial solution for the joint dispatching of multi-model ground support vehicles. A neighborhood generation strategy for destruction and repair is designed, and a rating function is solved to obtain a multi-model field vehicle dispatching scheme, which improves the efficiency of the airport ground support link and reduces the airport field operation cost.

[0043] The present invention is based on the proposed joint dispatching method and system for multi-model ground support vehicles for large airports, and uses the T2 terminal of a large airport as an example study area to design a multi-model field vehicle dispatching scheme. Through comparative analysis with other heuristic solution results, it can be obtained that the NSGA-Ⅱ algorithm proposed in the present invention, which integrates adaptive local search, has good global optimization ability, and after the joint optimization of the TS algorithm, the number of vehicles used by each vehicle has basically not changed, but the total driving distance of some models has a certain degree of slight decrease, indicating that the multi-model joint TS algorithm designed in this paper has the ability to continue to optimize the objective function while solving the multi-model joint scheme. In addition, after the joint dispatching of multi-model ground support vehicles, the balance degree of the task volume of each model is similar to that of a single model, and a relatively balanced task distribution is still maintained. The joint dispatching method for multi-model ground support vehicles for large airports can provide a better solution for the multi-model dispatching of airport scenes, has high practical value, and is of great significance for providing guidance on the joint dispatching scheme of multi-model vehicles for airports. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of the joint dispatch method for multi-type ground support vehicles for large airports.

[0045] Figure 2 Figure 1 shows the three operating modes of ground support vehicles, where (1) is the continuous operating mode, (2) is the continuous operating mode with resource capacity constraints, and (3) is the round-trip operating mode.

[0046] Figure 3Schematic diagram of elite selection strategy.

[0047] Figure 4 The following is a sample airport parking space and lane distribution map.

[0048] Figure 5 Scatter plots of the solution results of LS-ANSGAⅡ, ANSGAⅡ, TS, LNS and FCFS, where (a) is a conveyor belt car, (b) is a water truck, and (c) is a baggage tractor. DETAILED DESCRIPTION

[0049] 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.

[0050] The present invention proposes a joint dispatching method for multi-type ground support vehicles for large airports. The flow chart is as follows: Figure 1 As shown. Considering the actual scenario of multi-model vehicle dispatching at the airport, the ground support vehicle operation modes are divided into three types: continuous operation, continuous operation with resource capacity constraints, and round-trip operation. Conveyor vehicles, water trucks, and baggage tractors are selected as key research models in the three operation modes. The total number of ground support vehicles in use and the total operating distance are taken as optimization objects. A multi-model ground support vehicle joint dispatching model is established, and an adaptive NSGA-Ⅱ algorithm integrating local search is designed to solve the single-model ground support vehicle dispatching model. The weight of each objective function in the decision-making is determined, and then the optimal solution is selected and the taboo search (TS) algorithm is designed to solve the multi-model joint dispatching plan. The following steps are included:

[0051] (1) Considering the actual scenario of multi-type vehicle dispatching at the airport, the ground support vehicle operation modes are divided into three types: continuous operation, continuous operation with resource capacity constraints, and round-trip operation. A multi-type ground support vehicle joint dispatching model is established.

[0052] The specific steps in step (1) include:

[0053] 1.1, Problem Description and Assumptions: According to the three operating modes of ground support vehicles, such as Figure 2 As shown in the figure, the conveyor truck, water truck and baggage tractor are selected as the key research models. Considering the computational complexity and model adaptability, information assumptions, flight assumptions and vehicle performance assumptions are made for the multi-model vehicle joint scheduling optimization model.

[0054] 1.2. Model establishment: The model establishment part first explains the numbers and parameters, and then establishes a multi-objective optimization model in which the first-level decision is the dispatch of ground support vehicles of a single type, the second-level decision is to solve the joint dispatch plan of ground support vehicles of multiple types, and the goal is to minimize the total number of support vehicles and the minimum total operating distance.

