Main-branch multi-layer transportation network optimization method for general aviation demand

Abstract

The invention discloses a general aviation demand-oriented trunk and branch multi-layer transportation network optimization method, relates to the technical field of aviation hub site selection, and aims to solve the problems of accurate prediction of general aviation short-distance transportation demands and connection of general aviation and trunk and branch aviation networks. The method comprises the following steps: firstly, establishing an airline passenger transport demand influence factor index system, and extracting a similar airline reference sample set; secondly, considering characteristics of different types of airports and airlines, and designing a gravity model based on similarity to carry out airline demand prediction; then, by taking minimization of transportation cost and hub construction cost, minimization of longest transportation time and maximization of general aviation short-distance transportation demand coverage rate as targets, constructing a single-allocation allowed direct connection shaft spoke type network hub site selection model; and finally, designing an NSGA-III algorithm to solve the multi-objective optimization model. According to the method, economic benefits, transportation efficiency and demand coverage can be comprehensively considered, a hub layout scheme is obtained, and reference is provided for decision makers of governments, airports and airlines.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of aviation hub site selection, and particularly relates to a multi-layer transport network optimization method for general aviation demand. BACKGROUND

[0002] Under the background of continuous upgrading of aviation transport networks, building a multi-level and fully connected air traffic system has become a development trend in the industry. As an important extension of regional route networks, general aviation short-haul transport connects resource-based areas through branch nodes and has strategic value in promoting three-dimensional traffic coordination. However, there are two major bottlenecks at present: first, the lack of general aviation short-haul transport operation data makes it difficult to predict the demand of small and medium-sized airports, and traditional models relying on historical data are not applicable; second, the existing trunk-branch network model cannot adapt to the connection demand, and the hub site selection research is mostly limited to a single objective, lacking consideration of the high accessibility characteristics of short-haul transport and the connection rules of three or more network levels. In addition, the network optimization problem is essentially a mixed integer nonlinear programming (NP difficult), and the efficiency of traditional solvers decreases sharply when the number of airports exceeds 30, making it difficult to meet actual needs. To overcome the above challenges, breakthroughs are needed in three aspects: first, by capturing the characteristics of different types of routes, the demand prediction problem under the condition of lack of historical data is solved; second, a multi-objective optimization model is built, with the minimum cost, transportation time compression and maximum demand coverage rate as the objectives, allowing ordinary hubs to be directly connected and integrating short-haul transport constraints; finally, an efficient NSGA-III algorithm is designed to improve the solving ability in high-dimensional objective space and realize the rapid optimization of large-scale networks. SUMMARY

[0003] The application discloses a multi-layer transport network optimization method for general aviation demand, aiming to solve the problems of accurate prediction of general aviation short-haul transport demand and connection of general aviation and trunk-branch aviation network.

[0004] To achieve the above-mentioned purpose, the technical scheme provided by the application is as follows:

[0005] A multi-layer transport network optimization method for general aviation demand, characterized in that it comprises the following steps:

[0006] Step 1: Considering the airport attributes of route origin and destination points, the city attributes of airports and the route attributes itself, an index system of route passenger demand influencing factors is established from the diversity and quantifiability of indexes, and a reference sample set of similar routes is extracted;

[0007] Step 2: Considering the characteristics of different types of airports and routes, selecting reference sample routes with similar characteristics, and designing a gravity model based on similarity for route demand prediction;

[0008] Step 3: A single distribution allowed direct connection hub location model of the hub-and-spoke network is constructed, i.e. a multi-layer network optimization model of hub-and-spoke network, with the objectives of minimizing transportation cost and hub construction cost, minimizing the longest transportation time, and maximizing the coverage rate of general aviation short-haul transportation demand;

[0009] Step 4: The NSGA-III algorithm is designed to solve the proposed multi-layer network optimization model of hub-and-spoke network.

[0010] To optimize the above technical solutions, the specific measures / limitations also include:

[0011] In step 1, considering the attributes of the airport origin and destination, the attributes of the city where the airport is located, and the attributes of the route itself, the relevant factors are selected to form the demand influence factor index system as follows:

[0012] Airport-related indicators: airport throughput, number of navigable cities;

[0013] City-related indicators: city population, total GDP, per capita GDP, fiscal budget expenditure, city commercial grade, tourism income, number of A-level scenic spots, and proportion of tertiary industry;

[0014] Route-related indicators: route distance.

