EVTOL vertical take-off and landing airport site selection method considering travel mode selection behavior

Abstract

The invention discloses an eVTOL vertical take-off and landing airport site selection method considering travel mode selection behaviors, and relates to the field of urban low-altitude traffic facility site selection. The method comprises the following steps: firstly, considering a travel mode selection behavior, and constructing an urban air traffic (UAM) travel demand prediction model; secondly, solving an eVTOL vertical take-off and landing airport candidate point set by adopting an iterative weighted clustering algorithm according to an original traffic mode time value of a demand point; and finally, constructing a site selection model by taking reduction of the total cost of the network, reduction of the total travel time of the network and improvement of the UAM service rate as targets, and solving by adopting an NSGA-III algorithm. According to the method, the optimal eVTOL vertical take-off and landing airport site selection scheme can be given, meanwhile, the optimal travel mode of each journey of the whole network is solved, the model is more practical, and the urban travel efficiency is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of urban low-altitude transportation facility site selection, and particularly relates to a vertical take-off and landing airport site selection method for eVTOL considering travel mode selection behavior. BACKGROUND

[0002] With the acceleration of urbanization, major cities around the world are facing increasingly serious traffic congestion problems, causing travel delays, energy waste and environmental pollution. Urban air mobility (UAM) as a new transportation mode has gradually become an important means to alleviate the above problems. Electric vertical take-off and landing aircraft (eVTOL) as the core transportation tool of UAM can significantly improve travel efficiency, reduce noise and air pollution, and achieve efficient and green urban travel. Infrastructure construction is the basic work of developing new transportation modes, so it is necessary to study the site selection of eVTOL vertical take-off and landing airports.

[0003] At present, most of the eVTOL vertical take-off and landing airport site selection problems are single-oriented eVTOL mode research, lacking comprehensive consideration of multi-mode travel in the city, such as travelers can choose taxi, bus, subway and other pure ground transportation modes or eVTOL mode to complete the trip. In addition, most of the researches directly take demand points as potential eVTOL vertical take-off and landing airport candidate points, ignoring the eVTOL travel multi-modal transport process, i.e. two ground transportation from the departure point to the first vertical take-off and landing airport, from the second vertical take-off and landing airport to the destination, and one air transportation between the two vertical take-off and landing airports. Therefore, it is necessary to consider the travel mode selection behavior and the eVTOL travel multi-modal transport process to provide a more realistic eVTOL vertical take-off and landing airport site selection scheme. SUMMARY

[0004] The present application discloses a vertical take-off and landing airport site selection method considering travel mode selection behavior, which aims to consider multi-mode travel in the city and eVTOL travel multi-modal transport process, provide vertical take-off and landing airport site selection scheme and travel mode selection results for each trip, and improve network travel efficiency.

[0005] To achieve the above object, the technical scheme provided by the present application is as follows:

[0006] A vertical take-off and landing airport site selection method considering travel mode selection behavior, comprising the following steps:

[0007] Step 1: considering travel mode selection behavior, constructing an urban air mobility (UAM) travel demand prediction model;

[0008] Step 2: according to the original traffic mode time value of the demand point, using an iterative weighted clustering algorithm to solve the eVTOL vertical take-off and landing airport candidate point set;

[0009] Step 3: Build a site selection model with the goal of reducing the total network cost, reducing the total network travel time, and improving the UAM service rate, and solve it using the NSGA-III algorithm.

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

[0011] In step 1, taxis, buses, and subways are selected as potential competitive modes for eVTOL travel. Annual travel data for the three ground transportation modes are collected and preprocessed to remove abnormal data.

[0012] Considering the purpose of the study and the limitations of the sample data, travel time, distance, and cost are selected as factors affecting travel mode choice behavior. Realistic route simulation is used to supplement the time, distance, and cost data for each trip in the three ground transportation modes.

[0013] Random utility theory is a key factor in quantifying the impact of travel time, distance, and cost on travel mode choice behavior, and is the basis for building a UAM travel demand forecasting model. The utility U mn of trip n choosing transportation mode m is as follows:

[0014] U mn = β1t mn + β2d mn + β3f mn + ∈ mn

[0015] where [β1, β2, β3] represents the time, distance, and cost parameter vector to be estimated, [t mn , d mn , f mn ] represents the time, distance, and cost feature vector of trip n choosing transportation mode m, and ∈ mn is the random error term of the utility function.

