MGWR-based general aviation short-distance transportation demand spatial heterogeneity analysis method

Abstract

The invention discloses a general aviation short-distance transportation demand spatial heterogeneity analysis method based on multi-scale geographically weighted regression (MGWR), relates to the technical field of aviation transportation demand prediction and influence factor analysis, and aims to accurately predict general aviation short-distance transportation demands. And the spatial heterogeneity characteristics of the key factors influencing the general aviation requirements are accurately revealed. The method comprises the following steps: firstly, constructing a generalized cost function containing different types of costs in combination with an inter-city traffic trip mode; secondly, introducing an improved condition Logit model, constructing a competition model among different intercity transportation modes, and predicting potential demands of general aviation short-distance transportation; thirdly, screening factor indexes influencing general aviation short-distance transportation; and finally, constructing an MGWR model, and revealing changes and differences of influence degrees of different factors in space. According to the method, the spatial heterogeneity of the regional general aviation short-distance transportation demand influence factors can be effectively evaluated, and technical support is provided for general aviation airport layout planning and operation management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air transportation demand prediction and influencing factor analysis, and particularly relates to a general aviation short-haul transportation demand spatial heterogeneity analysis method based on multiscale geographically weighted regression (MGWR). BACKGROUND

[0002] General aviation short-haul transportation is an important part of intercity transportation and is playing an increasingly important role and showing unprecedented prospects. Accurate prediction and analysis of general aviation short-haul transportation demand is the prerequisite for general aviation route design, general aviation planning layout and construction, so it is particularly important to study the spatial heterogeneity characteristics of key factors affecting general aviation demand.

[0003] In the analysis and prediction of influencing factors of general aviation short-haul transportation demand, existing researches mostly start from economy, policy and transportation, and most of the researches focus on the qualitative analysis of the influence of a single influencing factor on general aviation demand prediction and development, lack systematic comprehensive analysis of general aviation industry, and ignore the influence of multiple variables on general aviation transportation. In addition, due to the lack of data, how to accurately predict general aviation short-haul transportation demand also needs to be solved. Therefore, qualitative and quantitative analysis of general aviation short-haul transportation demand still needs further exploration.

[0004] In the application of GWR model, existing researches are mostly applied in the fields of city and regional planning, transportation, environment and ecology, etc. However, the traditional GWR model adopts fixed bandwidth and ignores the difference in the action scale of different variables. Although the emerging MGWR allows the optimal bandwidth of each variable to be adjusted adaptively, the application of MGWR in the field of general aviation is still rare. Therefore, it is necessary to introduce MGWR into the field of general aviation short-haul transportation demand analysis, combine scientific demand prediction, capture the multiscale spatial heterogeneity between general aviation short-haul transportation demand and influencing factors, and provide a more adaptive spatial analysis method for general aviation network planning. SUMMARY

[0005] The present application discloses a general aviation short-haul transportation demand spatial heterogeneity analysis method based on MGWR, which aims to accurately predict and analyze general aviation short-haul transportation demand and build a multi-model comparison and evaluation of the spatial heterogeneity of factors affecting general aviation short-haul transportation demand.

[0006] To achieve the above object, the technical scheme provided by the present application is as follows:

[0007] A general aviation short-haul transportation demand spatial heterogeneity analysis method based on MGWR, comprising the following steps:

[0008] Step 1: Consider the characteristics and scenarios of different intercity short-distance transportation modes and construct a generalized cost function that includes time, economy, distance, safety, and delay costs;

[0009] Step 2: Combined with the generalized cost function, an improved conditional logit model is constructed to analyze the passenger flow share of each mode of transportation and predict the potential demand and distribution of general aviation short-haul transportation;

[0010] Step 3: Select factors that affect general aviation short-haul transport from the perspectives of society, economy, transportation, and policy, and screen the influencing factors to extract indicators with significant spatial heterogeneity;

[0011] Step 4: Construct the MGWR model, independently calibrate the optimal bandwidth of each variable, and quantify the spatial heterogeneity of the impact of each influencing factor on general aviation short-haul transport demand.

