An airport flight adjustment method based on an improved entropy value gravity model

Abstract

The application discloses an airport flight adjustment method based on an improved entropy gravity model, acquires multi-source data about an airport, constructs a relevant evaluation system of the airport, analyzes the weight through a TOPSIS algorithm, calculates the coupling degree between two airports in a region, obtains the distance between the airports based on the latitude and longitude coordinates of the airports, solves traffic generation and traffic attraction functions by using fuzzy compromise planning, calculates the entropy values of traffic generation, traffic attraction and airport distance by using an entropy weight method, and constructs an improved entropy gravity model, calculates the OD (Origin-Destination) quantity between the airports, obtains an ideal flight schedule in combination with flight passenger capacity and airport proportion, and adjusts original flights in combination with actual flight numbers. The application combines the coupling degree between the airports, enhances traffic impedance, and is more accurate in ideal OD quantity prediction. The ideal flight information obtained in combination with the flight passenger capacity and the airport proportion is suitable for route recommendation, is helpful to the development of small airports, and can also relieve the passenger carrying pressure of hub airports.

Description

Technical Field

[0001] This invention belongs to the field of transportation, specifically relating to an airport flight adjustment method based on an improved entropy gravity model. Background Technology

[0002] In recent years, my country's civil aviation has continued to develop, with its total air transport turnover, passenger throughput, and cargo and mail throughput all ranking second in the world. By the end of 2022, the passenger throughput of the national airport clusters had reached 520.033 million. Among them, the number of routes of large hub airports such as Beijing Daxing, Shanghai Pudong, and Hangzhou Xiaoshan exceeds half of the total number of routes. The routes of hub airports are oversaturated, and small and medium-sized airports are not developing sufficiently, which seriously restricts the further development of the national airport clusters.

[0003] International research on multi-airport routes began in the 1990s, mostly focusing on flight frequency and airport arrival times, without providing specific methods for predicting flight numbers. Domestic research on multi-airport systems primarily concentrates on qualitative analysis based on existing problems, with limited quantitative research demonstrating scientific rigor and persuasiveness. Therefore, this paper proposes an airport flight adjustment method based on an improved entropy-gravity model. By allocating surplus passenger traffic to other small and medium-sized airports, the flight pressure on hub airports is alleviated, resulting in optimized ideal flight routes. Summary of the Invention

[0004] Purpose of the invention: To address the shortcomings of existing technologies, this invention proposes an airport flight adjustment method based on an improved entropy gravity model, which combines multi-source airport data to adjust airport flights.

[0005] Technical solution: The airport flight adjustment method based on an improved entropy gravity model described in this invention includes the following steps:

[0006] (1) Acquire relevant data from each airport in advance and establish a multi-source data network;

[0007] (2) Establish an airport comprehensive evaluation system based on multi-source data, and use the TOPSIS algorithm to calculate the airport score and solve the coupling degree between airports in the region;

[0008] (3) Based on the latitude and longitude coordinates of the airports and the distance between airports, use fuzzy compromise programming to solve the traffic generation and traffic attraction functions;

[0009] (4) Use the entropy weight method to calculate the entropy values ​​of traffic generation, traffic attraction, and airport distance, and construct an improved entropy gravity model;

[0010] (5) The computer calculates the OD (Original Dispatch) between airports, and combines it with the passenger capacity of flights and the airport ratio to obtain the ideal flight schedule; the original flight schedule is then adjusted based on the actual number of flights.

[0011] Furthermore, the multi-source data in step (1) includes economic factor data, transportation factor data, and conditional factor data affecting flight schedules for each airport; the economic factor data includes GDP, tertiary industry output, and average travel cost; the transportation factor data includes route type, cargo throughput, passenger flow, and takeoff and landing frequency; the conditional factor data affecting flight schedules includes the number of airlines, the number of universities, and the number of listed companies.

[0012] The average travel cost is the average of the travel costs from the city where the target airport is located to the cities where other airports within the region are located.

[0013]

[0014] Among them, C i C represents the average travel cost in city i, where the airport is located. ij The price is the airfare between city i and city j, where the airport is located.