[0016] The standard conditional probability P mn of trip n choosing transportation mode m is as follows:

[0017]

[0018] where M is the set of available travel modes.

[0019] The actual selected transportation mode for each trip and the time, distance, and cost data in the three ground transportation modes are input into the model to estimate the parameter vector [β1, β2, β3]; if the actual selected transportation mode for trip n is m, then P mn = 1, otherwise P mn = 0.

[0020] By assuming the eVTOL unit distance transportation cost, eVTOL flight speed, average waiting time and other parameters, the time, distance and cost data of each trip in the eVTOL travel mode are calculated, and the data are combined with the three ground transportation mode data.

[0021] The parameter estimation results are updated into the UAM travel demand prediction model, and the combined data (time, distance and cost data of each trip in the three ground transportation modes and the eVTOL mode) are input into the model to predict the eVTOL mode selection probability P eVTOL,n .

[0022] The city air traffic travel demand is obtained by P eVTOL,n .

[0023] In step 2, according to the original transportation mode time value of the demand point, the eVTOL vertical take-off and landing airport candidate point set is solved by using the iterative weighted clustering algorithm, and the specific steps include:

[0024] Step 1: According to the city air traffic travel demand prediction results, the trip data (including trip origin and destination geographical location, original transportation mode, city air traffic demand) are combined as the trip origin data and destination data to obtain the city air traffic demand point set.

[0025] Step 2: The trip with eVTOL mode has high time value characteristics, so the eVTOL vertical take-off and landing airport should be close to the high time value demand point, and the demand point time value is determined by its original transportation mode. According to the time value of taxi, bus and subway three transportation modes, the weights of the three are designed, and if the original transportation mode of demand point i is m, then its weight w m :

[0026]

[0027] Among them, VOTT m represents the time value of transportation mode m, and VOTT taxi represents the time value of taxi mode.

[0028] Randomly select k sample points from the demand point set as the initial clustering center, and the weighted distance D ij of demand point i and clustering center j is as follows:

[0029]

[0030] Among them, w m represents the weight of the original transportation mode of demand point m, r represents the earth radius 6371km, lon i and lat i represent the longitude and latitude of demand point i, respectively, and lonj and lat j denote the longitude and latitude of the cluster center j, respectively.

[0031] Each demand point is assigned to the nearest cluster center.

[0032] Step 3: Calculate the weighted average position of all demand points in the cluster, get the new cluster center of the k cluster, and reassign all demand points based on the weighted distance.

[0033] Step 4: Compare the position change of the new and old cluster center, judge whether it reaches the convergence condition. If not, repeat Step 3 until convergence, then calculate the average silhouette coefficient S of the demand point set:

[0034]

[0035] Where n represents the total number of demand points in the data set, s(i) represents the silhouette coefficient of each demand point i, a(i) represents the average distance between demand point i and other points in the same cluster, which measures the tightness of the cluster; b(i) represents the average distance from demand point i to all points in the nearest neighbor cluster, which measures the separation degree between clusters.

[0036] If demand point i belongs to cluster C i , then the intra-cluster tightness a(i) and inter-cluster separation b(i) are:

[0037]

[0038] Where |C i | represents the number of demand points in cluster C i , d(i,j) represents the distance between demand point i and demand point j, |C k | represents the number of demand points in cluster C k , d(i,j) represents the distance between demand point i and demand point j.

[0039] Step 5: Replace k value and repeat Step 2-4, calculate the average silhouette coefficient of the clustering result under different k value, select the k value corresponding to the peak value of the average silhouette coefficient as the best candidate point number, and its clustering result is the best candidate point set.

[0040] In Step 3, first arrange the candidate point set data, and the trip data including the geographical position of the trip start and end point, trip demand, three ground transportation modes (taxi, bus, subway) and eVTOL mode characteristics, etc.