[0012] To optimize the above technical solutions, specific measures / limitations adopted also include:

[0013] In step 1, the competitive and cooperative relationship among intercity transportation modes is very complex, involving multiple modes of transportation. We analyze the characteristics, usage scenarios, advantages, and disadvantages of different intercity short-distance transportation modes, and consider the multi-type costs when passengers choose different intercity transportation modes to construct a generalized cost function:

[0014] Time cost: All the time a traveler spends while traveling, including the time spent buying tickets, waiting, riding, arriving at, and leaving the waiting points of the transportation mode.

[0015]

[0016] Where T i is the time cost of the transportation mode i selected by the passenger, V(T) is the time value of the passenger, t i The time it takes for the passenger to travel by the mode of transportation he / she chooses, t o The time spent on arriving at and leaving the waiting point for transferring to public transportation, C j is the sum of waiting time of the selected travel mode, t car When passengers choose private cars, taxis and online ride-hailing services, the vehicle's operating time, t p The waiting time for passengers to wait for taxis or online ride-hailing vehicles, the waiting time for private car travel p =0;

[0017] Economic cost: includes fares, luggage fees, insurance premiums, etc. paid for public transportation, as well as fares and tolls for private travel. The specific economic cost E i The calculation formula is:

[0018]

[0019] where F i is the ticket price of the transportation mode selected by the passenger, F o is the cost of urban public transportation between the station of the intercity transportation mode selected by the passenger and the origin-destination of the passenger, F e is the taxi and online car-hailing travel fare, F r is the toll generated by taxi and online car-hailing travel;

[0020] Distance cost: including the running distance of the transportation mode taken by the passenger, the distance of entering and leaving the station, and the running distance of transferring urban public transportation, the specific distance cost L i The calculation formula is:

[0021] L i = D i + D0

[0022] where D i is the distance of the intercity transportation mode selected by the passenger, and D0 is the distance of urban public transportation to and from the station of the transportation mode selected by the passenger;

[0023] Safety cost: refers to the safety risk that needs to be borne when selecting transportation mode i, which is quantified as the annual passenger accident casualty number of the transportation mode, and the specific safety cost S i The calculation formula is:

[0024]

[0025] where Z i is the annual casualty number of transportation mode i, and p i is the annual passenger flow of transportation mode i;

[0026] Delay cost: for private travel, the deviation rate of the expected arrival time and the actual arrival time represents the delay cost, and the punctuality of other transportation tools is measured by the punctuality rate of each transportation mode. The delay cost is represented by Z i .

[0027] In step 2, the utility theory is the basis of the Logit model. Combined with the generalized cost function, the uncertain utility part in the utility function is quantified, and the improved conditional Logit model is established to predict the potential demand of general aviation short-haul transportation. Based on the utility theory, the utility U ij of traveler i selecting intercity transportation mode j is:

[0028]

[0029] where x ti , x ei , x liThe weights of time cost, economic cost and distance cost of each mode of transportation, respectively ij is the random term in the utility function, which contains the utility generated by factors that cannot be directly observed but can affect the travel individual's choice behavior, such as comfort level, service level, etc.

[0030] According to the random utility theory, the probability P ij of traveler i choosing mode j is:

[0031]

[0032] The demand evaluation formula of general aviation short-haul transportation based on the improved conditional Logit model is:

[0033]

[0034] where D GA is the demand of general aviation short-haul transportation, and D o is the total transportation volume between regions.

[0035] In step 3, the key factors affecting the demand of general aviation short-haul transportation are classified and described, and the social economy, transportation mode, land use and other service facilities are selected as the influencing factor indicators. The Pearson correlation coefficient test method and the variance inflation factor test method are used to test the multicollinearity of the influencing factor indicators:

[0036] Pearson correlation coefficient test method: the Pearson correlation coefficient of each influencing factor is tested, and the factors with r coefficient greater than 0.8 are excluded. The formula for calculating the r value of the correlation coefficient is:

[0037]

[0038] where x i is the value of the i-th region, y i is the value of the dependent variable of the i-th region, and represent the sample mean of x and y, respectively, and n is the total number of regions.