[0015] Furthermore, the process of establishing an airport comprehensive evaluation system based on multi-source data as described in step (2) is as follows:

[0016] The evaluation indicators for airports include route type, shortest travel cost, number of airlines, passenger throughput, and cargo throughput. An initial data matrix is ​​constructed from the statistical data of these evaluation indicators, and a normalized matrix is ​​formed using vector normalization. For benefit-related attributes, the transformation formula is as follows:

[0017]

[0018] For cost-type attributes, the transformation formula is:

[0019]

[0020] In the formula, x ij To convert data into benefits, y ij For the positive rating of the i-th object on the j-th evaluation metric, z ij Cost conversion data; benefit-type attributes are those with higher values, while cost-type attributes are those with lower values, which are better.

[0021] Vector normalization transforms all attribute values ​​in the decision matrix into numbers in the range (0,1), which is beneficial for comparing attributes.

[0022] Determine the set R of positive and negative ideal solutions. + R -In evaluation indicators, larger values ​​are considered better; smaller values ​​are considered better. R0 is constructed using the maximum value of the benefit-type indicator and the minimum value of the cost-type indicator. + Conversely, it constructs R. - ,Right now:

[0023]

[0024]

[0025] In the formula: R + R - Let r be the set of positive and negative ideal solutions respectively. ij For the standardized matrix, j + It belongs to the set of benefit-type indicators, j - It belongs to the set of cost-type indicators;

[0026] Calculate the Euclidean distance between each influencing index and the positive and negative ideal solutions.

[0027]

[0028] In the formula: Let r be the Euclidean distance between the positive and negative ideal solutions. ij For a standardized matrix, The set of maximum values ​​for each column. It is the set of minimum values ​​for each column.

[0029] Calculate the relative closeness Ci of each influencing index to the ideal solution, and rank the schemes according to their values:

[0030]

[0031] In the formula, C i The relative similarity of the proposed solutions.

[0032] Furthermore, the process for solving the coupling degree between airports within the region described in step (2) is as follows:

[0033] Based on the coupling coordination degree model, the degree of correlation between the two airports is analyzed, and the coupling degree value between the two locations is calculated using the following formula:

[0034]

[0035] Where f(x) and g(y) are two airports in the region, and x and y are the index values ​​describing the characteristics of each indicator, and both are dimensionless values.

[0036] Furthermore, the distance between airports mentioned in step (3) is:

[0037] L ij =R·arccos(cos(lat) i ·π / 180))·cos(lat j ·π / 180)·cos(lag i ·π / 180-lat j ·π / 180)+sin(lat i ·π / 180)·sin(lat j ·π / 180) (9)

[0038] Where R is the Earth's radius; lat i lat j It refers to the longitudes of airport i and airport j; lag i lag j These are the latitudes of i and j.

[0039] Furthermore, the process of solving the traffic generation and traffic attraction functions using fuzzy compromise programming in step (3) is as follows:

[0040] Traffic volume is composed of passenger flow, freight throughput, and takeoffs and landings; traffic attraction is composed of the number of universities, the number of listed companies, GDP, and the value of the tertiary industry.

[0041] Construct the preference matrix R a :

[0042]

[0043] Specifically, when comparing the preference degree of the i-th indicator and the j-th indicator, if the latter is much greater than the former, the preference matrix is ​​set to 0 in row i and 2 in column j; if the latter is greater than the former, the preference matrix is ​​set to 0 in row i and 1 in column j; if the two are approximately the same, the preference matrix is ​​set to 1 in row i and 1 in column j.

[0044] The preference matrix R is obtained as follows:

[0045]

[0046] Among them, α, β, γ, δ, ε are real numbers between (0, 1), and α+β=δ+ε=1, δ<α<γ<β<ε;

[0047] Let matrix R define a directed weighted graph G(A,R). The outgoing edge values ​​of this graph are:

[0048]

[0049] Where a is a variable, a∈A; c is a variable, c∈A / {a};

[0050] Obtain the corresponding weighting coefficients:

[0051]

[0052] Where T is the index number of the influencing indicator; t i As a variable, t i ∈T;

[0053] Fuzzy compromise multi-objective programming is used to solve for the indicator weights, and then the traffic generation and traffic attraction functions are solved sequentially:

[0054] M1=∑εG i (14)

[0055] M2=∑σA i (15)

[0056] Where ε represents the weight of each factor in traffic attraction, and G i For the indicators involved in the attraction function, σ represents the weight of each factor in traffic generation, and A... i The indicators involved in the occurrence function are i = 1, 2, 3... n.

