A multi-level and multi-mode dual-objective optimization method for route networks

Abstract

The present invention discloses a multi-level and multi-mode dual-objective optimization method for route networks, which belongs to the field of aviation technology. First, a gravity model is used to predict the passenger flow of newly opened routes. Secondly, the route network cost and the aviation accessibility of branch cities are comprehensively considered, and a multi-level and multi-mode dual-objective optimization design model for trunk and branch route networks is constructed, which solves the technical problem of integrating the general navigable short-distance transport route network into the existing trunk and branch route network, and effectively improving the service efficiency and operation efficiency of the route network, ensuring that while reducing operating costs, the aviation connectivity of remote areas is improved, reducing decision-making risks, and being able to achieve more accurate resource allocation, avoid resource waste, and ensure the rationality and effectiveness of network layout. While optimizing the overall efficiency of the network, the present invention can also take into account the aviation connection needs of branch cities, achieving a balance between global perspective and local optimization.

Description

Technical Field

[0001] The invention belongs to the field of aviation technology, and in particular relates to a multi-level and multi-mode route network dual-objective optimization method. Background Art

[0002] At present, in the traditional passenger route network, the problems of unbalanced route ratios and low airport connectivity are particularly prominent. The travel time between third- and fourth-tier cities is about 2.5 to 3.0 times that of trunk cities, and residents cannot enjoy the same aviation services as residents of trunk cities.

[0003] The general aviation short-distance transport network is like a capillary network, while the trunk and branch route network is like arteries and veins. How to optimize the route network design to ensure that the trunk and branch route network is fully connected with the general aviation short-distance transport network to form network effects, scale effects and flow effects, thereby improving the air accessibility of branch cities and meeting the air travel needs of people in remote areas, is of great practical significance.

[0004] Deficiencies of existing technology:

[0005] 1. There are many studies on route network planning and hub location selection, but the models are usually built based on operational efficiency and economic benefits, ignoring the aviation accessibility of feeder cities, resulting in large regional differences and failure to meet the aviation travel needs of people in feeder cities;

[0006] 2. General aviation short-distance transportation is becoming more and more important. Current research mainly focuses on the development status of general aviation short-distance transportation, general airport site selection and passenger demand forecasting. Research on the construction of general aviation short-distance transportation network lacks theoretical basis and corresponding quantitative analysis. Summary of the invention

[0007] The purpose of the present invention is to provide a multi-level and multi-mode route network dual-objective optimization method, which solves the technical problem of integrating the general navigable short-distance transport route network into the existing trunk and branch route network and effectively improving the service effectiveness and operation efficiency of the route network.

[0008] To achieve the above object, the present invention adopts the following technical solution:

[0009] A multi-level and multi-mode route network dual-objective optimization method includes the following steps:

[0010] Step 1: Establish a distributed server cluster, and establish a data processing module, a gravity model calculation module, an optimization design module, an algorithm solution module, a data storage management module and a user interface module in the distributed server cluster;

[0011] The data processing module collects flight data and pre-processes the flight data;

[0012] Step 2: The gravity model calculation module uses the gravity model to calculate and predict the passenger flow of the new route, and transmits the prediction results to the optimization design module;

[0013] Step 3: The optimization design module constructs a multi-level and multi-mode dual-objective optimization design model for the route network, which includes setting the objective function, which includes minimizing the network cost and maximizing the aviation accessibility of branch cities; configuring the model parameters and performing linearization processing; and establishing optimization constraints.

[0014] Step 4: The algorithm solving module uses the NSGA-II algorithm to solve the dual-objective optimization design model of the route network and obtain the Pareto frontier solution according to the set model parameters and constraints;

[0015] Step 5: The storage management module stores the optimization results and analysis data in the database, and the user interface module provides a visual display interface for retrieving and displaying the optimization results and analysis data in the database.