[0039] Variance inflation factor test method: VIF is used to measure the severity of multicollinearity, and its calculation formula is:

[0040]

[0041] where R 2 is the coefficient determined by OLS regression of the dependent variable and other independent variables.

[0042] The exploratory spatial data analysis is used to analyze the spatial correlation of the air passenger transport volume and the influencing factors, and the key influencing factors are determined by the significant map of the variables. The global spatial autocorrelation is measured by the global Moran's I:

[0043]

[0044] where x i is the value of the i-th region, n is the total number of regions, w ij is the spatial weight matrix between the i-th and j-th regions, and S o is the aggregation of the spatial weights.

[0045] The local spatial autocorrelation analysis is used to analyze the spatial distribution characteristics of the variables in the local region, and the hot spots, cold spots and spatial outliers are identified. The local Moran's I is used for measurement:

[0046]

[0047] where I i is the local Moran's I of the i-th region, z i and z j are the standardized variable values of the i-th region, and w ij is the spatial weight matrix between the i-th and j-th regions.

[0048] In step 4, the MGWR model is constructed using the screened influencing factors with significant spatial relationship to analyze the spatial differences of the influence degree of each factor. The model performance is evaluated by the RMSE, R 2 , AICc and other indicators. The MGWR is compared with the OLS and the geographically weighted regression model to verify the effectiveness of the MGWR in balancing the fitting degree and complexity:

[0049] OLS model: the OLS model is constructed based on the screened influencing factors by the Pearson correlation test, and the formula is:

[0050]

[0051] where i is the i-th region, y i is the dependent variable of the i-th region, β i0 is the regression constant term, β ik is the regression coefficient of the k-th independent variable, x ik is the k-th independent variable, and ε i is the error term of the i-th region;

[0052] GWR model: the GWR model is constructed by introducing the spatial relationship based on the OLS model, and the expression is:

[0053]

[0054] wherein (u i ,v i ) is the longitude and latitude coordinate of region i, β i0 (u i ,v i ) is the regression constant term, β ik (u i ,v i ) is the regression coefficient of the kth independent variable.

[0055] MGWR model: introduce the spatial scale and the individual bandwidth of each influencing factor to construct the MGWR model, and calculate the action bandwidth of each influencing factor individually. The formula of the MGWR model is:

[0056]

[0057] wherein β bw0 (u i ,v i ) is the regression constant term, β bwk (u i ,v i ) is the regression coefficient of the kth independent variable.

[0058] Compared with the prior art, the beneficial effects of the present application are:

[0059] The present application proposes a general aviation short-haul transport demand prediction and spatial heterogeneity analysis method based on an improved conditional Logit model and MGWR in view of the deficiencies in the existing general aviation short-haul transport demand prediction and spatial heterogeneity analysis. Compared with the traditional method, the improved conditional Logit model comprehensively considers various travel costs to construct a generalized cost function, optimizes the quantification of the uncertainty utility term, and significantly improves the calculation accuracy of the potential demand of general aviation. At the same time, compared with the traditional GWR and OLS model, the MGWR model allows different influencing factors to adopt different bandwidths, and can effectively capture the regional difference of key influencing factors. This method not only intuitively presents the difference in the influence strength of key driving factors in each region, provides technical support for general airport site selection and route optimization, but also fills the research gap of MGWR in the field of general aviation short-haul transport, provides high-precision demand prediction and spatial policy adaptation scheme, and promotes the fine development of general aviation short-haul transport. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is a flowchart of the method of the present application.

[0061] Figure 2 is a technical roadmap of spatial heterogeneity analysis.

[0062] Figure 3 is a model calculation residual contrast chart. DETAILED DESCRIPTION

[0063] The above content of the present application is further illustrated in detail by 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 falls within the scope of the present application.

[0064] The present application proposes a general aviation short-distance transportation demand spatial heterogeneity analysis method based on MGWR, and the flow chart is as shown in Figure 1 The method comprises the following steps:

[0065] (1) considering the characteristics and scenarios of different intercity short-distance transportation modes, a generalized cost function including time, economy, distance, safety and delay cost is constructed.