[0055] In step 1.1, the different assumptions made for the drone inspection model are as follows:

[0056] Information assumptions: The basic information of flights and vehicles is complete and correct and will not be changed;

[0057] Flight assumptions: Once each ground support service begins, it will not be interrupted due to any circumstances until the support mission is completed; transit flights are divided into inbound flights and outbound flights;

[0058] Vehicle Performance Assumptions: There is no performance difference between ground support vehicles of the same type, and the relevant vehicle operators have the same working ability and efficiency. All vehicles depart from the depot and return to the depot after completing the service, and the driving speed remains unchanged.

[0059] The specific steps in step 1.2 include:

[0060] 1.2.1, the total number of ground support vehicles used and the total distance of operation are optimized. Combined with the unique operating characteristics of different models, considering that some support links require different models to complete jointly, a joint scheduling model for ground support vehicles with multiple models is established. The model objectives are as follows:

[0061]

[0062] Formula (1) represents the minimization of the number of vehicles in use of the three types of vehicles studied in the present invention; Formula (2) represents the minimization of the total distance traveled by the three types of vehicles when in operation.

[0063] 1.2.2, Constraints are established.

[0064]

[0065] Formula (3) to (8) are common constraints that all vehicle types must follow. Formula (3) is the flight assignment uniqueness constraint, which means that each type of service flight is accepted only once; Formulas (4) to (6) are vehicle service uniqueness constraints, which means that each type of ground support vehicle will not repeatedly serve a flight; Formula (7) is the vehicle quantity constraint, which means that the number of each type of vehicle used does not exceed the number of available vehicles; Formula (8) is the vehicle service start time constraint, which means that the start time of flight i accepting a certain vehicle type service is within its start time window.

[0066]

[0067] Formula (9) and (10) are specific constraints for the clean water truck. Formula (9) indicates that the clean water truck is empty when it leaves the parking lot; Formula (10) indicates that the remaining clean water volume of the clean water truck during operation cannot exceed the maximum volume of the water tank, nor can it be less than 0.

[0068]

[0069] Formula (11) to (13) are the joint operation constraints of the conveyor belt vehicle and the baggage tractor. Formula (11) indicates that the first conveyor belt vehicle used for service must arrive before the first baggage tractor; Formula (12) indicates that baggage service can only be started when the conveyor belt vehicle and the baggage tractor arrive at the service point at the same time; Formula (13) indicates that the conveyor belt vehicle can only end its service after all baggage tractors serving a certain flight have finished their service.

[0070] (2) This problem is a variant application of the VRP problem and is an NP-Hard problem. Therefore, an adaptive NSGA-Ⅱ algorithm integrating local search is first designed to solve the ground support vehicle scheduling model of a single vehicle type.

[0071] The specific steps of the adaptive NSGA-Ⅱ algorithm integrated with local search to solve the ground support vehicle scheduling model of a single vehicle type are as follows:

[0072] Step 1: Generate the initial parent population;

[0073] Step 2: Calculate the crowding of the parent population. The essence of crowding is to reflect the similarity between individual f and other individuals. The greater the crowding, the smaller the similarity. Therefore, when selecting in the same frontier, individuals with greater crowding are more likely to be selected to maintain the diversity of the population.

[0074] Step 3: Initialize the offspring population; the selection operation adopts the binary tournament principle. Randomly select two individuals from the parent population and choose the one with the smaller non-dominated rank; if the non-dominated ranks are the same, choose the one with the larger crowding degree;

[0075] Step 4: Perform adaptive crossover, mutation and local search operations. The adaptive crossover operator is used for the crossover operation, and the adaptive mutation operator is used for the mutation operation. The core of the local search operation comes from the neighborhood search algorithm, which removes the flights that need to be adjusted from the individuals through the destruction operator, and then reinserts these flights into the vehicle path according to the repair operator to obtain a new vehicle path. The present invention uses two types of destruction operators, namely the worst removal destruction operator and the random destruction removal operator; the repair operators used are of two types, namely the minimum cost insertion repair operator and the maximum regret value insertion repair operator;

[0076] Step 5: Merge the parent population and the child population, and perform non-dominated sorting, crowding calculation, and offset calculation on the merged population.