[0016] Preferably, when executing step 2, the following steps are specifically included:

[0017] Step 2-1: Forecast the passenger flow of the OD pairs, including collecting data on key influencing factors of the OD pairs, including per capita GDP, population and distance;

[0018] Step 2-2: Build a gravity model to predict passenger demand q ij , as shown in the following formula:

[0019]

[0020] Among them, q ij is the passenger demand from city i to city j; E i With E j are the per capita GDP of city i and city j respectively; P i With P j are the populations of city i and city j respectively; d ij is the flight distance from city i to city j; K is the gravity coefficient; a, b and c are the elasticity coefficient of per capita GDP, population elasticity coefficient and distance damping coefficient respectively;

[0021] Step 2-3: Use the regression equation to solve the parameters of the gravity model and obtain the route demand function, as shown below:

[0022]

[0023] Step 2-4: Substitute the passenger demand q obtained in step 2-3 into ij Transfer to the optimization design module.

[0024] Preferably, when executing step 3, the following steps are specifically included:

[0025] Step 3-1: Setting the objective function specifically includes the objective function f1 of route network cost and the objective function f2 of branch city aviation accessibility;

[0026] The objective function f1 of the route network cost is shown in the following formula:

[0027] max f1=f 直飞 +f 中转 ;

[0028] Among them, f 直飞 is the objective function of the transportation cost of direct passengers, f 中转 is the objective function of the transportation cost of transfer passengers;

[0029]

[0030]

[0031] Where N represents the set of general airports; α represents the unit transportation cost discount factor for general aviation short-distance transportation; β represents the unit transportation cost discount factor when transferring through hub airports; c represents the passenger-kilometer cost of the flight; q ij represents the passenger demand from airport i to airport j, in person-years -1 ,q ij =q ji ;d ij Represents the flight distance from airport i to airport j, in km, d ij =d ji ;d il Represents the flight distance from airport i to hub airport l, in km, d il =d li ;d jl Represents the flight distance from airport j to hub airport l, in km, d jl =d lj ; is a continuous decision variable, representing the passenger volume from airport i to airport j via regional hub airport l; is a continuous decision variable, representing the passenger volume between airport i and airport j via the central hub airport l; x ij is a 0-1 decision variable, which is equal to 1 when airport i and airport j have a short-haul transport route, and 0 otherwise;

[0032] When i and j are the same, if x i =1, select airport i to build a regional hub airport; y ijis a 0-1 decision variable, which is equal to 1 when a public transport route is opened from airport i to airport j, and 0 otherwise. ii =1, select airport i to build a central hub airport; H represents the set of regional hub airports; CH represents the set of central hub airports;

[0033] The objective function f of the transportation cost for direct passengers is 直飞 Perform linearization processing;

[0034] The objective function f2 of branch city aviation accessibility is as follows:

[0035]

[0036]

[0037] Among them, C1, C2, and C3 are used to unify the dimensions and adjust the weights;

[0038] Step 3-2: Construct constraints, including hub airport quantity constraints, flight distance limit constraints, airport allocation constraints, flow distribution constraints, objective function linearization constraints, flow balance constraints and variable range constraints;

[0039] The hub airport quantity constraint is used to indicate the number of central hub airports and regional hub airports to be built;

[0040] The distance limit constraint is used to ensure that the distance of general short-distance transport routes does not exceed 500 kilometers, and the distance of public transport routes is not less than 500 kilometers;

[0041] Airport allocation constraints are used to ensure that general airports must be allocated to regional hub airports, that regional hub airports must be allocated to central hub airports, and that the central hub is fully connected to the network;

[0042] Traffic distribution constraints are used to ensure that traffic is only distributed to established connections;

[0043] Objective function linearization constraints are used to linearize the objective function;

[0044] Variable range constraints are used to define the value range of decision variables.

[0045] Preferably, when the NSGA-II algorithm is used in step 4, the chromosome is represented by two arrays, namely a hub array and an allocation array, and the length of the hub array and the length of the allocation array are both equal to the number of airport nodes.

[0046] Preferably, the hub array consists of three values ​​0, 1 and 2, where 0 represents a general airport, 1 represents a regional hub, and 2 represents a central hub;

[0047] The allocation array represents the regional hubs allocated by the general airport and the central hubs allocated by the regional hubs;

[0048] The distribution gene of the central hub is itself.