[0041] Considering the multi-mode travel in the city and the UAM intermodal process, a site selection model is constructed to reduce the total network cost, reduce the total travel time of the network, and improve the UAM service rate.

[0042] The model allows each trip p to select a pure ground transportation mode (taxi, bus, subway) or an eVTOL mode, wherein the eVTOL mode is a multimodal process including a first ground transportation, a second air flight and a third ground transportation, and the two ground transportations select one of the taxi, bus and subway modes.

[0043] The target of reducing the total network cost and the total network travel time is respectively:

[0044]

[0045] Wherein K, P and M represent the candidate point set, the trip set and the ground transportation mode set respectively, m and n represent the vertical airports closest to the start and end points of the trip p respectively, y k , c c represent whether to build a vertical take-off and landing airport at the candidate point k and the construction cost, D p represents the travel demand of the trip p, represents whether the trip p selects a pure ground transportation mode c, the corresponding cost and time, x p , c pmn , t pmn , t w represents whether the trip p selects a UAM service, the cost, flight time and transfer waiting time of flying from the vertical airport m to the vertical airport n, represents whether the trip p selects a ground transportation mode a from the departure place to the vertical airport m, the corresponding cost and time, represents whether the trip p selects a ground transportation mode b from the vertical airport n to the destination, the corresponding cost and time.

[0046] The target of improving the UAM service rate is:

[0047]

[0048] Wherein D p represents the travel demand of the trip p, s p represents whether the trip p selects an eVTOL trip.

[0049] The model constraint conditions cover the number of vertical airports, the uniqueness of the travel mode selection, the vertical airports passed by the eVTOL trip must be the constructed airports and the maximum acceptable time of the two ground transportations of the eVTOL trip.

[0050] The NSGA-III algorithm is used to solve the model, and the site selection scheme and the travel mode selection result of each trip are output.

[0051] Compared with the prior art, the beneficial effects of the present application are:

[0052] The present application aims at the insufficient research, and proposes an eVTOL vertical take-off and landing airport site selection method considering travel mode selection behavior. At present, most of the researches are single-oriented eVTOL mode, and lack of comprehensive consideration of multi-mode travel in the city. The present application considers the competition prediction of UAM demand of various ground transportation modes and eVTOL mode, and allows each trip to select pure ground transportation (taxi, bus, subway) or eVTOL in the mathematical model, which improves the deficiency of this aspect of research. In addition, most of the researches directly take the demand point as the potential eVTOL vertical take-off and landing airport candidate point, ignoring the multi-modal transport process of eVTOL travel. The present application screens the candidate point set close to the demand point with high time value based on the iterative weighted clustering algorithm, and describes the eVTOL travel process as three stages, including the first stage of ground transportation, the second stage of air flight and the third stage of ground transportation, and allows each trip selecting UAM service to select the first and third stages of ground transportation mode. The present application comprehensively considers the travel mode selection behavior and the multi-modal transport process of eVTOL travel, constructs a multi-objective site selection model with low cost, high efficiency and high service satisfaction rate, and provides an eVTOL vertical take-off and landing airport site selection scheme and travel mode selection result of each trip which is more in line with the actual situation. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is a flow chart of the method of the present application.

[0054] Figure 2 is a schematic diagram of multi-mode travel in the city and multi-modal transport process of eVTOL travel.

[0055] Figure 3 is a flow chart of the NSGA-III algorithm.

[0056] Figure 4 is a broken line graph of the average silhouette coefficient of the clustering results under different k values.

[0057] Figure 5 is a distribution diagram of eVTOL vertical take-off and landing airport candidate points.

[0058] Figure 6 is a distribution diagram of 20 vertical take-off and landing airports. DETAILED DESCRIPTION

[0059] The above content of the present application will be further described in the form of examples, 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.

[0060] The present application proposes an eVTOL vertical take-off and landing airport site selection method considering travel mode selection behavior, and the flow chart is as shown in Figure 1 , which comprises the following steps:

[0061] (1) Considering the travel mode choice behavior, a model for predicting the demand of urban air mobility (UAM) travel is constructed.