[0015] The concepts of cosine similarity and distance similarity are defined to calculate the similarity values of each indicator between the target route sample and the reference route sample:

[0016] The cosine similarity θ is established as follows: ij The calculation formula is as follows:

[0017]

[0018] Where, the two-dimensional vector a j represents the value of the jth indicator, and b ij represents the value of the jth indicator of the ith reference sample route.

[0019] The distance similarity ξ is established as follows: ij The calculation formula is as follows:

[0020]

[0021] The comprehensive similarity Sim(a j ,b ij ) is established as follows:

[0022]

[0023] Where, z ij represents the similarity between the jth indicator and the jth indicator vector of the ith reference sample, and its value is the product of the cosine similarity and the distance similarity, and CVj is the coefficient of variation of the jth index, which is the standard deviation of the jth index divided by the mean.

[0024] Extract the reference samples with a comprehensive similarity higher than the set threshold to form a certain number of similar route reference sample sets.

[0025] In step 2, considering the characteristics of different types of airports and routes, a gravity model based on similarity is designed for route demand prediction to solve the problem of relatively scarce data for general aviation short-haul transport routes, and the formula is as follows:

[0026]

[0027] where W ij represents the demand from region i to region j. q i , q j represents the size of region i, j, d ij represents the distance between regions i and j. k, a, b, c are model parameters to be calibrated.

[0028] In step 3, a hub-and-spoke multi-level transport network optimization model is established for general aviation demand. With the objectives of minimizing cost, minimizing maximum transport time, and maximizing demand coverage, a single distribution allowed hub-and-spoke network multi-objective optimization objective function is constructed:

[0029] The first objective is to minimize the total cost, and the specific formula is:

[0030]

[0031] where the first three terms are transportation costs, and the last two terms are hub construction costs. In the formula: N, H, CH are the sets of all airports, candidate points for hub airports, and candidate points for central hubs, respectively; w is is the OD flow from airport i to airport s; c0 is the unit transportation cost of general aviation short-haul transport; α1 is the discount factor for the unit transportation cost required for transportation between central hubs compared to general aviation short-haul transport; α2 is the discount factor for the unit transportation cost required for transportation between ordinary hubs and between ordinary hubs and central hubs compared to general aviation short-haul transport; β is the discount factor for the fixed cost required for constructing a central hub compared to ordinary hubs; d ij is the distance from airport i to airport j; F i is the fixed cost required for selecting airport i to construct an ordinary hub; x ij is a 0-1 decision variable, equal to 1 when airport i is assigned to airport j, and 0 otherwise, if x jj = 1, then select airport j to construct a hub; y ii is a 0-1 decision variable, if yii = 1, then select airport i to build a central hub, if y jl = 1 and j≠l, y jl = 1, then select airport i to build a central hub, if y ijl is an integer decision variable, which represents the demand from airport i via hub j to hub l, where at least one of hub j and hub l is a general hub; h ijl is an integer decision variable, which represents the demand from airport i via central hub j to central hub l.

[0032] The second objective is to minimize the maximum travel time between the origin and destination, which is defined as:

[0033]

[0034] where the upper formula is the travel time of the direct route via the general hub, and the lower formula is the travel time of the route via the central hub. In the formula: is a decision variable, which represents the earliest arrival time of all demands assigned to central hub k; is a decision variable, which represents the earliest arrival time of all demands assigned to hub i; λ1 is the discount factor of the required travel time between central hubs relative to the general aviation short-haul transportation; λ2 is the discount factor of the required travel time between general hubs, between general hubs and central hubs relative to the general aviation short-haul transportation; t ij is the travel time from airport i to airport j; s ij is a 0-1 decision variable, which equals to 1 when general hub i is directly connected to general hub j, and equals to 0 otherwise; u kl is a 0-1 decision variable, which equals to 1 when both airport k and airport l are selected as central hubs, and equals to 0 otherwise; M is a positive integer; X ijkl is a 0-1 decision variable, which equals to 1 when the travel path from airport i to airport j is via central hub k and central hub l, and equals to 0 otherwise.

[0035] The third objective is to maximize the coverage rate of the general aviation short-haul transportation demand, which is defined as:

[0036]

[0037] where the numerator is the demand that can be covered by the general aviation short-haul transportation service, and the denominator is the total demand of the general aviation short-haul transportation. In the formula: m ij is an input parameter, which equals to 1 when the distance between airport i and j is less than or equal to 500 km, and equals to 0 otherwise. ij

[0038] ​To adapt to the dry branch pass multi-layer network scene, the constraint conditions include hub and location allocation, central hub connection, ordinary hub direct connection constraint and flow balance constraint multiple aspects, ensure that the model adapts to the connection relationship between different levels of network and the allocation rule.