[0057] Furthermore, the improved entropy-value gravity model described in step (4) is as follows:

[0058]

[0059] Among them, Q ij Let M1 be the OD (Original Distance) quantity of airport i and airport j; M2 be the traffic attraction function of airport i and airport j; L be the traffic generation function of airport i and airport j; ij denoted as , where is the distance between airports i and j; C is the coupling degree between the two airports, used to measure the degree of connection between the origin and destination; w1, w2, and w3 are parameters used to adjust the weight of each factor on traffic flow.

[0060] Furthermore, the process of calculating the entropy values ​​of traffic generation, traffic attraction, and airport distance using the entropy weight method in step (4) is as follows:

[0061] The traffic generation, traffic attraction, and airport distance data are standardized to eliminate the influence of dimensions, resulting in the standardized matrix F:

[0062]

[0063] Calculate the weight of the index value of the i-th option under the j-th index:

[0064]

[0065] Calculate the entropy value of the j-th index:

[0066]

[0067] Calculate the entropy weight of the j-th indicator, and define the deviation of the j-th indicator as:

[0068] g j =1-e j (20)

[0069] If the decision-maker has no obvious preference for any of the n evaluation indicators, then the entropy weight of the j-th indicator is:

[0070] and

[0071] Performing a logarithmic transformation on the entropy-gravity model, we obtain:

[0072] ln Q ij =w2lnM1+w3lnM2-w1lnL ij C (22)

[0073] In the formula: Q ij Let M1 be the OD (Original Distance) from airport i to airport j; M2 be the traffic attraction function for airports i and j; and L be the traffic generation function for airports i and j. ij denoted as , where is the distance between airports i and j; C is the coupling degree between the two airports, used to measure the degree of connection between the origin and destination; w1, w2, and w3 are the airport distance weights, traffic attraction weights, and traffic generation weights, respectively, used to adjust the weights of each factor on traffic flow.

[0074] Furthermore, the implementation process of step (5) is as follows:

[0075] The OD (Original Distance) value is calculated by inputting data such as traffic attraction, traffic generation, airport distance, and coupling degree between the two airports.

[0076] Calculate the number of flights:

[0077] Number of flights = OD ÷ (Flight passenger capacity × Aircraft type ratio)

[0078] Obtain the idealized flight schedule and compare it with the actual flights. If the allocated number is less than the actual number of flights between airport i and airport j, then routes should be added between airport i and airport j. If the allocated number is greater than the actual number of flights between airport i and airport j, then routes should be reduced between airport i and airport j.

[0079] Beneficial Effects: Compared with existing technologies, the beneficial effects of this invention are as follows: This invention takes airports as the research object, acquires multi-source data, organizes different data, establishes a comprehensive airport evaluation system, uses the TOPSIS algorithm to preprocess the data and obtain airport scores, calculates the correlation between two airports through the coupling degree algorithm, and uses fuzzy compromise programming to weight and solve the gravity model parameters. It uses Python to obtain airport latitude and longitude data and determine the distance between airports, establishes an entropy gravity model suitable for the aviation system through the entropy weight method, and uses coupling degree to strengthen the geographical impedance between two regions, solving the OD quantity between the two airports to make the prediction results more accurate. Furthermore, by combining flight passenger capacity and aircraft type ratio with the average load factor of the studied airports, an ideal flight schedule is obtained. Based on the original routes, the actual number of flights is compared to adjust the flights, ultimately obtaining a better flight adjustment scheme. This invention is applicable to airports of all sizes, suitable for route recommendation, and can alleviate the burden on core hub airports in the Yangtze River Delta region, changing the oversaturated state of hub airports and allowing more other smaller airports to handle surplus passenger volume. It not only helps the development of small airports but also alleviates the passenger load pressure on hub airports. Attached Figure Description

[0080] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0081] The present invention will now be described in further detail with reference to the accompanying drawings.

[0082] like Figure 1 As shown, this invention proposes an airport flight adjustment method based on an improved entropy-gravity model. Taking Shanghai Pudong Airport in the Yangtze River Delta region as an example, the method includes the following steps:

[0083] Step 1: Collect multi-source data from the airport and establish a multi-source data network.