[0066] The specific steps in step (1) include:

[0067] 1.1, analyze the characteristics, use scenarios, advantages and disadvantages of different intercity short-distance transportation modes, and qualitatively analyze the characteristics of general aviation demand and the relationship between influencing factors.

[0068] 1.2, considering the multi-type cost of passengers in different intercity traffic modes, a generalized cost function including time cost, economic cost, distance cost, safety cost and delay cost is constructed:

[0069] Time cost: all the time spent by travelers when traveling, including ticket purchase, waiting, riding, arriving at and leaving the waiting point of the transportation mode.

[0070]

[0071] Where T i is the time cost of the selected transportation mode i of the passenger, V(T) is the time value of the passenger, t i is the time spent by the passenger in the selected transportation mode, t o is the time spent by the passenger in the selected transportation mode, t j is the sum of waiting time of the selected travel mode, t car is the running time of the vehicle when the passenger chooses private car, taxi and online car, t p is the waiting time of the passenger waiting for a taxi or online car, t p = 0;

[0072] Economic cost: including ticket price, luggage fee, insurance fee, etc. in public transportation, and car fee and road toll, etc. in private travel, and the specific economic cost E i is calculated by the formula:

[0073]

[0074] wherein F i is the ticket price of the travel mode selected by the passenger, F o is the cost of the urban public transport mode between the station of the intercity travel mode selected by the passenger and the origin-destination point of the passenger, F e is the taxi and online car-hailing travel fare, F r is the toll generated by the taxi and online car-hailing travel;

[0075] Distance cost: including the running distance of the transport mode selected by the passenger, the distance of entering and exiting the station, and the running distance of transferring the urban public transport, the specific distance cost L i The calculation formula is:

[0076] L i = D i + D0

[0077] wherein D i is the running distance of the intercity travel mode selected by the passenger, and D0 is the distance of the urban public transport to and from the station of the travel mode selected by the passenger;

[0078] Safety cost: refers to the safety risk to be borne when selecting the travel mode i, which is quantified as the annual passenger accident casualty number of the travel mode, and the specific safety cost S i The calculation formula is:

[0079]

[0080] wherein Z i is the annual casualty number of the travel mode i, and p i is the annual passenger flow of the travel mode i;

[0081] Delay cost: for private travel, the deviation rate of the scheduled arrival time and the actual arrival time is used to represent the delay cost, and the punctuality of other transport tools is measured by the punctuality rate of each travel mode. The delay cost is represented by Z i .

[0082] (2) Combined with the generalized cost function, an improved conditional Logit model is constructed to analyze the passenger flow sharing rate of each transport mode and predict the potential demand and distribution of general aviation short-haul transport.

[0083] The specific steps in the step (2) include:

[0084] 2.1, based on the generalized cost function, the uncertain utility part in the utility function is quantified to obtain the utility function of the individual selecting different travel modes, so as to obtain the conditional probability function of the individual selection and the demand evaluation function of the general aviation short-haul transport:

[0085] Utility theory is the basis of Logit model, the utility U ij :

[0086]

[0087] where x ti , x ei , x li are the weights of time cost, economic cost and distance cost of each mode of transport, respectively, and ε ij is the random term in the utility function, which contains the utility generated by factors that cannot be directly observed but can affect the individual's choice behavior, such as comfort, service level, etc.

[0088] According to the random utility theory, the probability P ij of traveler i choosing mode j is:

[0089]

[0090] The demand estimation formula of general aviation short-haul transport based on the improved conditional Logit model is:

[0091]

[0092] where D GA is the demand of general aviation short-haul transport, and D o is the total transport volume of intercity transport between regions.

[0093] 2.2, the improved conditional Logit model is introduced, and the intercity traffic multi-source data (OD flow, fare, timetable, etc.) is introduced to determine the potential demand and distribution of general aviation short-haul transport in the study area.

[0094] (3) From the perspective of society, economy, transportation and policy, select the factors affecting general aviation short-haul transport, and screen the influencing factors to extract the index with significant spatial heterogeneity.