[0049] Preferably, the flight data set includes passenger demand, route passenger kilometer cost, route information, flight segment distance, airport data and flight frequency.

[0050] The multi-level and multi-mode dual-objective optimization method for route network described in the present invention solves the technical problem of integrating the general navigable short-distance transport route network into the existing trunk and branch route network, and effectively improving the service performance and operation efficiency of the route network. The present invention simultaneously considers the minimization of network costs and the maximization of the aviation accessibility of branch cities, ensuring that the aviation connectivity of remote areas is improved while reducing operating costs, and helping to achieve a balanced aviation network layout. Through gravity models and regression analysis, the scheme can accurately predict the passenger flow of new routes, thereby providing reliable data support for optimization design and reducing decision-making risks. The NSGA-II algorithm is used to solve the dual-objective optimization model, which can effectively handle complex multi-objective optimization problems, obtain Pareto frontier solutions, and help users make the best choice among multiple alternative solutions. Through the linearization of route network costs and the introduction of multiple constraints, the present invention can achieve more accurate resource allocation, avoid resource waste, and ensure the rationality and effectiveness of network layout. While optimizing the overall efficiency of the network, the present invention can also take into account the aviation connection needs of branch cities, achieving a balance between global perspective and local optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is the main flow chart of the present invention;

[0052] Figure 2 is an example of a multi-level multi-mode route network of the present invention;

[0053] Figure 3 is a flow chart of the NSGA-II algorithm of the present invention;

[0054] Figure 4 It is a schematic diagram of the chromosome structure in the NSGA-II algorithm of the present invention;

[0055] Figure 5 Schematic diagram of the Pareto frontier solution obtained by 200 iterations of the present invention;

[0056] Figure 6 It is a schematic diagram of the trunk-branch route network of the present invention;

[0057] Figure 7 It is a schematic diagram of the effect of improving the aviation accessibility of some branch cities according to the present invention. DETAILED DESCRIPTION

[0058] Parameter Description:

[0059] S represents the set of all airports;

[0060] N represents the set of general airports;

[0061] H represents the collection of regional hub airports;

[0062] CH represents the collection of central hub airports;

[0063] p1 represents the number of central hub airports to be built;

[0064] p2 represents the number of regional hub airports to be built;

[0065] α represents the unit transportation cost discount factor for short-distance transportation;

[0066] β represents the unit transportation cost discount factor when transferring through a hub airport;

[0067] c represents the passenger-kilometer cost of the flight;

[0068] Ta represents the aviation accessibility of airport a;

[0069] q ij represents the passenger demand from airport i to airport j, in person-years -1 ,q ij =q ji ;

[0070] d ij Represents the flight distance from airport i to airport j, in km, d ij =d ji ;

[0071] d il Represents the flight distance from airport i to hub airport l, in km, d il =d li ;

[0072] d jl Represents the flight distance from airport j to hub airport l, in km, d jl =d lj ;

[0073] n ij represents the flight frequency from airport i to airport j (times per year -1 ), n ji =n ij ;

[0074] is a continuous decision variable, representing the passenger volume from airport i to airport j via regional hub airport l;

[0075] is a continuous decision variable, representing the passenger volume between airport i and airport j via the central hub airport l;

[0076] x ij is a 0-1 decision variable, which is equal to 1 when airport i and airport j have a short-haul transport route, otherwise it is 0. i =1, select airport i to build a regional hub airport;

[0077] y ij is a 0-1 decision variable, which is equal to 1 when a public transport route is opened from airport i to airport j, and 0 otherwise. i =1, select airport i to build a central hub airport.

[0078] x ij =1 means i and j are connected (i≠j); when i and j are the same, generally x ij =0; This describes a special case and is represented in the model by the constraint on the number of hub airports.

