A flight four-dimensional track intelligent regulation method based on a regular penalty strategy

Abstract

The present application relates to a kind of flight four-dimensional track intelligent regulation and control method based on regular penalty strategy, belong to air traffic flow management field, the flight regulation and control method in prior art cannot effectively and accurately identify the problem of key delay flight is solved.The method of the present application includes step 1, establishes air traffic network flight regulation and control problem model;Step 2, design the regular penalty regulation and control strategy of air traffic network;Step 3, based on the air traffic network flight regulation and control problem model of step 1 and the regular penalty strategy regulation and control flight four-dimensional track of step 2.Compared with prior art, the present application is based on air traffic network delay mechanism Design targeted regulation and control method, with low delay cost, low operation cost Realize the mutual matching of space resources and flight flow demand of aviation network, can provide auxiliary decision support for air traffic management department, airline, airport and other stakeholders.

Description

Technical Field

[0001] This invention belongs to the field of air traffic flow management technology, specifically relating to a method for intelligent control of four-dimensional flight paths based on a regularized penalty strategy. Background Technology

[0002] Air traffic networks, comprised of airports, airspace control units, and air routes, serve as the "nerve center" supporting the safe and efficient execution of a series of flight plans by aircraft. However, with the rapid development of the air transport industry, the contradiction between continuously increasing flight demand and limited airspace resources has become increasingly prominent. Problems such as air traffic congestion, flight safety accidents, and large-scale flight delays are intensifying, severely restricting the high-quality and sustainable development of air transport. Optimizing air traffic network operations through technologies such as flight delay control and networked collaborative regulation has become an effective means to overcome the bottlenecks in air traffic safety and operational efficiency.

[0003] Currently, there are two main types of flight operation control methods: one is the traditional air traffic network control method, which formulates the operation rules into mathematical programming models and optimizes the allocation of air traffic flow by implementing control operations on aircraft. However, the solution methods mainly rely on exact solution algorithms such as branch and bound, resulting in low control efficiency. In response, intelligent control methods for air traffic network operation have become one of the hot research directions in the domestic and international air traffic control academic and industrial communities in recent years. These methods explore the inherent operational laws of air traffic and design targeted heuristic control strategies. They use iterative optimization or data-driven strategies to simplify the cumbersome traversal search tasks in combinatorial optimization methods. However, existing work lacks sufficient depth in its heuristic strategies, and the control algorithms have limitations. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes an intelligent four-dimensional flight trajectory control method based on a regularized penalty strategy. Based on the delay evolution mechanism, it analyzes the causal relationship between airports and delays, examines the ripple effects of delays on the air traffic network, and designs a targeted regularized penalty control strategy for flights, focusing on busy operating hours, bottleneck airspace units, and critical flights. Furthermore, an intelligent four-dimensional flight trajectory control algorithm is designed based on the regularized penalty strategy, thereby significantly improving the operational safety and efficiency of the air traffic network. This solves the problem that existing flight control methods cannot effectively and accurately identify critical delayed flights.

[0005] This invention provides a method for intelligent control of four-dimensional flight paths based on a regularized penalty strategy, specifically including the following steps:

[0006] Step 1: Establish a model for the air traffic network flight control problem;

[0007] Step 2: Design a regularized penalty control strategy for the air traffic network. The specific steps are as follows:

[0008] Step 21: Use transitive entropy to quantify the causal relationship of delays for any airport and establish a weighted causal relationship network for delays.

[0009] Step 22: Perform a random walk in the delay causal weighted association network to obtain the delay penalty coefficient between any two airports;

[0010] Step 3: Based on the air traffic network flight control problem model in Step 1 and the regularized penalty strategy in Step 2, control the four-dimensional flight trajectory.

[0011] Optionally, the specific steps for establishing the air traffic network flight control problem model in step 1 are as follows:

[0012] Step 11: Based on the three-dimensional bounded region model of the air traffic network, set the auxiliary variables and decision variables of the model;

[0013] The expressions for the auxiliary variables of the model are:

[0014]

[0015] in, This indicates the state of flight f arriving at node k at time t with altitude layer l;

[0016] The expression for the decision variables of the model is:

[0017]

[0018] in, This indicates that flight f has arrived at node k at altitude layer l before time t;

[0019] Step 12: Based on the auxiliary and decision variables of the model, construct a model for the flight control problem in the air traffic network, expressed as:

[0020]

[0021] Where x is the flight's scheduled takeoff time; C(x) is the operational efficiency cost; and E(x) represents the fairness of airline delay allocation.

[0022] Optionally, operational efficiency costs can be obtained based on ground waiting costs and in-flight waiting costs.

[0023] Optionally, airline delay allocation fairness can be obtained based on the delay costs of all flights and the average delay cost.

[0024] Optionally, when establishing the air traffic network flight control problem model in step 1, constraints are set for the air traffic network flight control problem model; the constraints include restrictions on ground waiting time, air waiting time, safe flight interval, node continuity, altitude layer uniqueness, altitude layer adjustment, altitude layer before aircraft landing, aircraft landing time, takeoff airport capacity, landing airport capacity, sector capacity, and time continuity.

[0025] Optionally, in step 21, the specific steps for quantifying the causal relationship of delays for any airport pair using transfer entropy and establishing a weighted causal relationship network for delays are as follows:

[0026] The m-th order transfer entropy between any two airports is expressed as:

[0027]

[0028] in, Airport v i airport v j m-order transfer entropy; Airport v i The historical delay value at time t; Airport v j The historical delay value at time t+1; Airport v j The sequence of historical delay values ​​of length m; H(·) represents the information entropy operator; I(·) is the conditional mutual information operator; x1, x2 and x3 respectively represent Elements in;

[0029] The m-order transit entropy between any two airports constitutes the set of m-order transit entropy between any two airports.