[0095] The specific steps in step (3) include:

[0096] 3.1, classify and describe the key factors affecting the demand of general aviation short-haul transport, and select the influencing factor indexes of social economy, transportation mode, land use and other service facilities.

[0097] 3.2, use multiple collinearity test to screen the influencing factors, and through the calculation of correlation coefficient and variance inflation factor, eliminate the strongly correlated influencing factors and retain the influencing factors that meet the index. The correlation test method of influencing factor variables is as follows:

[0098] Pearson correlation coefficient test method: Pearson correlation coefficient test is conducted on each influencing factor, and factors with r coefficient greater than 0.8 are excluded. The r value of the correlation coefficient is calculated according to the following formula:

[0099]

[0100] where x i is the value of the i-th region, y i is the dependent variable value of the i-th region, and respectively represent the sample mean of x and y, and n is the total number of regions.

[0101] Variance inflation factor test method: VIF is used to measure the severity of multicollinearity, and its calculation formula is as follows:

[0102]

[0103] where R 2 is the coefficient determined after OLS regression of the dependent variable and other independent variables.

[0104] 3.3. Exploratory spatial data analysis is conducted on the air passenger transport volume and each influencing factor to test its spatial autocorrelation, exclude the influencing factors that do not meet the requirements, generate a significant map of the variable, and determine the key influencing factors input into the regression model. Exploratory spatial analysis method is carried out from two angles of global and local:

[0105] Global spatial autocorrelation: global Moran's index is used to measure:

[0106]

[0107] where x i is the value of the i-th region, n is the total number of regions, w ij is the spatial weight matrix between the i-th and j-th regions, and S o is the aggregation of spatial weights.

[0108] Local spatial autocorrelation: the spatial distribution characteristics of the variable in the local region are analyzed, and hot spots, cold spots and spatial outliers are identified. Local Moran's index is used to measure:

[0109]

[0110] where I i represents the local Moran's index of the i-th region, z i , z j represent the standardized variable value of the i-th region, and w ij represents the spatial weight matrix between the i-th and j-th regions.

[0111] (4) Constructing MGWR model, independently calibrating the optimal bandwidth of each variable, quantifying the spatial heterogeneity of the influence of each factor on the demand of general aviation short-haul transport.

[0112] The specific steps in step (4) include:

[0113] 4.1, using the screened impact factors to construct OLS, geographically weighted regression model and MGWR model to analyze the spatial differences of the influence degree of each factor, and to comprehensively evaluate the model performance through RMSE, R 2 , AICc and other indicators, and to verify the effectiveness of MGWR in balancing fitting degree and complexity. The three spatial heterogeneity regression models constructed are:

[0114] OLS model: based on the Pearson coefficient test to screen the impact factors to construct OLS model, which is a global regression model without considering spatial variation, assuming that all observation values have the same spatial characteristics, and the model formula is:

[0115]

[0116] Where i is the ith region, y i is the dependent variable of the ith region, β i0 is the regression constant term, β ik is the regression coefficient of the kth independent variable, x ik is the kth independent variable, and ε i is the error term of region i;

[0117] GWR model: based on the OLS model, the spatial relationship is introduced to construct the GWR model, and the data within the bandwidth action scale of the GWR model is weighted in each region to calibrate a separate regression model, and the observation values from each region are weighted by distance, and the regression region is represented by the geographic coordinates (μ i ,v i ) of its centroid. The expression of GWR model is:

[0118]

[0119] Where (u i ,v i ) are the latitude and longitude coordinates of region i, β i0 (u i ,v i ) is the regression constant term, and β ik (u i ,v i ) is the regression coefficient of the kth independent variable.

[0120] MGWR model: The GWR model uses a global unified bandwidth to blur the spatial differences of influencing factors between regions. It introduces spatial scale and the individual bandwidth of each influencing factor to construct the MGWR model. The bandwidth of each influencing factor is calculated separately to analyze the spatial differences of influencing factors and the differences between adjacent regions. The MGWR model formula is:

[0121]

[0122] where β bw0 (u i ,v i ) is the regression constant, β bwk (u i ,v i ) is the regression coefficient of the kth independent variable.