[0079] Depend on Figure 1-Figure 7 A multi-level and multi-mode route network dual-objective optimization method is shown, comprising the following steps:

[0080] Step 1: Establish a distributed server cluster, and establish a data processing module, a gravity model calculation module, an optimization design module, an algorithm solution module, a data storage management module and a user interface module in the distributed server cluster;

[0081] The data processing module collects flight data and pre-processes the flight data;

[0082] The flight data set includes passenger flow, route cost, route information, airport data and flight frequency.

[0083] In this embodiment, Figure 2 An example of a multi-level multimodal route network consisting of 8 general airports, 6 feeder airports and 4 trunk airports is shown. In this network, trunk airports serve as central hubs to connect feeder airports, while feeder airports serve as regional hubs to connect general aviation short-distance transportation with public transportation. The first-level transportation network adopts a full network structure and uses the transportation mode of public transportation; the second-level adopts a star network structure and uses the transportation mode of public transportation; the third-level adopts a star network structure and uses the transportation mode of general aviation short-distance transportation.

[0084] The traditional route network optimization model only has two types of nodes: hubs and non-hubs. It does not comprehensively consider the route network cost and the aviation accessibility of feeder cities, and does not perform route network optimization design. To address this problem, this paper constructs a route network optimization design model. The model's assumptions are as follows:

[0085] 1) The number of central hubs and regional hubs is known.

[0086] 2) Hubs and routes have sufficient capacity to handle all transportation needs.

[0087] 3) Regional hubs can be assigned to multiple central hubs, and general airports can be assigned to multiple regional hubs.

[0088] 4) Direct connection between any two central hubs; direct connection is not allowed between regional hubs, but is allowed between general airports.

[0089] 5) The maximum number of transfers a passenger can make is once.

[0090] 6) General aviation short-distance transport uses small aircraft with a small number of passengers. It is difficult to support the operation of general aviation companies by relying on market-based ticket revenue. The routes need government subsidies in the initial operation stage. In this article, the unit transportation cost discount factor of general aviation short-distance transport is set at 0.7.

[0091] Step 2: The gravity model calculation module uses the gravity model to calculate and predict the passenger flow of the new route and transmits the prediction results to the optimization design module, which specifically includes the following steps:

[0092] Step 2-1: Forecast the passenger flow of the OD pairs, including collecting data on key influencing factors of the OD pairs, including per capita GDP, population and distance;

[0093] Step 2-2: Build a gravity model to predict passenger demand q ij , as shown in the following formula:

[0094]

[0095] Among them, q ij is the passenger demand from city i to city j; E i With E j are the per capita GDP of city i and city j respectively; P i With P j are the populations of city i and city j respectively; d ij is the flight distance from city i to city j; K is the gravity coefficient; a, b and c are the elasticity coefficient of per capita GDP, population elasticity coefficient and distance damping coefficient respectively;

[0096] Step 2-3: Use the regression equation to solve the parameters of the gravity model and obtain the route demand function, as shown below:

[0097]

[0098] Step 2-4: Substitute the passenger demand q obtained in step 2-3 into ij Transfer to the optimization design module.

[0099] Step 3: The optimization design module constructs a multi-level and multi-mode dual-objective optimization design model for the route network, which specifically includes setting the objective function, which includes minimizing the network cost and maximizing the aviation accessibility of branch cities; configuring the model parameters and performing linearization processing; and establishing optimization constraints, which specifically includes the following steps:

[0100] Step 3-1: Setting the objective function specifically includes the objective function f1 of route network cost and the objective function f2 of branch city aviation accessibility;

[0101] The objective function f1 of the route network cost is shown in the following formula:

[0102] max f1=f 直飞 +f 中转 ;

[0103] Among them, f 直飞 is the objective function of the transportation cost of direct passengers, f 中转 is the objective function of the transportation cost of transfer passengers;

[0104]

[0105] The objective function f of the transportation cost for direct passengers is 直飞 Perform linearization processing;

[0106] The objective function f2 of branch city aviation accessibility is as follows:

[0107]

[0108] Among them, C1, C2, and C3 are used to unify the dimensions and adjust the weights. i represents the aviation accessibility of airport i;

[0109] Step 3-2: Construct constraints:

[0110] The hub airport quantity constraint is used to indicate the number of central hub airports and regional hub airports to be built:

[0111]

[0112] Among them, if x i=1, select airport i to build a regional hub airport; if y i =1, select airport i to build a central hub airport;

[0113] Distance Limit Constraints:

[0114]

[0115] This constraint ensures that the distance of general short-haul transport routes does not exceed 500 kilometers.