[0062] The specific steps in step (1) include:

[0063] 1.1, Selecting a taxi, bus and subway as potential competitive modes of eVTOL travel, collecting annual travel data of the three ground transportation modes and performing data preprocessing to eliminate abnormal data.

[0064] 1.2, Considering the research purpose and the limitations of sample data, selecting travel time, distance and cost as factors affecting travel mode choice behavior, and supplementing the time, distance and cost data of each trip in the three ground transportation modes by simulating real routes.

[0065] 1.3, Based on the random utility theory, a utility function is constructed to quantify the influence of travel time, distance and cost on travel mode choice behavior. The utility U mn of trip n choosing transportation mode m is as follows:

[0066] U mn = β1t mn + β2d mn + β3f mn + ∈ mn

[0067] Where [β1, β2, β3] represents the time, distance and cost parameter vector to be estimated, [t mn , d mn , f mn ] represents the time, distance and cost feature vector of trip n choosing transportation mode m, and ∈ mn is the random error term of the utility function.

[0068] The standard conditional probability P mn of trip n choosing transportation mode m is as follows:

[0069]

[0070] Where M is the set of available travel modes.

[0071] The actual selected transportation mode of each trip and the time, distance and cost data in the three ground transportation modes are input into the model to estimate the parameter vector [β1, β2, β3]; if the actual selected transportation mode of trip n is m, then P mn = 1, otherwise P mn = 0.

[0072] 1.4. Assuming an eVTOL unit distance transportation cost of 5.5 yuan per kilometer, an eVTOL flight speed of 200 kilometers per hour, and average waiting time and other parameters as shown in Table 1, calculate the time, distance, and cost data for each trip under the eVTOL travel mode and merge them with the data from the three ground transportation modes.

[0073] Table 1 Time, distance, and cost calculation parameter settings in eVTOL mode

[0074]

[0075] 1.5. Update the parameter estimation results into the UAM travel demand forecast model and input the combined data (time, distance, and cost data of each trip in the three ground transportation modes and the eVTOL mode) into the model to predict the probability of eVTOL mode selection P eVTOL,n Urban air traffic demand is driven by P eVTOL,n The travel demand is obtained by taking trip n.

[0076] (2) Based on the time value of the original traffic mode of the demand point, an iterative weighted clustering algorithm is used to solve the candidate point set of eVTOL vertical take-off and landing airports.

[0077] The specific steps in step (2) include:

[0078] 2.1. Based on the urban air traffic demand forecast results, the trip data (including the geographical locations of the starting and ending points, the original traffic mode, and the urban air traffic demand) are used as the departure and destination data of the trip to obtain the urban air traffic demand point set.

[0079] 2.2. The eVTOL mode has a high time value. Therefore, the eVTOL vertical take-off and landing airport should be close to the high time value demand point. The time value of the demand point is determined by its original transportation mode. The weights of the three transportation modes of taxi, bus and subway are designed according to their time value. If the original transportation mode of demand point i is m, then its weight w is m :

[0080]

[0081] Among them, VOTT m represents the time value of traffic mode m, VOTT taxi It represents the time value of the taxi mode. The time value parameter settings of the three ground transportation modes are shown in Table 2.

[0082] Table 2 Time value and weight parameter settings for three ground transportation modes

[0083]

[0084] Randomly select k sample points from the demand point set as initial cluster centers, and calculate the weighted distance D between demand point i and cluster center j ij :

[0085]

[0086] where w m represents the weight of the original traffic mode of demand point i, r represents the radius of the earth 6371km, lon i and lat i represent the longitude and latitude of demand point i respectively, lon j and lat j represent the longitude and latitude of cluster center j respectively.

[0087] Assign each demand point to the nearest cluster center.

[0088] 2.3, calculate the weighted average position of all demand points in the cluster, get the new cluster center of the k cluster, and reassign all demand points based on the weighted distance.

[0089] 2.4, compare the position change of new and old cluster centers, judge whether to reach the convergence condition. If not, repeat 2.3 until convergence, then calculate the average silhouette coefficient S of the demand point set:

[0090]

[0091] where n represents the total number of demand points in the data set, s(i) represents the silhouette coefficient of each demand point i, a(i) represents the average distance between demand point i and other points in the same cluster, which measures the tightness of the cluster; b(i) represents the average distance from demand point i to all points in the nearest neighbor cluster, which measures the separation degree between clusters.