[0123] 4.2. Use ArcGIS to visualize the regression coefficients of the influencing factors with greater spatial heterogeneity, and intuitively reflect the impact of a certain influencing factor on the general aviation short-distance transportation demand in various regions.

[0124] The technical solution of the present invention is further illustrated by a specific embodiment below.

[0125] This article uses the general aviation industry in a specific region as an example. As of 2022, the region had established 20 general aviation airports, representing 6% of the nation's total. This article extracts intercity transportation data from county-level cities within the region, including social, economic, road, rail, and air passenger and freight data, for 2022. This data is then mapped to the corresponding urban areas using ArcGIS.

[0126] (1) Analysis of general aviation short-haul transport demand based on the conditional logit model

[0127] The improved conditional logit model was calibrated using partial data on some intercity travel modes in a certain region. The results are shown in Table 1. The model was then applied to the intercity travel mode data in the region to calculate the ratio of travelers choosing general aviation short-distance transportation, thereby obtaining the potential general aviation intercity transportation demand.

[0128] Table 1 Results of improved conditional Logit model

[0129]

[0130] By improving the conditional Logit model, the share of general aviation short-distance transportation in the inter-city short-distance transportation competition market is obtained, and the distribution of general aviation short-distance demand in a certain area is obtained.

[0131] (2) Screening of factors influencing general aviation short-haul transport demand

[0132] The 31 factors including social economic indicators, service economic indicators, service facility indicators and land use indicators in a certain area were selected as the influencing factors of general short-distance transportation demand. First, some influencing factors were removed through Pearson correlation coefficient test. Then, the screened data were input into the OLS model to obtain the VIF value, as shown in Table 2.

[0133] Table 2 Multiple collinearity and spatial autocorrelation test of influencing factors

[0134]

[0135]

[0136] Table 2 (continued) Multiple collinearity and spatial autocorrelation test of influencing factors

[0137]

[0138] (3) Results of spatial heterogeneity model

[0139] After the screening of influencing factors through multiple collinearity and spatial autocorrelation test, the influencing factors with significant spatial relationship were left to construct GWR model and MGWR model. Seven indicators such as residual sum of squares (RSS), log-likelihood, degree of freedom, etc. were selected to evaluate the goodness of fit and complexity of OLS, GWR and MGWR regression models. The technical route is shown in Figure 2 , and the results are shown in Table 3.

[0140] Table 3 Results of OLS, GWR and MGWR regression models

[0141]

[0142] Comparing the goodness of fit of each model, the RSS, Log-likelihood, RMSE, R 2 , Adj R 2 , AICc values of MGWR model are the best among the three regression models, indicating that the fitting effect on the data set in this paper is better, the regression result of general aviation short-distance transportation demand in a certain area is more significant, and the regression result closer to the true situation is obtained. The box plot and violin plot are combined to compare the residuals between the actual observed values and estimated values of each influencing factor in the calculation of OLS, GWR and MGWR three models, as shown in Figure 3 .

[0143] Combined with the coefficients and P values given by GWR and MGWR model regression, the regression coefficients of the factors with greater spatial heterogeneity are visualized and analyzed using ArcGIS, and thus the spatial distribution of the regression coefficients of GWR and MGWR can be obtained. The significant and positive regression coefficients indicate that the greater the influence on the short-distance passenger traffic volume of general aviation, the greater the demand; the significant and negative regression coefficients also have greater influence, but have a negative impact and reduce the demand for navigation; if the regression coefficient is not significant, that is, the P value is greater than 0.05, it is considered that the spatial heterogeneity of the factor in the region is close to zero.

[0144] 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, without departing from the technical solution of the present application, can make any simple modification, equivalent replacement and improvement of the above embodiment according to the technical essence of the present application, which still belongs to the protection scope of the technical solution of the present application.