[0116]

[0117] This constraint ensures that the distance of public transport routes is no less than 500 kilometers.

[0118] Airport allocation constraints:

[0119]

[0120] This constraint ensures that general airports must be allocated to regional hub airports.

[0121]

[0122] This constraint ensures that regional hub airports must be assigned to central hub airports.

[0123]

[0124] This constraint ensures full network connectivity of the central hub.

[0125] Traffic distribution constraints are used to ensure that traffic is only distributed to established connections:

[0126]

[0127] Objective function linearization constraints are used to linearize the objective function:

[0128]

[0129] in, and The objective function f is the transportation cost of direct passengers. 直飞 The symbols introduced during linearization are: M represents an arbitrarily large positive number;

[0130] The flow balance constraints are as follows:

[0131]

[0132] Variable range constraints are used to define the value range of decision variables:

[0133]

[0134] Step 4: The algorithm solving module uses the NSGA-II algorithm to solve the dual-objective optimization design model of the route network and obtain the Pareto frontier solution according to the set model parameters and constraints;

[0135] The NSGA-II algorithm first initializes the population, then performs a fast non-dominated sorting method on the individuals in the population, and performs crossover and mutation operations on the individuals while performing selection operations based on the sorting of non-dominated and crowdedness. The NSGA-II algorithm is an existing technology that can reduce algorithm complexity, accelerate algorithm convergence performance and ensure algorithm diversity, and is widely used in solving multi-objective hub location problems.

[0136] When the NSGA-II algorithm is used in step 4, the chromosome is represented by two arrays, namely the hub array and the allocation array. The length of the hub array and the length of the allocation array are both equal to the number of airport nodes.

[0137] The hub array consists of three values: 0, 1 and 2, where 0 represents a general airport, 1 represents a regional hub, and 2 represents a central hub;

[0138] The allocation array represents the allocation relationship between general airports and regional hubs and between regional hubs and central hubs. Specifically, if airport node i is allocated to hub airport k, then the i-th gene value on the allocation array is k.

[0139] Since the central hubs are fully connected, the distribution gene of the central hub is itself.

[0140] Step 5: The storage management module stores the optimization results and analysis data in the database, and the user interface module provides a visual display interface for retrieving and displaying the optimization results and analysis data in the database.

[0141] In this embodiment, 20 existing and planned general airports in Province A, 14 feeder airports in Province A and 5 trunk airports in Province A and its neighboring provinces are selected. Through the route network optimization design and in accordance with the "general aviation + transportation" organizational model, a short-distance transport route network structure of "hub-and-spoke distribution within the league and city, and two centripetal fans in the autonomous region" is formed. The airport nodes are selected as shown in Table 1:

[0142]

[0143] Table 1

[0144] Parameter settings:

[0145] The aviation demand data for city pairs with opened routes comes from the Internet, and the aviation demand data for city pairs without opened routes are predicted through the gravity model.

[0146] According to relevant research on airline transportation costs, this embodiment sets the cost per passenger kilometer to 0.45 yuan; when transferring through a hub airport, the discount on the unit transportation cost is set to 0.6.

[0147] After experiments, the following parameter settings in the NSGA-II algorithm can produce better results: population size N = 500, maximum number of generations G = 200, crossover probability P c =0.8, the mutation probability is P m =0.3.