[0077] Step 6: Apply elite selection strategy to the merged population. Sort the parent population in the order of non-dominated level, crowding degree and deviation degree, and select a specified number of individuals as the offspring population.

[0078] Step 7: Update the number of iterations, repeat steps 3 to 6 until the maximum number of iterations is reached, return to find the final population and stop the program.

[0079] The specific steps in step (2) include:

[0080] 2.1. Generation of initial solution. In order to ensure the quality and iteration efficiency of the initial solution for single vehicle type scheduling, the present invention uses the CW algorithm to generate an initial solution for the ground support vehicle scheduling model that satisfies a single vehicle type.

[0081] 2.2, use the adaptive NSGA-Ⅱ algorithm integrated with local search to solve the ground support vehicle scheduling model of a single vehicle type.

[0082] In step 2.1:

[0083] Assuming that H flights are served by H vehicles, H loops will be generated. Calculate the reduction in transportation distance after merging any two loops, and merge the two loops with the largest reduction value until no other loops can be merged.

[0084] For the airport ground support vehicle scheduling of a single vehicle model, the present invention adopts a method of inserting a single flight into the vehicle path according to the size of the savings value in the path merging. The specific process is as follows: Whenever a new path set VC is generated, a savings value matrix sav[i][k][n][d] of size n*4 will be obtained at the same time. The meaning of each row of sav is that the path length savings value generated by the selected flight i inserted into the nth gap of path k is d, and the insertion operation represented by each row satisfies the time window constraint, and the entire sav is arranged in descending order according to the size of d. Subsequently, the insertion operation represented by the first row in sav will be used to generate the expected new path set pre_VC, and it will be judged whether the path set meets the constraints specific to each vehicle model. If the relevant constraints are violated, the insertion operation represented by the next row in sav will be selected until an insertion operation that meets all constraints is found and a new path set VC is generated. After several cycles of operation, when a certain VC can no longer generate sav, it means that the CW algorithm cannot continue to merge the paths, so that the VC is used as the initial solution for the vehicle scheduling of the corresponding vehicle model.

[0085] The specific steps in step (3) are:

[0086] 3.1, Standardization of objective function values. For the dual-objective problem, the Pareto optimal solution set of the problem is obtained using a heuristic algorithm, and the individual solutions in the obtained Pareto optimal solution set are numbered from top to bottom as 1, 2, ..., f, ..., g. The objective function value Z of each individual solution using the number of vehicles and the driving distance is 1 (f) and Z 2 (f) Divided by the maximum value of the number of vehicles used and the driving distance in the optimal solution set and Complete standardization;

[0087] 3.2, calculate the weight of the combination of each objective function value of each individual and the weight of each objective;

[0088] 3.3, the standardized values ​​of the objective functions of all individuals are multiplied by their corresponding objective function values ​​to obtain the weighted objective function values ​​and generate a weighting matrix;

[0089] 3.4. Find the maximum and minimum values ​​of each weighted objective function value, and then calculate the distance between each individual and the optimal level and the worst level. The larger the distance, the closer it is to the optimal level of optimization, and the corresponding vehicle model scheduling is the optimal solution.

[0090] The specific steps in step 3.2 are:

[0091] 3.2.1, Calculation of weights of each objective function and combined weights:

[0092] The calculation formula for the combined weight of each objective function is as follows:

[0093]

[0094] Where C represents the number of combinations.

[0095] The C-OWA operator is a weighted average method for processing continuous data. The C-OWA operator is introduced to calculate the weight of each objective function. The calculation formula is as follows:

[0096]

[0097] where w 1 ' and w 2 'represent the absolute weights of the two objective functions respectively; in formula (4-20), w 1 and w 2 It represents the relative weights corresponding to the two objective functions.

[0098] The specific steps in step 3.4 are:

[0099] 3.4.1, Calculation of the distance between the best level and the worst level:

[0100] The calculation formula of the weighted objective function value is as follows:

[0101]

[0102] Where Y 1 (f) and Y 2 (f) represent the weighted objective function values ​​of the number of vehicles used and the total distance traveled, respectively.