[0084] Due to the rapid economic development and dense population of the Yangtze River Delta region, as well as its more developed transportation compared to other regions, the data involved is more diversified. The data collected in this invention includes economic factors such as GDP, tertiary industry output, and average travel costs; transportation factors such as route type, cargo throughput, passenger flow, and takeoff and landing frequency; and factors affecting flight frequency such as the number of airlines, universities, and listed companies.

[0085] Considering that airfare prices are affected by factors such as tourism and returning to work, we collected airfare prices from various airports to other airports, built a dataset, and used Python to randomly select values ​​to calculate the average travel cost. The average travel cost is calculated by taking the average of the travel costs from the city where the target airport is located to the cities where other airports in the region are located.

[0086]

[0087] Among them, C i C represents the average travel cost in city i, where the airport is located. ij Let i be the airfare between city i and city j, where the airport is located; the results are shown in Table 1.

[0088] Table 1: Shortest Travel Costs to Each Airport

[0089]

[0090] Step 2: Based on multi-source data and airport-related information, the comprehensive airport evaluation system is finally determined, and the airport score is calculated using the TOPSIS algorithm (Technique for Order Preference by Similarity to Ideal Solution, often referred to as the superior-inferior solution distance method in China), and the coupling degree between airports in the region is calculated.

[0091] The evaluation indicators include route type, shortest travel cost, number of airlines, passenger throughput, and cargo throughput. Among them, route type refers to whether there are direct routes between the two airports, number of airlines refers to the number of the same airlines in the two airports, and average travel cost, passenger throughput, and cargo throughput reflect the transportation scale and traffic of the airports, and indirectly reflect the connection between the two airports.

[0092] The initial data matrix is ​​constructed from the determined statistical data of the indicators, and a normalized matrix is ​​formed by using the vector normalization method.

[0093] For benefit-type attributes, the transformation formula is:

[0094]

[0095] For cost-type attributes, the transformation formula is:

[0096]

[0097] Where, x ij For effect transformation data, y ij For the positive rating of the i-th object on the j-th evaluation metric, z ij Cost conversion data; benefit-type attributes are those with higher values, while cost-type attributes are those with lower values.

[0098] The normalized matrices are shown in Table 2:

[0099] Table 2 Normalized Matrix

[0100]

[0101]

[0102] Vector normalization transforms all attribute values ​​in the decision matrix into numbers in the range (0,1), which is beneficial for comparing attributes.

[0103] Determine the set R of positive and negative ideal solutions. + ,R - Indicators that are better the larger they are are considered benefit-oriented, while those that are better the smaller they are considered cost-oriented. R is constructed using the maximum value of the benefit-oriented indicator and the minimum value of the cost-oriented indicator. + Conversely, it is constructed as follows:

[0104]

[0105]

[0106] Among them, R + R - Let r be the set of positive and negative ideal solutions respectively. ij For the standardized matrix, j + It belongs to the set of benefit-type indicators, j - It belongs to the set of cost-type indicators.

[0107] As shown in Table 3, the Euclidean distances between each influencing index and the positive and negative ideal solutions are calculated.

[0108]

[0109] In the formula: Let r be the Euclidean distance between the positive and negative ideal solutions. ij For a standardized matrix, The set of maximum values ​​for each column. It is the set of minimum values ​​for each column.

[0110] Table 3. European Distance Table

[0111]

[0112]

[0113] As shown in Table 4, the relative closeness Ci between each influencing index and the ideal solution is calculated, and the schemes are ranked according to their values:

[0114]

[0115] Among them, C i The relative similarity of the proposed solutions.

[0116] Table 4 Final Scores and Rankings

[0117]

[0118] Using f(x) as the score for Shanghai Pudong Airport and g(y) as the score for other airports, and substituting them into the coupling degree formula, the results are shown in Table 5.

[0119]

[0120] Where x and y are the index values ​​describing the characteristics of each index, and both are dimensionless values.

[0121] Table 5 Airports and their coupling values

[0122]

[0123] Step 3: Use Python to crawl the latitude and longitude coordinates of the airports and calculate the distance between airports. Then, use fuzzy compromise programming to solve for the traffic generation and traffic attraction functions.