[0148] Result analysis:

[0149] The Pareto frontier solution obtained after 200 iterations is as follows: Figure 5 As shown. It can be seen that as the cost of the route network increases, the aviation accessibility of branch cities is also gradually increasing. The final set of Pareto solutions is shown in Table 2:

[0150]

[0151] Table 2

[0152] Figure 6 The trunk-branch route network of Pareto solution 3 is given. The trunk-branch route network of Pareto solution 3 is compared with the traditional route network that does not consider short-distance transportation, as shown in Table 3:

[0153]

[0154] Table 3

[0155] Verification summary:

[0156] 1. Compared with the traditional trunk and branch route network, the aviation accessibility of branch cities in the route network of the present invention has increased by 22.81%. By integrating the general navigable short-distance transport route network into the trunk and branch route network, the aviation accessibility of branch cities has been greatly improved.

[0157] 2. The total cost of the route network increased by only 7.02%. The trunk and branch route network remained relatively stable in terms of cost control and did not cause a significant economic burden.

[0158] 3. The number of travelers on the trunk-branch route network was 4,230,145, an increase of 10.64% compared to 3,823,341 on the traditional network. This result shows that the designed trunk-branch route network effectively meets the air travel needs of residents in third- and fourth-tier cities and stimulates the potential of the air transport market.

[0159] 4. In terms of the number of transfers and transfer rate, the number of transfers on the trunk and branch route network increased by 59.94%, and the transfer rate increased by 48.59%, providing passengers with more flight options and enabling more passengers to achieve efficient and convenient travel through transfers.

[0160] Figure 7 The results show the improvement of aviation accessibility of some branch cities. Among them, the growth rate of aviation accessibility of DSN, BAV, HLD, HLH and CIF airports exceeded 25%, indicating that these branch cities have effectively promoted the route interconnection and airport interconnection between trunk, branch and short-distance transportation in the trunk-branch route network, and played an important role as a regional hub.

[0161] The multi-level and multi-mode dual-objective optimization method for route network described in the present invention solves the technical problem of integrating the general navigable short-distance transport route network into the existing trunk and branch route network, and effectively improving the service performance and operation efficiency of the route network. The present invention simultaneously considers the minimization of network costs and the maximization of the aviation accessibility of branch cities, ensuring that the aviation connectivity of remote areas is improved while reducing operating costs, and helping to achieve a balanced aviation network layout. Through gravity models and regression analysis, the scheme can accurately predict the passenger flow of new routes, thereby providing reliable data support for optimization design and reducing decision-making risks. The NSGA-II algorithm is used to solve the dual-objective optimization model, which can effectively handle complex multi-objective optimization problems, obtain Pareto frontier solutions, and help users make the best choice among multiple alternative solutions. Through the linearization of route network costs and the introduction of multiple constraints, the present invention can achieve more accurate resource allocation, avoid resource waste, and ensure the rationality and effectiveness of network layout. While optimizing the overall efficiency of the network, the present invention can also take into account the aviation connection needs of branch cities, achieving a balance between global perspective and local optimization.