[0103] The specific steps in step 3.3 are:

[0104] 3.3.1, Calculation of weighted objective function value:

[0105] The calculation formula of the distance between the best level and the worst level is as follows:

[0106]

[0107] (4) Design a taboo search (TS) algorithm to solve the scheduling solution for multiple vehicle types

[0108] The specific steps in step (4) are:

[0109] 4.1, construct a feasible initial solution for the joint dispatch of multi-type ground support vehicles

[0110] 4.2, for the current local optimal solution X * Iterative learning is performed to obtain the overall neighborhood movement set and the overall neighborhood candidate solution set according to the neighborhood generation strategy, and the corresponding rating function value of each candidate solution is calculated.

[0111] 4.3, select the neighborhood candidate solution X with the smallest evaluation function value min , if Z(X min )<Z(X * ), then X min Become the new local optimal solution X * , and its corresponding neighborhood moves are included in the taboo table. Then, best For comparison, if Z(X min )<Z(X best ), then X min Become the new global optimal solution. If the neighborhood move that can generate the global optimal solution or the local optimal solution is in the taboo table, then the neighborhood move is unbanned. After the comparison is completed, update the taboo table and determine whether the current number of iterations has reached the maximum number of iterations. If not, return to step 4.2; if reached, go to step 4.4.

[0112] 4.4, stop searching after reaching the maximum number of iterations, and output the current global optimal solution X best , the solution is an airport ground support vehicle scheduling scheme for multiple vehicle types.

[0113] The specific steps in step 4.1 are:

[0114] 4.1.1, Initial solution construction and initialization:

[0115] According to the optimal solution for the dispatch of single-model support vehicles, the chromosomes of the optimal dispatch solutions of the three models are merged into a matrix-encoded chromosome, and the generated chromosome is corrected by the destruction and repair method to obtain a feasible initial solution for the joint dispatch of multi-model ground support vehicles. Clear the taboo table and set the taboo length.

[0116] The specific steps in step 4.2 are:

[0117] 4.2.1, Neighborhood generation action: The steps of neighborhood generation are as follows:

[0118] First, the neighborhood generation operation is performed on the vehicle paths of pipeline refueling trucks and water trucks, and the paths of these two types of vehicles are optimized separately. Because the path planning of these two types of vehicles is not affected by the collaborative constraints, their optimized paths can be determined first. The chromosomes corresponding to the above two types of vehicles and the neighborhood generation strategy using the "destruction and repair" concept are used.

[0119] Secondly, separate neighborhood move sets for baggage tractors and conveyor belt vehicles are generated, and combined with the optimal neighborhood move operations for pipeline refueling trucks and water trucks to obtain the overall neighborhood move set. The "destroy and repair" neighborhood generation strategy is also used for baggage tractors and conveyor belt vehicles.

[0120] 4.2.2 Calculation of evaluation function: The calculation formula of the evaluation function of the multi-vehicle collaborative scheduling taboo search process is as follows:

[0121]

[0122] Where Z(X) is the evaluation function, which is composed of a multi-objective single compromise function and a penalty function; X best The global optimal solution in the current iteration; Z 1 (X) and Z 2 (X) represents the two objective function values ​​corresponding to the solution X; and Z 1 With Z 2 The combined weight of the two objective functions; violation(X) is the number of times the conveyor belt car generated by the solution X violates the coordination constraint, and α is the penalty coefficient.

[0123] In summary, the present invention establishes a single scheduling model for each vehicle type according to the characteristics of the operation modes of the above three types of vehicles. Then, the single scheduling model of each vehicle type is integrated into a multi-vehicle joint ground support vehicle scheduling model, which takes into account the joint operation constraints between the baggage tractor and the conveyor belt vehicle, and sets a joint objective function to treat the scheduling models of all vehicle types as a whole.