[0124] The flight path distance between airport i and airport j is calculated using the formula for flight path distance calculation and Jupyter Notebook is used as follows:

[0125] L ij =R·arccos(cos(lat) i ·π / 180))·cos(lat j ·π / 180)·cos(lag i ·π / 180-lat j ·π / 180)+sin(lat i ·π / 180)·sin(lat j ·π / 180) (9)

[0126] In the formula: R is the Earth's radius, denoted as 6373 km; lat i lat j It refers to the longitudes of i and j; lag i lag j These are the latitudes of i and j.

[0127] Python was used to crawl the latitude and longitude coordinates of 20 airports in the Yangtze River Delta region, and the flight distance was calculated according to the flight route distance calculation formula. Jupyter Notebook was used to calculate the flight route distance between Pudong Airport and other airports, as shown in Table 6.

[0128] Table 6 Airport Distances Between Shanghai Pudong Airport and Other Airports

[0129]

[0130] Fuzzy compromise multi-objective programming is used to solve for the weights of the indicators, and traffic generation and traffic attraction are solved sequentially.

[0131] Traffic attraction: number of universities, number of listed companies, GDP, tertiary industry value; Traffic generation: passenger flow, freight throughput, takeoffs and landings. The specific steps are as follows:

[0132] There are many factors affecting airport traffic volume; we will analyze three main ones: Passenger Traffic: Passenger traffic is a crucial indicator of airport busyness, directly reflecting the number of passengers. Cargo Throughput: Cargo throughput refers to the total amount of cargo handled by the airport, including goods, baggage, and express deliveries transported by cargo flights. With economic development and the rise of e-commerce, increased cargo throughput will drive up the number of cargo flights and impact airport traffic demand. Takeoffs and Landings: Takeoffs and landings are closely related to passenger traffic and cargo throughput, but their impact on airport traffic demand may be relatively smaller. Therefore, the order of importance of the three influencing indicators is: Passenger Traffic > Cargo Throughput > Takeoffs and Landings.

[0133] Several factors influence airport traffic volume. Number of Universities: Cities with a high number of universities typically have an advantage in academic research, talent cultivation, and technological innovation, attracting a large number of students, teachers, and researchers. This population flow increases airport passenger traffic, especially during holidays and academic conferences. GDP: Economically developed cities generally have greater business travel demand and more frequent domestic and international business exchanges. This generates greater flight demand and airport traffic attractiveness. Number of Listed Companies: Cities with a higher number of listed companies typically have more active business exchanges and travel demand. Tertiary Sector Value: The tertiary sector, including services, finance, and information technology, plays a vital role in the economic structure. Cities with higher tertiary sector output typically have more business activities and business travel demand, significantly impacting airport traffic and increasing passenger volume and airport traffic attractiveness. Considering all these factors, the order of importance is: City GDP > Number of Universities > Number of Listed Companies > Tertiary Sector Value.

[0134] Construct the preference matrix R a :

[0135]

[0136] Among them, when comparing the preference degree of the i-th indicator and the j-th indicator, when the latter is much greater than the former, the i-th row and j-th column of the preference matrix is ​​0 and the j-th row and i-th column is 2; when the latter is greater than the former, the i-th row and j-th column of the preference matrix is ​​0 and the j-th row and i-th column is 1; when the two are approximately the same, the i-th row and j-th column of the preference matrix is ​​1 and the j-th row and i-th column is 1.

[0137] The preference matrix R is obtained as follows:

[0138]

[0139] Among them, α, β, γ, δ, and ε are real numbers between (0, 1), and α+β=δ+ε=1, δ<α<γ<β<ε. Generally, α=0.4, β=0.6, γ=0.5, δ=0.2, ε=0.8;

[0140] Let matrix R define a directed weighted graph G(A,R). The outgoing edge values ​​of this graph are:

[0141]

[0142] Where a is a variable, a∈A; c is a variable, c∈A / {a};

[0143] Obtain the corresponding weighting coefficients:

[0144]

[0145] Where T is the sequence number of the four influencing indicators; t i As a variable, t i ∈T;

[0146] Fuzzy compromise multi-objective programming is used to solve for the indicator weights, and then the traffic generation and traffic attraction functions are solved sequentially:

[0147] M1=∑εG i (14)

[0148] M2=∑σA i (15)

[0149] Where ε represents the weight of each factor in traffic attraction, and G i For the indicators involved in the attraction function, σ represents the weight of each factor in traffic generation, and A... i The indicators involved in the occurrence function are i = 1, 2, 3... n.