Claims

1. A multi-level and multi-mode route network dual-objective optimization method, characterized by: The steps include: Step 1: Establish a distributed server cluster, and establish a data processing module, a gravity model calculation module, an optimization design module, an algorithm solution module, a data storage management module and a user interface module in the distributed server cluster; The data processing module collects flight data and pre-processes the flight data; Step 2: The gravity model calculation module uses the gravity model to calculate and predict the passenger flow of the new route, and transmits the prediction results to the optimization design module; Step 3: The optimization design module constructs a multi-level and multi-mode dual-objective optimization design model for the route network, which includes setting the objective function, which includes minimizing the network cost and maximizing the aviation accessibility of branch cities; configuring the model parameters and performing linearization processing; and establishing optimization constraints. When executing step 3, the specific steps include: Step 3-1: Setting the objective function specifically includes the objective function f1 of route network cost and the objective function f2 of branch city aviation accessibility; The objective function f1 of the route network cost is shown in the following formula: my f1=f 直飞 +f 中转 ; Among them, f 直飞 is the objective function of the transportation cost of direct passengers, f 中转 is the objective function of the transportation cost of transfer passengers; Where N represents the set of general airports; α represents the unit transportation cost discount factor for general aviation short-distance transportation; β represents the unit transportation cost discount factor when transferring through hub airports; c represents the passenger-kilometer cost of the flight; q ij represents the passenger demand from airport i to airport j, in person-years -1 ,q ij =q ji ;d ij Represents the flight distance from airport i to airport j, in km, d ij =d ji ;d il Represents the flight distance from airport i to hub airport l, in km, d il =d li ;d jl Represents the flight distance from airport j to hub airport l, in km, d jl =d lj ; is a continuous decision variable, representing the passenger volume from airport i to airport j via regional hub airport l; is a continuous decision variable, representing the passenger volume between airport i and airport j via the central hub airport l; x ij is a 0-1 decision variable, which is equal to 1 when airport i and airport j have a short-haul transport route, and 0 otherwise; When i and j are the same, if x ii =1, select airport i to build a regional hub airport; y ij is a 0-1 decision variable, which is equal to 1 when a public transport route is opened from airport i to airport j, and 0 otherwise. ii =1, select airport i to build a central hub airport; H represents the set of regional hub airports; CH represents the set of central hub airports; The objective function f of the transportation cost for direct passengers is 直飞 Perform linearization processing; The objective function f2 of branch city aviation accessibility is as follows: Among them, C1, C2, and C3 are used to unify the dimensions and adjust the weights; Step 3-2: Construct constraints, including hub airport quantity constraints, flight distance limit constraints, airport allocation constraints, flow distribution constraints, objective function linearization constraints, flow balance constraints and variable range constraints; The hub airport quantity constraint is used to indicate the number of central hub airports and regional hub airports to be built; The distance limit constraint is used to ensure that the distance of general short-distance transport routes does not exceed 500 kilometers, and the distance of public transport routes is not less than 500 kilometers; Airport allocation constraints are used to ensure that general airports must be allocated to regional hub airports, that regional hub airports must be allocated to central hub airports, and that the central hub is fully connected to the network; Traffic distribution constraints are used to ensure that traffic is only distributed to established connections; Objective function linearization constraints are used to linearize the objective function; Variable range constraints are used to define the value range of decision variables; Step 4: The algorithm solving module uses the NSGA-II algorithm to solve the dual-objective optimization design model of the route network and obtain the Pareto frontier solution according to the set model parameters and constraints; Step 5: The data storage management module stores the optimization results and analysis data in the database, and the user interface module provides a visual display interface for retrieving and displaying the optimization results and analysis data in the database.

2. The multi-level and multi-mode route network dual-objective optimization method according to claim 1, characterized in that: When executing step 2, the specific steps include: Step 2-1: Forecast the passenger flow of the OD pairs, including collecting data on key influencing factors of the OD pairs, including per capita GDP, population and distance; Step 2-2: Build a gravity model to predict passenger demand q uv , as shown in the following formula: Among them, q uv is the passenger demand from city u to city v; E u With E v are the per capita GDP of city u and city v respectively; P u With P v are the populations of city u and city v respectively; d uv is the flight distance from city u to city v; K is the gravity coefficient; a, b and c are the elasticity coefficient of per capita GDP, population elasticity coefficient and distance damping coefficient respectively; Step 2-3: Use the regression equation to solve the parameters of the gravity model and obtain the route demand function, as shown below: Step 2-4: Substitute the passenger demand q obtained in step 2-3 into uv Transfer to the optimization design module.

3. The multi-level and multi-mode route network dual-objective optimization method according to claim 1, characterized in that: When the NSGA-II algorithm is used in step 4, the chromosome is represented by two arrays, namely the hub array and the allocation array. The length of the hub array and the length of the allocation array are both equal to the number of airport nodes.

4. The multi-level and multi-mode route network dual-objective optimization method according to claim 3, characterized in that: The hub array consists of three values: 0, 1 and 2, where 0 represents a general airport, 1 represents a regional hub, and 2 represents a central hub; The allocation array represents the regional hubs allocated by the general airport and the central hubs allocated by the regional hubs; The distribution gene of the central hub is itself.

5. The multi-level and multi-mode route network dual-objective optimization method according to claim 4, characterized in that: The flight data includes passenger demand, route passenger kilometer cost, route information, flight segment distance, airport data and flight frequency.

Images