[0124] For the airport ground support vehicle scheduling solution algorithm, it is necessary to fully understand whether the currently available ground support vehicle resources can complete the ground support services of all service flights under the maximum utilization degree. Therefore, the present invention designs an adaptive NSGAⅡ that integrates local search to solve the ground support vehicle scheduling model of a single vehicle type, and uses local search operations and adaptive crossover mutation operators to simultaneously enhance the global and local search capabilities. Subsequently, the obtained single-vehicle ground support vehicle scheduling model is fused and input into the taboo search algorithm for the joint scheduling of multi-vehicle ground support vehicles, and finally an airport ground support vehicle scheduling plan that meets various constraints is obtained.

[0125] Based on the above research, the present invention uses the real flight data of a large domestic airport as an example and conducts verification analysis. The T2 terminal of the airport is selected as the research space range, and October 2, 2019 10:00-18:00 is selected as the research time range. The example verification results show that the scheduling model and algorithm proposed in the present invention can effectively schedule airport ground support vehicles and thereby improve the efficiency of the airport ground support link, and have practicality and applicability.

Claims

1. A joint dispatching method for multi-type ground support vehicles for large airports, characterized in that: The following steps are involved: S1: Establish a joint dispatch model for multi-type ground support vehicles at large airports; S2: Design an adaptive NSGA-Ⅱ algorithm integrating local search to solve the single-type ground support vehicle scheduling model; S3: Determine the weight of each objective function in decision making and then select the optimal solution; S4: Design a taboo search algorithm to solve the optimal solution for the joint dispatch of multiple types of ground support vehicles; In step S1, the ground support vehicle operation modes are divided into three types: continuous operation, continuous operation with resource capacity constraints, and round-trip operation. In the three operation modes, conveyor belt vehicles, water cleaning vehicles, and baggage tractors are selected as research models, respectively. The total number of ground support vehicles used and the total operating distance are taken as optimization objects, and a joint scheduling model for ground support vehicles of multiple models and constraint conditions are established; In step S2, an integer coding form is used to construct a chromosome with the flight service order of each ground support vehicle as the gene value; In the initialization phase, a mileage-saving algorithm is used to generate an initial solution that satisfies the single-type ground support vehicle scheduling model; The design of the adaptive NSGA-Ⅱ algorithm integrating local search includes: calculation and selection of crowding degree, adaptive crossover and adaptive mutation operations, adaptive local search operations and elite selection strategy; In step S3, for multi-objective optimization problems, it is necessary to evaluate the importance of the optimization objectives, and select the objective assignment method based on the Pareto optimal solution set to introduce the C-OWA operator to determine the objective function weight and select the most suitable solution; In step S4, the chromosomes of the optimal scheduling solutions for the three types of vehicles are merged into a matrix-coded chromosome, and the generated chromosome is corrected by a destructive repair method to obtain a feasible initial solution for the joint scheduling of ground support vehicles of multiple types; the design content of the taboo search algorithm for the joint scheduling of ground support vehicles of multiple types includes: neighborhood generation, evaluation function and neighbor selection strategy, taboo table and special criteria.

2. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The objective function of the model in step S1 is as follows: Formula (1) represents the minimization of the number of three vehicle types used; Formula (2) represents the minimization of the total distance traveled by the three vehicle types when in operation; Where type is the ground support vehicle type, bv is a baggage tractor, cv is a conveyor belt vehicle, and wv is a water truck. is a 0-1 decision variable, indicating whether the k-th vehicle of type type is used; is a 0-1 decision variable, indicating whether the k-th vehicle in the type type serves flight i and flight j in sequence; is the transfer distance of type type vehicle from flight i to flight j; The constraints include: Formula (3) to (8) are common constraints that all vehicle types must follow; Formula (3) is the flight assignment uniqueness constraint, which means that each type of service flight is accepted only once; Formulas (4) to (6) are vehicle service uniqueness constraints, which means that ground support vehicles of different models will not repeatedly serve a flight; Formula (7) is the vehicle quantity constraint, which means that the number of vehicles of different models used does not exceed the number of available vehicles; Formula (8) is the vehicle service start time constraint, which means that the start time of a certain vehicle type service accepted by flight i is within its start time window; Formula (9)-Formula (10) are specific constraints for the clean water truck. Formula (9) indicates that the clean water truck is in an empty state when it leaves the parking lot. Formula (10) indicates that the remaining clean water volume of the clean water truck during operation cannot exceed the maximum volume of the water tank, nor can it be less than 0. Formula (11) to (13) are the joint operation constraints of the conveyor belt vehicle and the baggage tractor. Formula (11) indicates that the first conveyor belt vehicle used for service must arrive before the first baggage tractor. Formula (12) indicates that baggage service can only be started when the conveyor belt vehicle and the baggage tractor arrive at the service point at the same time. Formula (13) indicates that the conveyor belt vehicle can only end its service after all baggage tractors serving a certain flight have finished their service. Among them, H is the flight set; H type The flight set that needs to be served by vehicles of type type; N type is a set of vehicles of type type; Δ + (i) is the set of arcs starting from node i; Δ - (j) is the set of arcs returning to node j; E is the set of arcs; is the transfer distance of type type vehicle from flight i to flight j; MAX type The total number of vehicles of type type; is the time when vehicle k of type type arrives at the service point of flight i; ET i type The earliest start time of the service of type type vehicle for flight i; LT i type The latest start time of service for type type vehicle for flight i; The time when vehicle k of type type starts serving flight i; The time when vehicle k of type type finishes serving flight i; is the remaining clean water capacity of clean water truck k after serving flight i; cap is the maximum clean water capacity of the clean water truck.

3. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The adaptive NSGA-II algorithm integrated with local search adopted in step S2 solves the single-type ground support vehicle scheduling model. The specific steps are as follows: Step (1): Generate the initial parent population; Step (2): Calculate the crowding degree of the parent population; the crowding degree reflects the similarity between individual f and other individuals. The greater the crowding degree, the smaller the similarity. Therefore, when performing selection operations on the same frontier, individuals with greater crowding degrees are more likely to be selected to maintain the diversity of the population. Step (3): Initialize the offspring population; the selection operation adopts the binary tournament principle, randomly selects two individuals from the parent population, and selects the individual with the smaller non-dominated level; if the non-dominated levels are the same, selects the individual with the larger crowding degree; Step (4): Execute adaptive crossover, mutation and local search operations; the crossover operation uses the adaptive crossover operator, and the mutation operation uses the adaptive mutation operator; the core of the local search operation comes from the neighborhood search algorithm, which removes the flights that need to be adjusted from the individual through the destruction operator, and then reinserts these flights into the vehicle path according to the repair operator to obtain a new vehicle path; there are two types of destruction operators used, namely the worst removal destruction operator and the random destruction removal operator; there are two types of repair operators used, namely the minimum cost insertion repair operator and the maximum regret value insertion repair operator; Step (5): merge the parent population and the child population, and perform non-dominated sorting, crowding calculation and offset calculation on the merged population; Step (6): Apply elite selection strategy to the merged population; Sort the parent population in the order of non-dominance level, crowding degree and deviation degree, and select a specified number of individuals as the offspring population; Step (7): Update the number of iterations, repeat steps (3) to (6) until the maximum number of iterations is reached, return to find the final population and stop the program.

4. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The specific steps of the elite selection strategy in step S2 are as follows: Step (1): Sort the individuals according to the non-dominated level from small to large, starting from the non-dominated level 1, and put the individuals in the entire frontier into the next generation parent population, followed by the individuals of the next non-dominated level, until all the individuals of a certain level cannot be put into the next generation parent population, then go to the next step; Step (2): Arrange the individuals of this level in descending order of crowding, and place the individuals with higher crowding degrees into the next generation parent population in turn, until all individuals with the same crowding degree cannot be placed into the next generation parent population, and then proceed to the next step; Step (3): Sort the remaining individuals with the same crowding degree from small to large according to the degree of deviation, and prioritize individuals with small deviations into the next generation parent population until the next generation parent population is filled.

5. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The specific steps of step S3 are as follows: Step (1): Standardize the objective function value; for the dual-objective problem, use the heuristic algorithm to obtain the Pareto optimal solution set of the problem, and number the individual solutions in the obtained Pareto optimal solution set from top to bottom as 1, 2, ..., f, ..., g; divide the objective function values ​​Z1(f) and Z2(f) of the number of vehicles used and the driving distance of each individual solution by the maximum value of the number of vehicles used and the driving distance in the optimal solution set. and Complete standardization; Step (2): Calculate the weight of the combination of each objective function value of each individual and the weight of each objective; The calculation formula for the combined weight of each objective function is as follows: Where C represents the number of combinations; The C-OWA operator is a weighted average method for processing continuous data. The C-OWA operator is used to calculate the weight of each objective function. The calculation formula is as follows: Where w1' and w2' represent the absolute weights of the two objective functions respectively; w1 and w2 represent the relative weights of the two objective functions; Step (3): multiply the standardized values ​​of the objective functions of all individuals by their corresponding objective function values ​​to obtain the weighted objective function values ​​and generate a weighting matrix; The calculation formula of the weighted objective function value is as follows: Where Y1(f) and Y2(f) represent the weighted objective function values ​​of the number of vehicles used and the total distance traveled, respectively; Step (4): Find the maximum and minimum values ​​of each weighted objective function value, and then calculate the distance between each individual and the optimal level and the worst level. The larger the distance, the closer it is to the optimal level of optimization, and the corresponding vehicle model scheduling is the optimal solution. The calculation formula of the distance between the best level and the worst level is as follows:

6. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The specific steps of the taboo search algorithm in step S4 are as follows: Step (1): Based on the optimal solution for the dispatch of a single-model ground support vehicle, the chromosomes of the optimal dispatch solutions of the three models are merged into a matrix-coded chromosome, and the generated chromosome is corrected by a destructive repair method to obtain a feasible initial solution for the joint dispatch of multiple-model ground support vehicles, clear the taboo table and set the taboo length; Step (2): For the current local optimal solution X * Perform iterative learning, obtain the overall neighborhood movement set and the overall neighborhood candidate solution set according to the neighborhood generation strategy, and calculate the corresponding rating function value of each candidate solution; Step (3): Select the candidate solution X in the neighborhood with the smallest evaluation function value min , if Z(X min )<Z(X * ), then X min Become the new local optimal solution X * , and its corresponding neighborhood movement is included in the taboo table; then compared with the global optimal solution X best For comparison, if Z(X min )<Z(X best ), then X min becomes the new global optimal solution; if the neighborhood move that generates the global optimal solution or the local optimal solution is in the taboo table, then the neighborhood move is unbanned; after the comparison is completed, the taboo table is updated to determine whether the current number of iterations has reached the maximum number of iterations. If not, return to step (2); if so, go to step (4); Step (4): Stop searching after reaching the maximum number of iterations and output the current global optimal solution X best , the solution is a multi-type ground support vehicle scheduling solution for airports.

7. The method for joint dispatching of multi-type ground support vehicles for large airports according to claim 1 is characterized in that: The steps of generating the neighborhood are as follows: Step (1): First, perform neighborhood generation operation on the path of the water truck in the continuous operation mode, and optimize its path separately; because the path planning of vehicles in the continuous operation mode is not affected by the coordination constraint, their optimized paths are determined first; the corresponding chromosome X wv Neighborhood generation strategy using the "break and repair" concept; Step (2): Generate separate neighborhood move sets for the baggage tractor and conveyor truck, and combine them with the optimal neighborhood move operation of the water truck to obtain the overall neighborhood move set; the "destroy and repair" neighborhood generation strategy is also used for the baggage tractor and conveyor truck.

8. A multi-type ground support vehicle joint dispatching system for large airports, characterized by: include: The model building module is used to establish a joint dispatch model for ground support vehicles of multiple models. The model takes the total number of ground support vehicles in use and the total distance of operation as the optimization objects, combines the unique operation characteristics of different models, and takes into account that some support links require different models to complete jointly. A multi-objective optimization model for the joint dispatch of ground support vehicles of multiple models is established; The solution module is used to design an adaptive NSGA-Ⅱ algorithm integrating local search to solve the single-model ground support vehicle scheduling model, and to design a taboo search algorithm for joint scheduling to obtain a solution for multi-model joint scheduling.

Images