[0150] In this embodiment, the preference matrix of the four influencing indicators of traffic attraction is obtained: passenger flow > cargo throughput > takeoffs and landings > load factor.

[0151]

[0152] Obtain the preference matrix:

[0153]

[0154] The weights of the four influencing indicators are calculated as follows:

[0155] λ2=0.3, λ3=0.2,

[0156] Therefore, the preference matrix and results for the three influencing indicators of traffic volume are as follows:

[0157]

[0158]

[0159] μ3 = 0.2

[0160] Substituting into formulas (14) and (15), we finally obtain the numerical values ​​of traffic attraction and traffic generation, as shown in Table 7.

[0161] Table 7. Traffic attraction and traffic generation data for each airport.

[0162]

[0163]

[0164] Step 4: Use the entropy weight method to calculate the entropy values ​​of traffic generation, traffic attraction, and airport distance, and construct an improved entropy gravity model.

[0165] The improved entropy-gravity model incorporates inter-airport coupling, enhances traffic impedance, and yields more accurate predictions. The steps are as follows:

[0166]

[0167] Among them, Q ij Let M1 be the OD (Original Distance) quantity of airport i and airport j; M2 be the traffic attraction function of airport i and airport j; L be the traffic generation function of airport i and airport j; ij Let represent the distance between airports i and j; C represents the coupling degree between the two airports, used to measure the degree of connection between the origin and destination; w1, w2, and w3 are the airport distance weights, traffic attraction weights, and traffic generation weights, respectively.

[0168] Data such as traffic generation, traffic attraction, and airport distance are standardized to eliminate the influence of dimensions, resulting in the standardized matrix F:

[0169]

[0170] Calculate the weight of the index value of the i-th option under the j-th index:

[0171]

[0172] Calculate the entropy value of the j-th index:

[0173]

[0174] Calculate the entropy weight of the j-th indicator, and define the deviation of the j-th indicator as:

[0175] g j =1-e j (20)

[0176] If the decision-maker has no obvious preference for any of the n evaluation indicators, then the entropy weight of the j-th indicator is:

[0177] and

[0178] For ease of calculation, a logarithmic transformation is performed on the entropy gravity model, resulting in equation (4-7):

[0179] lnQ ij =w2lnM1+w3lnM2-w1lnL ij C (22)

[0180] In the formula: Q ij Let M1 be the OD (Original Distance) from airport i to airport j; M2 be the traffic attraction function for airports i and j; and L be the traffic generation function for airports i and j. ij denoted as , where is the distance between airports i and j; C is the coupling degree between the two airports, used to measure the degree of connection between the origin and destination; w1, w2, and w3 are the airport distance weights, traffic attraction weights, and traffic generation weights, respectively, used to adjust the weights of each factor on traffic flow.

[0181] In this embodiment, the data entropy weight initialization matrix is ​​established by integrating the information of each airport, as shown in Table 8.

[0182] Table 8. Traffic attraction and traffic generation data for each airport.

[0183]

[0184] The data was standardized using MATLAB, and the information entropy values ​​were calculated as shown in Table 9.

[0185] Table 9 Entropy Value Table

[0186]

[0187] The final weights of each primary indicator are shown in Table 10.

[0188] Table 10 Weights of Primary Indicators

[0189]

[0190] Step 5: Calculate the Origin Destination (OD) volume between airports and the number of relevant flights based on flight passenger capacity and airport ratio; adjust the original flights based on the actual number of flights.

[0191] The OD (October Distance) was calculated using the entropy gravity model by incorporating data such as traffic attraction, traffic generation, airport distance, and coupling degree between the two airports. The results are shown in Table 11.

[0192] Table 11 Airport Od Quantity

[0193]

[0194]

[0195] The calculation method for the number of flights is: number of flights = OD ÷ (flight passenger capacity × aircraft type ratio). The passenger capacity and ratio of various aircraft types for domestic flights in my country can be found on the Ctrip official website. This invention selects the number of passengers and the ratio of different aircraft types of different airlines for statistical analysis, as shown in Table 12.

[0196] Table 12 Aircraft Type Passenger Capacity Table

[0197]

[0198] In calculating OD (Occupancy Rate), this invention uses the weighted average of all statistical passenger capacity and aircraft type ratios in Table 12 as the number of passengers on each flight. The calculation process is as follows: Therefore, in the calculation of OD (Occupancy Rate), 143 is used as the number of passengers on each flight to obtain the ideal flight schedule.