[0039] In step 4, the NSGA-III is designed to solve the model, first design the chromosome:

[0040] The chromosome is divided into hub array and allocation array two parts, define the array length as |N|, the first |CH| genes represent the central hub airport candidate set, the first |H| genes represent the hub airport candidate set, and the last |N|-|H| genes are non-hub airports. The genes of the hub array are composed of numbers "0", "1", "2", which represent non-hub, ordinary hub and central hub respectively. The genes of the allocation array represent the allocation of each airport, and the allocation of the central hub is itself.

[0041] Then design selection, crossover and mutation operators: selection operator adopts binary tournament to get offspring chromosome. The crossover operator adopts single-point crossover, when the crossover probability P c is satisfied, select two individuals for crossover operation. The mutation operator adopts exchange mutation, when the mutation probability P m is satisfied, select the offspring generated by the crossover operation for mutation operation.

[0042] Then design the direct connection strategy between ordinary hubs based on transportation cost and time: for each two ordinary hubs, respectively calculate the transportation cost and time of direct connection between ordinary hubs and transfer through central hub. If the transportation cost and time of direct connection are both less than transfer, select direct connection; if one of the transportation cost and time of direct connection is less than transfer, select direct connection or transfer with equal probability; if the transportation cost and time of direct connection are both greater than transfer, select transfer.

[0043] Finally, the non-dominated sorting operation is performed on the combined population, and the operating individual is selected.

[0044] Compared with the prior art, the beneficial effects of the present application are:

[0045] The present application aims at the deficiency in the existing dry branch pass multi-layer network optimization research, and proposes a multi-objective network optimization method for general aviation demand. The present application simultaneously considers the minimization of cost, the minimization of maximum transportation time and the maximization of demand coverage, and can balance economic benefit, transportation efficiency and demand coverage. Meanwhile, through the route demand prediction based on the similarity-based gravity model, the present application can accurately predict the general aviation short-distance transportation route demand, and provide data support for network optimization. The NSGA-III algorithm is used to solve the model, and has advantages in convergence and diversity compared with other heuristic algorithms. When planning the route network, the decision maker can reasonably determine the number and distribution of hubs according to the different emphasis of demand, comprehensively consider the geographical location, transportation demand and market potential of the hub, and construct a reliable and efficient dry branch pass multi-layer transportation network. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 is a flow chart of the method of the present application.

[0047] Figure 2 is a dry branch pass multi-layer network structure diagram.

[0048] Figure 3 is a chromosome structure diagram of the NSGA-III algorithm design.

[0049] Figure 4 is a comparison result diagram of the Pareto optimal solution set of four heuristic solving methods.

[0050] Figure 5 is a dry branch pass multi-layer network optimization result diagram. DETAILED DESCRIPTION

[0051] The above content of the present application will be further described in detail in the form of examples below, but this should not be understood as the scope of the above subject matter of the present application being limited to the following examples only, and any technology realized based on the above content of the present application belongs to the scope of the present application.

[0052] The present application proposes a dry branch pass multi-layer transportation network optimization method for general aviation demand, and the flow chart is as shown in Figure 1 , and the network diagram is as shown in Figure 2 , and the present application comprises the following steps:

[0053] (1) considering the airport attribute of the route origin and destination, the city attribute where the airport is located and the route attribute itself, establishing a route passenger demand influence factor index system from the diversity and quantifiability of the index, and extracting a similar route reference sample set.

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

[0055] 1.1, considering the airport attributes of the route origin and destination, the city attributes of the airport and the route attributes itself, select relevant factors to form the demand influence factor index system:

[0056] Airport-related indicators: airport throughput, number of navigable cities;

[0057] City-related indicators: city population, total GDP, per capita GDP, fiscal budget expenditure, city commercial grade, tourism income, number of A-level scenic spots, proportion of tertiary industry;

[0058] Route-related indicators: route distance.