[0092] If demand point i belongs to cluster C i , then the intra-cluster tightness a(i) and inter-cluster separation b(i) are:

[0093]

[0094] where |C i | represents the number of demand points in cluster C i , d(i,j) represents the distance between demand point i and demand point j, |C k | represents the number of demand points in cluster C k , d(i,j) represents the distance between demand point i and demand point j.

[0095] 2.5, repeat 2.2-2.4 with different k values, calculate the average silhouette coefficient of clustering results under different k values, and select the k value corresponding to the peak value of the average silhouette coefficient as the best candidate point quantity, and the clustering result is the best candidate point set.

[0096] (3) A site selection model is constructed to reduce the total network cost, reduce the total travel time of the network, and improve the UAM service rate, and NSGA-III algorithm is used to solve.

[0097] The specific steps in step (3) include:

[0098] 3.1, first integrate the candidate point set data, and the trip data including the geographic location of the trip origin and destination, travel demand, characteristics of three ground transportation modes (taxi, bus, subway) and eVTOL mode.

[0099] 3.2, considering the multi-modal travel and UAM intermodal process in the city, a site selection model is constructed to reduce the total network cost, reduce the total travel time of the network, and improve the UAM service rate.

[0100] The model allows each trip p to choose a pure ground transportation mode (taxi, bus, subway) or an eVTOL mode, where the eVTOL mode is a multi-modal process including a first ground transportation, a second air flight, and a third ground transportation, and the two ground transportations choose one of the taxi, bus, and subway modes; the process diagram is shown in Figure 2 .

[0101] The targets of reducing the total network cost and reducing the total travel time of the network are respectively:

[0102]

[0103] Where K, P, and M represent the candidate point set, the trip set, and the ground transportation mode set respectively, m and n represent the nearest vertical airport from the origin and the destination of trip p, y k , c c represent whether to build a vertical take-off and landing airport at candidate point k and the construction cost, D p represents the travel demand of trip p, represents whether trip p chooses a pure ground transportation mode c, the corresponding cost and time, x p , c pmn , t pmn , t w represents whether trip p chooses UAM service, the cost of flying from vertical airport m to vertical airport n, flight time, and transfer waiting time, represents whether trip p chooses ground transportation mode a to go from the departure place to vertical airport m, the corresponding cost and time, denotes whether the trip p chooses the ground transportation mode b from the vertical airport n to the destination, the corresponding cost and time.

[0104] The goal of improving the UAM service rate is:

[0105]

[0106] wherein, denotes D p The trip demand of trip p, s p denotes whether the trip p chooses the eVTOL trip.

[0107] The model constraints include the number of vertical airport construction, the uniqueness of mode choice, the vertical airport that the eVTOL trip passes through must be a constructed airport, and the maximum acceptable time of the two ground transportation segments of the eVTOL trip.

[0108] 3.3, the NSGA-III algorithm is used to solve the model, and the site selection scheme and the mode selection result of each trip are output. The flow chart of the NSGA-III algorithm is shown in Figure 3 .

[0109] The technical scheme of the present application will be further illustrated below with a specific embodiment.

[0110] A certain city is selected as the research area. The data is derived from the 2019 commuting information dataset of a certain city, the 2017-2019 taxi information dataset, and the 2019-05 bus and subway card swiping dataset.

[0111] (1) Urban air transportation trip demand prediction

[0112] First, the dataset is preprocessed, and the data is supplemented and improved as shown in the example of Table 3.

[0113] Table 3 Example of data preprocessing results

[0114]

[0115] The estimated results of time, distance, and cost parameters are shown in Table 4.

[0116] Table 4 Estimated results of time, distance, and cost parameters

[0117]

[0118] The parameter estimation results are updated to the UAM trip demand prediction model, and the merged data (time, distance, and cost data of each trip under three ground transportation modes and eVTOL mode) are input into the model to predict the eVTOL mode selection probability and further obtain the urban air transportation trip demand distribution.