[0199] The idealized flight schedule was obtained. According to data survey, the average passenger load factor of Shanghai Pudong Airport is 82.6%. The solved data was rounded to integers, and only the decimal part was retained to meet the average passenger load factor. The data was then compared with the actual flight schedule. The results are shown in Table 13.

[0200] Table 13 Ideal Flight Schedule

[0201]

[0202] If the allocated number of flights is less than the actual number of flights between airport i and airport j, then an additional route should be added between airport i and airport j; if the allocated number of flights is greater than the actual number of flights between airport i and airport j, then an additional route should be added between airport i and airport j.

[0203] Based on the data in Table 13, the flights at Shanghai Pudong Airport were adjusted according to the original routes. The results of the adjustments are shown in Table 14.

[0204] Table 14 Adjustment Plan

[0205]

Claims

1. An airport flight adjustment method based on an improved entropy-gravity model, characterized in that, Includes the following steps: (1) Acquire relevant data from each airport in advance and establish a multi-source data network; (2) Establish an airport comprehensive evaluation system based on multi-source data, and use the TOPSIS algorithm to calculate the airport score and solve the coupling degree between airports in the region; (3) Based on the latitude and longitude coordinates of the airports and the distance between airports, use fuzzy compromise programming to solve the traffic generation and traffic attraction functions; (4) Use the entropy weight method to calculate the entropy values ​​of traffic generation, traffic attraction, and airport distance, and construct an improved entropy gravity model; (5) Calculate the OD (Original Dispatch) volume between the computer airports, combine it with the flight passenger capacity and airport ratio to obtain the ideal flight schedule; adjust the original flight schedule based on the actual number of flights. The process of solving the traffic generation and traffic attraction functions using fuzzy compromise programming in step (3) is as follows: Traffic volume is composed of passenger flow, freight throughput, and takeoffs and landings; traffic attraction is composed of the number of universities, the number of listed companies, GDP, and the value of the tertiary industry. Constructing the preference matrix : （10） Among them, compared with the first Type of indicator and the first The preference matrix is ​​determined by the degree of preference for certain indicators, and when the latter is much greater than the former. OK Take 0 in the column. OK The column takes 2; when the latter is greater than the former, the preference matrix is... OK Take 0 in the column. OK Columns are set to 1; when the two are approximately equal, the preference matrix... OK Take 1 in the column. OK Take 1; The preference matrix R is obtained as follows: （11） in, Let be a real number between (0, 1), and , ; Let matrix R define a directed weighted graph G(A,R). The outgoing edge values ​​of this graph are: （12） Where a is a variable, a∈A; c is a variable, c∈A / {a}; Obtain the corresponding weighting coefficients: （13） Where T is the index number of the influencing indicator; As variables, ; Fuzzy compromise multi-objective programming is used to solve for the indicator weights, and then the traffic generation and traffic attraction functions are solved sequentially: (14) （15） in, The weights of each factor in traffic attraction. The indicators involved in the attraction function, The weights of each factor in traffic volume, The indicators involved in the occurrence function, ; The improved entropy gravity model described in step (4) is as follows: （16） in, For the airport With the airport OD amount; for and The airport's traffic attraction function; for and Traffic generation function of the airport; for and Distance between airports; C is the coupling degree between two airports, used to measure the degree of connection between the origin and destination; These are parameters used to adjust the weight of each factor on traffic flow; The process of calculating the entropy values ​​of traffic generation, traffic attraction, and airport distance using the entropy weight method in step (4) is as follows: The traffic generation, traffic attraction, and airport distance data are standardized to eliminate the influence of dimensions, resulting in the standardized matrix F: （17） Calculate the first The first item under the indicator The weighting of the indicator values ​​for each scheme: （18） Calculate the entropy value of the j-th index: （19） Calculate the entropy weight of the j-th indicator, and define the deviation of the j-th indicator as: （20） If the decision-maker has no obvious preference for any of the n evaluation indicators, then the entropy weight of the j-th indicator is: （21） Performing a logarithmic transformation on the entropy-gravity model, we obtain: （22） In the formula: For the airport Arrive at the airport OD amount; for and The airport's traffic attraction function; for and Traffic generation function of the airport; for and The distance between airports; C is the coupling degree between two airports, used to measure the degree of connection between the origin and destination; These are the airport distance weight, traffic attraction weight, and traffic generation weight, used to adjust the weight of each factor on traffic flow.