[0059] 1.2, first define the concepts of cosine similarity and distance similarity to calculate the similarity values of each indicator between the target route sample and the reference route sample, the calculation formulas are respectively:

[0060] Cosine similarity θ ij :

[0061]

[0062] Where, two-dimensional vector a j represents the value of the jth indicator, b ij represents the value of the jth indicator of the ith reference sample route;

[0063] Distance similarity ξ ij :

[0064]

[0065] Then establish the comprehensive similarity Sim(a j ,b ij ) calculation formula:

[0066]

[0067] Finally, by extracting reference samples with a comprehensive similarity higher than the set threshold, a similar route reference sample set is formed.

[0068] (2) Considering the characteristics of different types of airports and routes, a gravity model based on similarity is designed to predict route demand.

[0069] In step (2), for the target sample route, select routes similar to its characteristics from the similar route reference sample set to calibrate the parameters, which can adapt to the particularity of different target sample routes, and the gravity model formula is as follows:

[0070]

[0071] Where, W ijIt represents the demand for the number of passengers departing from region i and arriving at region j. i ,q j represents the size of regions i and j, d ij represents the distance between regions i and j. k, a, b, and c are the model parameters to be calibrated.

[0072] (3) Establish a multi-layer transportation network optimization model for trunk and branch transportation to meet general aviation needs.

[0073] In step (3), with the goals of minimizing cost, minimizing the longest transportation time, and maximizing demand coverage, a multi-objective optimization objective function of a single-distribution hub-and-spoke network with direct connection is constructed:

[0074] The first goal is to minimize the total cost, and the specific formula is:

[0075]

[0076] Among them, the first three items are transportation costs, and the last two items are hub construction costs. Where: N, H, CH are the set of all airports, the set of candidate points for hub airports, and the set of candidate points for central hubs respectively; w is is the OD flow from airport i to airport s; c0 is the unit transportation cost of general aviation short-haul transportation; α1 is the discount factor of the unit transportation cost required for transportation between central hubs compared with general aviation short-haul transportation; α2 is the discount factor of the unit transportation cost required for transportation between ordinary hubs and between ordinary hubs and central hubs compared with general aviation short-haul transportation; β is the discount factor of the fixed cost required to build a central hub relative to ordinary hubs; d ij is the distance from airport i to airport j; F i The fixed cost required to build a general hub for airport i; x ij is a 0-1 decision variable, which is equal to 1 when airport i is assigned to airport j, and 0 otherwise. jj =1, then select airport j to build the hub; y ii is a 0-1 decision variable. If y ii =1, then choose airport i to build the central hub. If y jl =1 and j≠l, y jl indicates that ordinary hub j is assigned to central hub l; g ijl is an integer decision variable, representing the demand from airport i to hub l via hub j, where at least one of the hubs from hub j to hub l is a common hub; h ijl is an integer decision variable, which represents the demand for flights departing from airport i and passing through hub j to hub l.

[0077] The second goal is to minimize the maximum time between the starting and ending points of transportation. The specific formula is:

[0078]

[0079] where the upper formula is the air transport time of the route directly connected through the general hub, and the lower formula is the air transport time of the route transferred through the central hub. In the formula, t is the decision variable, representing the earliest arrival time of all demands allocated to the central hub k; is the decision variable, representing the earliest arrival time of all demands allocated to the hub i; λ1 is the discount factor of the required transport time between central hubs relative to the general air short-haul transport; λ2 is the discount factor of the required transport time between general hubs, between general hubs and central hubs relative to the general air short-haul transport; t ij is the transport time from airport i to airport j; s ij is the 0-1 decision variable, equal to 1 when the general hub i is directly connected with the general hub j, and 0 otherwise; u kl is the 0-1 decision variable, equal to 1 when the airport k and the airport l are both selected as central hubs, and 0 otherwise; M is a positive integer; X ijkl is the 0-1 decision variable, equal to 1 when the transport path from the airport i to the airport j is first through the central hub k and the central hub l, and 0 otherwise.

[0080] The third objective is to maximize the coverage rate of the general air short-haul transport demand, which is defined as that the demand can be covered when the distance between the non-hub and the allocated hub is less than 500 km, and the specific formula is:

[0081]

[0082] where the numerator is the demand that can be covered by the general air short-haul transport service, and the denominator is the total demand of the general air short-haul transport. In the formula, m ij is the input parameter, equal to 1 when the distance between the airport i and the airport j is less than or equal to 500 km, and 0 otherwise. ij

[0083] In the constraint conditions: the hub location and allocation constraints are established to ensure that the number of hub construction meets the requirements, and the non-hub and the general hub are allocated singly; the central hub connection constraints are established to ensure the complete connection between the central hubs; the general hub direct connection constraints are established to enable the direct connection between the general hubs to reduce the number of transfers; the demand flow balance constraints are established to ensure the rationality of the allocation results; and the time constraints are established to calculate the transport time between each demand point.