Claims

1. A method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR, characterized by: The following steps are involved: Step 1: Consider the characteristics and scenarios of different intercity short-distance transportation modes and construct a generalized cost function that includes time, economy, distance, safety, and delay costs; Step 2: Combined with the generalized cost function, an improved conditional logit model is constructed to analyze the passenger flow share of each mode of transportation and predict the potential demand and distribution of general aviation short-haul transportation; Step 3: Select factors that affect general aviation short-haul transport from the perspectives of society, economy, transportation, and policy, and screen the influencing factors to extract indicators with significant spatial heterogeneity; Step 4: Construct the MGWR model, independently calibrate the optimal bandwidth of each variable, and quantify the spatial heterogeneity of the impact of each influencing factor on general aviation short-haul transport demand.

2. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 1 is characterized in that: The specific process of step 1 is as follows: Step 1.1: Analyze the characteristics, usage scenarios, advantages, and disadvantages of different intercity short-distance transportation modes, and qualitatively analyze the characteristics and influencing factors of general aviation demand; Step 1.2: Consider the multi-type costs when passengers choose different intercity transportation modes, and construct a generalized cost function including time cost, economic cost, distance cost, safety cost, and delay cost.

3. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 2 is characterized in that: In step 1.2, the five travel costs are defined as follows: Time cost: all the time a traveler spends on a trip, including the time spent buying a ticket, waiting, riding, arriving at, and leaving the waiting point of the mode of transport; Where T i is the time cost of the transportation mode i selected by the passenger, V(T) is the time value of the passenger, t i The time it takes for the passenger to travel by the mode of transportation he / she chooses, t o The time spent on arriving at and leaving the waiting point for transferring to public transportation, C j is the sum of waiting time of the selected travel mode, t car When passengers choose private cars, taxis and online ride-hailing services, the vehicle's operating time, t p The waiting time for passengers to wait for taxis or online ride-hailing vehicles, the waiting time for private car travel p =0; Economic cost: includes fares, luggage fees, insurance premiums paid for public transportation, and fares and tolls for private travel. The specific economic cost E i The calculation formula is: Among them F i The fare for the mode of transportation chosen by the passenger, F o The cost of the urban public transportation between the station of the intercity transportation mode selected by the passenger and the passenger's starting and ending points, F e For taxi and online car-hailing fares, F r tolls generated for taxi and ride-hailing trips; Distance cost: includes the distance of the transportation mode used by the passenger, the distance of entering and exiting the station, and the distance of transferring to public transportation within the city. The specific distance cost L i The calculation formula is: L i =D i +D0 Among them D i is the distance traveled by the intercity transportation mode selected by the passenger, and D0 is the distance between the arrival and departure of urban public transportation at the station of the transportation mode selected by the passenger; Safety cost: refers to the safety risk that needs to be taken when choosing transportation mode i, which is quantified as the annual number of passenger casualties in accidents involving this transportation mode. The specific safety cost S i The calculation formula is: where Z i is the annual number of casualties in transport mode i, p i is the annual passenger volume of transport mode i; Delay cost: For private travel, the deviation rate between the estimated arrival time and the actual arrival time is used to represent the delay cost. The punctuality of other modes of transportation is measured by the punctuality rate of each mode of transportation. The delay cost is expressed as Z i express.

4. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 1 is characterized in that: The specific process of step 2 is as follows: Step 2.1: Quantify the uncertain utility part of the utility function based on the generalized cost function to obtain the utility function of the individual's choice of different travel modes, thereby obtaining the conditional probability function of the individual's choice and the demand assessment function of general aviation short-distance transportation; Step 2.2: Introduce the improved conditional logit model and import multi-source intercity transportation data to determine the potential general aviation short-distance transportation demand and distribution in the study area.

5. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 4 is characterized in that: In step 2.1, utility theory is the basis of the Logit model. The utility U of traveler i choosing intercity transportation mode j is ij : where x ti 、x ei 、x li are the weights of time cost, economic cost and distance cost of each mode of transportation, ε ij It is a random term in the utility function, which includes the utility generated by factors that cannot be directly observed but can affect the traveler's choice behavior; according to random utility theory, the probability P of traveler i choosing transportation mode j is ij The formula is: The demand assessment formula for general aviation short-distance transportation based on the improved conditional Logit model is: Among them D GA is the demand for general aviation short-distance transportation, D o It is the total transport volume of inter-city transport between regions.

6. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 1 is characterized in that: The specific process of step 3 is as follows: Step 3.1: Classify and describe the key factors affecting general aviation short-haul transport demand, and select indicators of factors affecting social economy, transportation mode, land use, and other service facilities; Step 3.2: Use multicollinearity test to screen the influencing factors, calculate the correlation coefficient and variance inflation factor, eliminate the strongly correlated influencing factors, and retain the influencing factors that meet the indicators; Step 3.3: Conduct exploratory spatial data analysis on air passenger volume and various influencing factors, test their spatial autocorrelation, eliminate incompatible influencing factors, and determine the key influencing factors to input into the regression model.

7. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 6 is characterized in that: The specific method for testing the correlation between influencing factor variables in step 3.2 is: Pearson correlation coefficient test method: Pearson correlation coefficient test is performed on each influencing factor. Factors with r coefficient greater than 0.8 are eliminated. The formula for calculating the r value of the correlation coefficient is: where x i is the value of the i-th region, y i is the value of the dependent variable in region i, and represent the sample means of x and y respectively, and n is the total number of regions; Variance Inflation Factor Test Method: VIF is used to measure the severity of multicollinearity. Its calculation formula is: Among them, R 2 It is the coefficient determined after OLS regression of the dependent variable and other independent variables.

8. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 6 is characterized in that: In step 3.3, the exploratory spatial analysis method is carried out from two perspectives: global and local: Global spatial autocorrelation: measured using the global Moran's index: Among them, x i is the value of the i-th region, n is the total number of regions, w ij is the spatial weight matrix between the i-th and j-th regions, S o is the aggregation of spatial weights; Local Spatial Autocorrelation: Analyze the spatial distribution characteristics of variables in a local area, identify hot spots, cold spots, and spatial outliers, and use the local Moran index to measure: Among them I i represents the local Moran index of region i, z i 、z j represents the standardized variable value of the i-th region, w ij represents the spatial weight matrix between the i-th and j-th regions.

9. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 1 is characterized in that: The specific process of step 4 is as follows: Step 4.1: Use the selected influencing factors to build the MGWR model and analyze the spatial differences in the impact of each factor. 2 , AICc index was used to comprehensively evaluate the model performance, and the OLS and geographically weighted regression models were compared to verify the effectiveness of MGWR in balancing goodness of fit and complexity; Step 4.2: Use ArcGIS to visualize the regression coefficients of the influencing factors with greater spatial heterogeneity, so as to intuitively reflect the impact of a certain influencing factor on the general aviation short-haul transport demand in various regions.

10. The method for analyzing spatial heterogeneity of general aviation short-haul transport demand based on MGWR according to claim 9, characterized in that: In step 4.1, the three spatial heterogeneity regression models constructed are: OLS model: The OLS model is constructed based on the influencing factors screened by the Pearson coefficient test. It is a global regression model that does not consider spatial variation and assumes that all observations have the same spatial characteristics. The model formula is: Where i is the i-th region, y i is the dependent variable of region i, β i0 is the regression constant term, β ik is the regression coefficient of the kth independent variable, x ik is the kth independent variable, ε i is the error term of region i; GWR model: Based on the OLS model, the spatial relationship is introduced to construct the GWR model. The GWR model calibrates a separate regression model in each region in a weighted manner based on the data within the bandwidth scale. The observations from each region are weighted by distance, and the regression region is determined by the geographic coordinates of its centroid (μ i ,v i ) represents; the expression of the GWR model is: Where (u i ,v i ) is the latitude and longitude coordinates of region i, β i0 (u i ,v i ) is the regression constant, β ik (u i ,v i ) is the regression coefficient of the kth independent variable; MGWR model: The GWR model uses a global unified bandwidth to blur the spatial differences of influencing factors among regions. The MGWR model is constructed by introducing spatial scale and the individual bandwidth of each influencing factor. The bandwidth of each influencing factor is calculated separately to analyze the spatial differences of influencing factors and the differences between adjacent regions. The MGWR model formula is: where β bw0 (u i ,v i ) is the regression constant, β bwk (u i ,v i ) is the regression coefficient of the kth independent variable.

Images