2. The airport flight adjustment method based on an improved entropy-gravity model according to claim 1, characterized in that, The multi-source data in step (1) includes economic factor data, transportation factor data, and conditional factor data affecting flight schedules for each airport; the economic factor data includes GDP, tertiary industry output value, and average travel cost; the transportation factor data includes route type, cargo throughput, passenger flow, and takeoff and landing frequency; the conditional factor data affecting flight schedules includes the number of airlines, the number of universities, and the number of listed companies.

3. The airport flight adjustment method based on an improved entropy-gravity model according to claim 1, characterized in that, The process of establishing an airport comprehensive evaluation system based on multi-source data in step (2) is as follows: The evaluation indicators for airports include route type, shortest travel cost, number of airlines, passenger throughput, and cargo throughput. An initial data matrix is ​​constructed from the statistical data of these evaluation indicators, and a normalized matrix is ​​formed using vector normalization. For benefit-related attributes, the transformation formula is as follows: （2） For cost-type attributes, the transformation formula is: （3） In the formula, To convert data into benefits, The positive rating of the i-th object on the j-th evaluation metric. Cost conversion data; benefit-type attributes are those with higher values, while cost-type attributes are those with lower values, which are better. Vector normalization transforms all attribute values ​​in the decision matrix into numbers in the range (0,1), which is beneficial for comparing attributes. Determine the set of positive and negative ideal solutions , Evaluation indicators that are larger are considered benefit-type indicators, while those that are smaller are considered cost-type indicators. The evaluation is constructed using the maximum value of the benefit-type indicators and the minimum value of the cost-type indicators. Conversely, it is constructed as ,Right now: = ,j=1,2,...,n (4) = ,j=1,2,...,n (5) In the formula: , These are the sets of positive and negative ideal solutions, respectively. For a standardized matrix, It belongs to the set of benefit-oriented indicators. It belongs to the set of cost-type indicators; Calculate the Euclidean distance between each influencing index and the positive and negative ideal solutions. , : ,i=1,2,...,m;j=1,2,...,n (6) In the formula: , The Euclidean distance between the positive and negative ideal solutions. For a standardized matrix, The set of maximum values ​​for each column. It is the set of minimum values ​​for each column. Calculate the relative closeness Ci of each influencing index to the ideal solution, and rank the schemes according to their values: = ,i=1,2,...,m (7) In the formula, The relative similarity of the proposed solutions.

4. The airport flight adjustment method based on an improved entropy-gravity model according to claim 1, characterized in that, The process of solving for the coupling degree between airports within the region described in step (2) is as follows: Based on the coupling coordination degree model, the degree of correlation between the two airports is analyzed, and the coupling degree value between the two locations is calculated using the following formula: （8） Where f(x) and g(y) are two airports in the region, and x and y are the index values ​​describing the characteristics of each indicator, and both are dimensionless values.

5. The airport flight adjustment method based on an improved entropy-gravity model according to claim 1, characterized in that, The distance between airports mentioned in step (3) is: （9） Where R is the Earth's radius; , These are the longitudes of airport i and airport j; , These are the latitudes of i and j.

6. The airport flight adjustment method based on an improved entropy-gravity model according to claim 1, characterized in that, The implementation process of step (5) is as follows: The OD (Original Distance) value is calculated by inputting data on traffic attraction, traffic generation, airport distance, and coupling degree between the two airports. Calculate the number of flights: Obtain an idealized flight schedule and then compare it with the actual flight schedule; If the allocated number is less than the actual number of flights between airport i and airport j, then an additional route should be added between airport i and airport j. If the allocated number of flights exceeds the actual number of flights between airport i and airport j, then the number of routes between airport i and airport j should be reduced.

7. The airport flight adjustment method based on an improved entropy-gravity model according to claim 2, characterized in that, The average travel cost is the average of the travel costs from the city where the target airport is located to the cities where other airports within the region are located. = （1） Among them, C i C represents the average travel cost in city i, where the airport is located. ij The price is the airfare between city i and city j, where the airport is located.

Images