[0119] (2) Screening of eVTOL vertical take-off and landing airport candidate points

[0120] The average silhouette coefficient of the clustering results corresponding to k values from 30 to 120 is counted and is shown in the histogram as Figure 4 The k value corresponding to the peak value 6.3 of the average silhouette coefficient is selected as the optimal clustering number, and the distribution of 109 eVTOL vertical take-off and landing airport candidate points is obtained as Figure 5 .

[0121] (3) Site selection scheme and travel mode selection results of each trip

[0122] A site selection model is constructed to reduce the total network cost, the total network travel time and improve the UAM service rate, and the NSGA-III algorithm is used to solve the model. A site selection scheme is obtained when the maximum acceptable time of the first and third ground traffic is 1200s and the number of vertical airports is 20 as Figure 6 , and the travel mode selection results of some trips are shown in Table 5.

[0123] Table 5 Travel mode selection results of some trips

[0124]

[0125] Under the site selection scheme, 28.5% of the travel demand in the whole network is satisfied by the urban air traffic service, and the eVTOL vertical take-off and landing airports are located in the central and southern regions of a city with high travel demand, effectively sharing the ground traffic pressure in the region and alleviating the urban traffic congestion problem.

[0126] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Any skilled person in the art can make any simple modification, equivalent replacement and improvement to the above embodiment according to the technical essence of the present application without departing from the scope of the technical solution of the present application, and all of the above still belong to the protection scope of the present application.

Claims

1. A method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior, characterized in that: The following steps are involved: Step 1: Considering travel mode selection behavior, build an urban air traffic (UAM) travel demand prediction model; Step 2: Based on the time value of the original traffic mode of the demand point, an iterative weighted clustering algorithm is used to solve the candidate point set of eVTOL vertical take-off and landing airports; Step 3: Construct a location selection model with the goal of reducing the total network cost, reducing the total network travel time, and improving the UAM service rate, and use the NSGA-III algorithm to solve it.

2. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1: Select taxis, buses, and subways as potential competing modes for eVTOL travel. Collect and pre-process the annual travel data for these three ground transportation modes to remove abnormal data. Step 1.2: Considering the research objectives and sample data limitations, we selected three factors influencing travel mode choice: time, distance, and cost. We then simulated real-world routes to supplement the time, distance, and cost data for each trip under the three ground transportation modes. Step 1.3: Based on random utility theory, construct a UAM travel demand forecasting model to estimate time, distance, and cost parameters; Step 1.4: By assuming parameters for eVTOL unit distance transportation cost, eVTOL flight speed, and average waiting time, calculate the time, distance, and cost data for each trip under the eVTOL travel mode and combine them with the data from the three ground transportation modes; Step 1.5: Update the parameter estimation results into the UAM travel demand forecasting model and input the combined data into the model to predict the eVTOL mode selection probability P eVTOL,n , the demand for urban air traffic is driven by P eVTOL,n The travel demand is obtained by taking trip n.

3. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 2, characterized in that: In step 1.3, the utility U of trip n choosing transportation mode m mn as follows: U mn =β1t mn +β2d mn +β3f mn +∈ mn Among them, [β1,β2,β3] represents the time, distance, and cost parameter vectors to be estimated, [t mn ,d mn ,f mn ] represents the time, distance, and cost feature vectors of trip n choosing transportation mode m, ∈ mn is the random error term of the utility function; The standard conditional probability P of choosing transportation mode m for trip n mn as follows: Among them, M is the set of optional travel modes; The actual transportation mode selected for each trip and the time, distance, and cost data under the three ground transportation modes are input into the model to estimate the parameter vector [β1,β2,β3]. If the actual transportation mode selected for trip n is m, then P mn =1, otherwise P mn =0.

4. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 1, characterized in that: The specific process of step 2 is as follows: Step 2.1: Based on the urban air traffic demand forecast results, the trip data is used as the departure and destination data of the trip to obtain the urban air traffic demand point set; Step 2.2: Design weights based on the time value of the three ground transportation modes: taxi, bus, and subway. Randomly select k sample points from the demand point set as initial cluster centers. Calculate the weighted distance between the demand point and each cluster center, and assign each demand point to the cluster center closest to it. Step 2.3: Calculate the weighted average position of all demand points in the cluster, obtain the new cluster centers of k clusters, and redistribute all demand points based on the weighted distance; Step 2.4: Compare the changes in the positions of the new and old cluster centers to determine whether the convergence condition has been met; if converged, calculate the average silhouette coefficient of the demand point set; if not, repeat step 2.3; Step 2.5: Repeat steps 2.2-2.4 with different k values, calculate the average silhouette coefficient of the clustering results under different k values, select the k value corresponding to the peak of the average silhouette coefficient as the optimal number of candidate points, and the clustering result is the optimal candidate point set.

5. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 4, characterized in that: In step 2.2, the weights of the three transportation modes, taxi, bus, and subway, are designed according to their time values. If the original transportation mode of demand point i is m, then its weight w m : Among them, VOTT m represents the time value of traffic mode m, VOTT taxi represents the time value of the taxi model; The weighted distance D between demand point i and cluster center j ij as follows: Among them, w m The weight of the original traffic mode of the demand point is m, r represents the radius of the earth 6371km, lon i and lat i Represent the longitude and latitude of demand point i, lon j and lat j denote the longitude and latitude of cluster center j respectively.

6. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 4, characterized in that: In step 2.4, the average silhouette coefficient S of the demand point set is: Where n represents the total number of demand points in the data set, s(i) represents the silhouette coefficient of each demand point i, a(i) represents the average distance between demand point i and other points in the same cluster, which measures the closeness within the cluster; b(i) represents the average distance from demand point i to all points in the nearest neighbor cluster, which measures the separation between clusters. If demand point i belongs to cluster C i , then the intra-cluster compactness a(i) and inter-cluster separation b(i) are: Among them, |C i | represents cluster C i The number of internal demand points, d(i,j) represents the distance from demand point i to demand point j, |C k | represents cluster C k The number of internal demand points, d(i,j) represents the distance from demand point i to demand point j.

7. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 1, characterized in that: The specific process of step 3 is as follows: Step 3.1: Organize candidate point set data and trip data including the geographic locations of the starting and ending points, travel demand, and characteristics of the three ground transportation modes and the eVTOL mode; Step 3.2: Considering the multimodal travel and UAM intermodal transport process within the city, a location selection model is constructed with the goal of reducing the total network cost, reducing the total network travel time, and improving the UAM service rate; Step 3.3: Use the NSGA-III algorithm to solve the model and output the site selection plan and the travel mode selection results for each trip.

8. The method for selecting an eVTOL vertical take-off and landing airport site considering travel mode selection behavior according to claim 7, characterized in that: In step 3.2, the model allows each trip p to choose either a pure ground transportation mode or an eVTOL mode. The eVTOL mode is a multimodal transportation process, including a first ground transportation segment, a second air flight segment, and a third ground transportation segment. The two ground transportation segments can be selected from one of the modes: taxi, bus, or subway. The goals of reducing the total network cost and the total network travel time are: Among them, K, P, and M represent the candidate point set, itinerary set, and ground transportation mode set, respectively. m and n represent the vertical airport closest to the starting point and end point of itinerary p, respectively. k 、c c Indicates whether to build a vertical take-off and landing airport at candidate point k and the construction cost, D p represents the travel demand for trip p, Indicates whether trip p chooses pure ground transportation mode c, the corresponding cost and time, x p 、c pmn , t pmn , t w Indicates whether trip p chooses UAM service, the cost of flying from vertiport m to vertiport n, the flight time and transfer waiting time, Indicates whether trip p chooses ground transportation mode a from the departure point to the vertical airport m, the corresponding cost and time, Indicates whether trip p chooses ground transportation mode b to travel from vertical airport n to the destination, the corresponding cost and time; The goals of improving UAM service rates are: Where, represents D p The travel demand of trip p, s p Indicates whether trip p chooses eVTOL travel; The model constraints include the number of vertical airports to be built, the uniqueness of travel mode selection, the vertical airport that eVTOL travel passes through must be a construction airport, and the maximum acceptable time for the two ground transportation sections of eVTOL travel.

Images