[0084] (4) The model is solved by using NSGA-III.

[0085] The specific steps in the step (4) include: ​

[0086] 4.1, design NSGA-III to solve the model, first design chromosome, structure as Figure 3 shown.

[0087] The chromosome is divided into two parts, the hub array and the distribution array, and the array length is defined as |N|. The first |CH| genes represent the candidate set of central hub airports, the first |H| genes represent the candidate set of hub airports, and the last |N|-|H| genes are non-hub airports. The genes of the hub array are composed of numbers "0", "1", "2", representing non-hub, ordinary hub, and central hub respectively. The genes of the distribution array represent the distribution of each airport, and the distribution of the central hub is itself.

[0088] 4.2, design selection, crossover and mutation operators.

[0089] The selection operator uses binary tournament to get the offspring chromosome. The crossover operator uses single-point crossover. When the crossover probability P c is satisfied, two individuals are selected for crossover operation. The mutation operator uses exchange mutation. When the mutation probability P m is satisfied, the offspring generated by the crossover operation is subjected to mutation operation.

[0090] 4.3, design the direct connection strategy between ordinary hubs based on transportation cost and time.

[0091] For each two ordinary hubs, the transportation cost and time of direct connection and transit through the central hub are calculated respectively. If the transportation cost and time of direct connection are both less than transit, direct connection is selected. If one of the transportation cost and time of direct connection is less than transit, direct connection or transit is selected with equal probability. If the transportation cost and time of direct connection are both greater than transit, transit is selected.

[0092] 4.4, non-dominated sorting operation is performed on the merged population, and elite individuals are selected.

[0093] The technical scheme of the present application will be further illustrated by a specific embodiment.

[0094] The known route demand data is obtained from the FlightCommons website, and the demand influencing factor indicators are obtained from the 2022 National Economic and Social Development Statistical Bulletin and the airport official website. Taking a 19-seat double-actuated non-pressurized aircraft as an example, the unit cost of general aviation short-distance transportation is set as c0=3 yuan·passenger kilometer -1 , the discount factors are α1=0.25 and α2=0.4. Assuming that the construction cost of an ordinary hub is 1×105 yuan, the discount factor β=5. The speed of the general aircraft used for short-distance transportation is 360km·h -1 , the discount factors are λ1=0.4 and λ2=0.5. A larger positive integer M=100000.

[0095] (1) Route demand prediction results

[0096] A total of 4705 reference samples of known route demand were statistically obtained, including both trunk routes between hub airports and large airports and branch routes between small and medium-sized airports. It was stipulated that routes with a comprehensive similarity greater than 0.6 would be added to the similar reference sample set for parameter calibration. Ten routes were randomly selected from the reference sample, and the passenger flow of the routes was predicted using the method of the present application and the general gravity model, respectively, and the results were compared, as shown in Table 1.

[0097] Table 1 Comparison of prediction results of two models and their errors

[0098]

[0099]

[0100] From the results, it can be seen that the maximum absolute error and relative error obtained by the gravity model based on similarity are 0.229 and 9.41%, respectively, both of which are smaller than those of the general gravity model. The mean absolute error (MAE), mean square error (MSE), mean relative error (MRE) and goodness of fit (R 2 ) were selected as evaluation indexes, and the comparison results are shown in Table 2. It can be found that the method of the present application is superior to the general gravity model in the four indexes, indicating that the gravity model based on similarity has a higher goodness of fit with the actual demand and a better prediction effect.

[0101] Table 2 Comparison of error indexes of two models

[0102]

[0103] (2) Algorithm performance analysis

[0104] To verify the effectiveness of NSGA-III, we introduce an exact solution algorithm: ε-constraint, and two other heuristic algorithms: NSGA-II and MOEA / D for comparison. The solution set obtained by ε-constraint is regarded as the Pareto optimal solution set. Four indicators are selected to evaluate the performance of heuristic algorithms: IGD (Inverted generational distance, the smaller the value, the better the convergence and diversity of the solution set), NPS (Number of Pareto solutions, the larger the value, the more the Pareto solutions), QM (Quality metric, the larger the value, the higher the quality of the solution), and SM (Spacing metric, the smaller the value, the more uniform the distribution of the solution). Since Gurobi cannot effectively solve large-scale instances, we take 20 airports in a province as an example to conduct a small-scale numerical experiment.

[0105] Figure 4 When P CH = 2, P H = 6, the comparison results of the Pareto solution set of the four methods are shown. It can be found that among the three heuristic algorithms, the solution set obtained by NSGA-III is closest to the optimal solution set. When P H = 6, the small-scale route network can basically cover the demand for short-haul transport by general aviation. The IGD indicators of the three heuristic algorithms are 0.225, 0.352, and 0.388, respectively, indicating that NSGA-III has better convergence and diversity.

[0106] (3) Network structure optimization results

[0107] Taking 63 airports in a certain region as an example, we conduct a numerical experiment. When P CH = 2, P H = 10, the visualization network optimization results corresponding to the lowest cost objective are shown in Fig. 6. Figure 5 ​

Claims

1. A trunk-branch multi-layer transportation network optimization method for general aviation needs, characterized by: The following steps are involved: Step 1: Considering the attributes of the route's origin and destination airports, the cities where the airports are located, and the route itself, we establish an index system for factors influencing route passenger demand based on the diversity and quantifiability of the indicators and extract a reference sample set of similar routes. Step 2: Consider the characteristics of different types of airports and routes, select reference sample routes with similar characteristics, and design a gravity model based on similarity to predict route demand; Step 3: With the goals of minimizing transportation costs and hub construction costs, minimizing the longest transportation time, and maximizing the coverage of general aviation short-haul transportation demand, a single-distribution hub location model that allows direct connection is constructed, namely, a trunk-branch multi-layer network optimization model; Step 4: Design the NSGA-III algorithm to solve the proposed trunk-branch multi-layer network optimization model.

2. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1: Consider the attributes of the route's origin and destination airports, the cities where the airports are located, and the route itself, and select relevant factors to form a demand influencing factor index system; Step 1.2: Define the concepts of cosine similarity and distance similarity to calculate the similarity values ​​of various indicators between the target route sample and the reference route sample, establish a comprehensive similarity calculation formula, extract reference samples with comprehensive similarity higher than the set threshold, and form a certain number of similar route reference sample sets.

3. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 2 is characterized in that: In step 1.1, considering the diversity and quantifiability of the indicators, 11 relevant factors were selected to form the indicator system, as follows: Airport-related indicators: airport throughput, number of cities served by flights; City-related indicators: urban population, total GDP, per capita GDP, fiscal budget expenditure, urban commercial grade, tourism revenue, number of A-level scenic spots, and proportion of the tertiary industry; Route-related indicators: route distance.

4. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 2, characterized in that: In step 1.2, the following three similarity concepts are defined: Cosine similarity θ ij The calculation formula is: Among them, the two-dimensional vector a j represents the value of the jth indicator, b ij represents the value of the jth indicator of the i-th reference sample route; Distance similarity ξ ij The calculation formula is: Comprehensive similarity Sim(a j ,b ij ) is calculated as: Among them, z ij It represents the similarity between the jth indicator and the jth indicator vector of the i-th reference sample. Its value is the product of cosine similarity and distance similarity. CV j is the coefficient of variation of the j-th indicator, and its value is the standard deviation of the j-th indicator of all target sample routes divided by the mean.

5. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 1, characterized in that: In step 2, the gravity model formula based on similarity is constructed as follows: Among them, W ij represents the demand for the quantity of goods departing from region i and arriving at region j, q i ,q j represents the size of regions i and j, d ij represents the distance between regions i and j, and k, a, b, and c are the model parameters to be calibrated.

6. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 1, characterized in that: The specific process of step 3 is as follows: Step 3.1: Construct a single-assignment-allowed directly connected hub-and-spoke network multi-objective optimization objective function; The first goal is to minimize the total cost, and the specific formula is: Among them, the first three items are transportation costs, and the last two items are hub construction costs; where: N, H, CH are the set of all airports, the set of candidate points for hub airports, and the set of candidate points for central hubs respectively; w is is the OD flow from airport i to airport s; c0 is the unit transportation cost of general aviation short-haul transportation; α1 is the discount factor of the unit transportation cost required for transportation between central hubs compared with general aviation short-haul transportation; α2 is the discount factor of the unit transportation cost required for transportation between ordinary hubs and between ordinary hubs and central hubs compared with general aviation short-haul transportation; β is the discount factor of the fixed cost required to build a central hub relative to ordinary hubs; d ij is the distance from airport i to airport j; F i The fixed cost required to build a general hub for airport i; x ij is a 0-1 decision variable, which is equal to 1 when airport i is assigned to airport j, and 0 otherwise. jj =1, then select airport j to build the hub; y ii is a 0-1 decision variable. If y ii =1, then choose airport i to build the central hub. If y jl =1 and j≠l, y jl indicates that ordinary hub j is assigned to central hub l; g ijl is an integer decision variable, representing the demand from airport i to hub l via hub j, where at least one of the hubs from hub j to hub l is a common hub; h ijl is an integer decision variable, representing the demand from airport i to hub l via hub j; The second goal is to minimize the maximum time between the starting and ending points of transportation. The specific formula is: The above formula in the brackets is the transportation time for routes that directly connect via ordinary hubs, and the following formula is the transportation time for routes that transfer via central hubs; where: k 1 is a decision variable, representing the earliest arrival time of all demands assigned to hub k; r i 2 is a decision variable, representing the earliest arrival time of all demands assigned to hub i; λ1 is the discount factor for the transportation time between central hubs relative to general aviation short-haul transportation; λ2 is the discount factor for the transportation time between ordinary hubs and between ordinary hubs and central hubs relative to general aviation short-haul transportation; t ij is the transportation time from airport i to airport j; s ij is a 0-1 decision variable, which is equal to 1 when ordinary hub i is directly connected to ordinary hub j, and 0 otherwise; u kl is a 0-1 decision variable, which is equal to 1 when both airport k and airport l are selected as central hubs, and 0 otherwise; M is a positive integer; X ijkl is a 0-1 decision variable, which is equal to 1 when the transportation path from airport i to airport j first passes through hub k and hub l, otherwise it is 0; The third goal is to maximize the coverage of general aviation short-haul transport demand. It stipulates that when the distance between the non-hub and the assigned hub is less than 500km, the demand can be covered. The specific formula is: The numerator is the demand that can be covered by general aviation short-haul transport services, and the denominator is the total demand for general aviation short-haul transport; where: m ij is an input parameter. When the distance between airports i and j is less than or equal to 500 km, m ij =1, otherwise 0; Step 3.2: Construct constraints. To adapt to the multi-layer network scenario of trunk and branch connections, the constraints cover hub and site allocation, central hub connection, ordinary hub direct connection constraints, and flow balance constraints, ensuring that the model adapts to the connection relationship and allocation rules between networks at different levels.

7. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 1, characterized in that: The specific process of step 4 is as follows: Step 4.1: Design the chromosome, including the hub array and the allocation array; Step 4.2: Design selection, crossover, and mutation operators; Step 4.3: Design direct connection strategies between common hubs based on transportation costs and time; Step 4.4: Perform a non-dominated sorting operation on the merged population and select elite individuals.

8. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 7, characterized in that: In step 4.1: The chromosome is divided into two parts: the hub array and the allocation array. The array length is defined as |N|. The first |CH| genes represent the candidate set of central hub airports, the first |H| genes represent the candidate set of hub airports, and the last |N|-|H| genes represent non-hub airports. The genes of the hub array are composed of the numbers "0", "1", and "2", representing non-hubs, ordinary hubs, and central hubs respectively. The genes of the allocation array represent the allocation of each airport, where the allocation of the central hub is itself.

9. The method for optimizing a trunk-branch multi-layer transportation network for general aviation needs according to claim 7, characterized in that: In step 4.2: The selection operator uses a binary tournament to obtain offspring chromosomes; the crossover operator uses a single-point crossover. When the crossover probability P is satisfied, c When two individuals are selected for crossover operation, the mutation operator adopts exchange mutation. When the mutation probability P is satisfied m When , the offspring generated by the crossover operation are selected for mutation operation.

10. The trunk-branch multi-layer transportation network optimization method for general aviation needs according to claim 7, characterized in that: In step 4.3: For every two ordinary hubs, the transportation cost and time of direct connection and transit through the central hub are calculated respectively; if the transportation cost and time of direct connection are both less than transit, direct connection is selected; if one of the objective functions of direct connection transportation cost and time is less than transit, direct connection or transit is selected with equal probability; if the transportation cost and time of direct connection are both greater than transit, transit is selected.

Images