[0030] Using transfer entropy to quantify the causal relationship of delays for any airport, a weighted causal relationship network for delays is established, expressed as:

[0031] G * =(V,E) * ,W)

[0032] Where V represents the set of airport nodes; E * Let W represent the set of flight routes; let W represent the set of m-order transit entropy between any two airports.

[0033] Optionally, the specific steps in step 22 to perform a random walk in the delay causal weighted association network to obtain the delay penalty coefficient between any two airports are as follows:

[0034] Random walks are performed in the delayed causal weighted association network to obtain a delayed propagation path of length Q.

[0035] By adjusting the number of airports affected by delay propagation in a random walk of length Q, the spatiotemporal network of delay propagation is obtained, expressed as:

[0036] G cau =(V,E) cau ,D);

[0037] Among them, E cau Let represent the set of propagation paths of delays; D represents the set of propagation delay values ​​for airports.

[0038] The propagation delay values ​​of all airports constitute the set of airport propagation delay values. The expression for the propagation delay value of each airport is as follows:

[0039]

[0040] in, Airport v i The propagation delay value; Indicates upstream airport v i-1 The propagation delay value, Airport v i The absorption delay value; Indicates upstream airport v i-1 airport v i m-order transfer entropy;

[0041] The delay penalty coefficient between any two airports can be obtained based on the delay propagation spatiotemporal network.

[0042]

[0043] in, Airport v i With airport v j Delay penalty coefficient between Airport v i With airport v j The set of all delayed propagation paths between them; Indicates upstream airport v i-1 With airport v i The shortest distance between them.

[0044] Optionally, the specific steps in step 3 for regulating the four-dimensional flight trajectory based on the air traffic network flight regulation problem model in step 1 and the regularized penalty strategy involved in step 2 are as follows:

[0045] Step 31: Based on the regularized penalty control strategy of the air traffic network in Step 2, generate a set of feasible takeoff times for flights.

[0046] Step 32: Arrange all flights in the set of feasible departure times according to their control order, encode the flights by chromosome, and obtain the initial population;

[0047] Step 33: Based on the bi-objective integer programming model in Step 1, use a greedy strategy to process each chromosome in the initial population in turn to obtain feasible solutions for the chromosomes; put the feasible solutions for the chromosomes into the pre-trained encoding library.

[0048] Step 34: Use a hash table to search for efficiency and fairness target values ​​for each chromosome in the pre-trained encoding library, and generate the optimal feasible solution for each chromosome; based on the optimal feasible solution for each chromosome, obtain the dominance level and crowding degree of each chromosome.

[0049] Step 35: Based on the dominance level and crowding of all chromosomes, apply an elite strategy to obtain the parent population;

[0050] Step 36: Perform selection and gene neighbor shift genetic operations on the parent population to generate the offspring population; merge and iterate the parent and offspring populations to generate a new generation population;

[0051] Step 37: Determine whether the generated new generation population meets the convergence condition. If the convergence condition is met, obtain the optimal control takeoff time for each flight and proceed to step 38. If not, return to step 34.

[0052] Step 38: Obtain the four-dimensional control trajectory based on the optimal control takeoff time for each flight.

[0053] Optionally, the feasible solution for each chromosome is the efficiency and fairness target value of the combination of regulatory sequences for each flight.

[0054] Optionally, based on the delay penalty coefficient between any two airports in step 22, the feasible departure time for each flight is obtained, expressed as:

[0055]

[0056] in, Indicates the scheduled departure time of flight f; This indicates the maximum ground waiting time for flight f; Let Ω represent the set of feasible flight times for flight f at node k; Ω represents the set of flights. Airport v i With airport v j Delay penalty coefficient between;

[0057] The feasible departure times for each flight are sorted in ascending order of adjustment to obtain a set of feasible departure times.

[0058] The advantages and positive effects of the large-scale intelligent flight control method for air traffic networks proposed in this invention are at least as follows:

[0059] (1) Taking into account the causal relationship between airports and actual flight operation information, the delay ripple effect of air traffic network was analyzed, revealing the inherent operational characteristics of air traffic network and the spatiotemporal evolution law of delay.

[0060] (2) By exploring the inherent operational laws of air traffic, this invention designs a regularized penalty control strategy based on the delay evolution mechanism and proposes a four-dimensional flight trajectory intelligent control method based on the regularized penalty strategy, which can quickly and accurately adjust the takeoff time of each flight and generate a more economical and fair decision-making scheme. Attached Figure Description

[0061] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0062] Figure 1 This is a flowchart of the intelligent control method for four-dimensional flight trajectory based on regularity penalty strategy of the present invention.

[0063] Figure 2 This is a schematic diagram of flight operations within the air traffic network. Specific implementation methods

[0064] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0065] A specific embodiment of the present invention, such as Figure 1-2 As shown, a method for intelligent control of four-dimensional flight paths based on a regularized penalty strategy is disclosed, which specifically includes the following steps:

[0066] Step 1: Establish a model for the air traffic network flight control problem;

[0067] The air traffic network flight control problem model of the present invention can allocate a reasonable takeoff time for each flight.

[0068] Step 11, as follows Figure 2 As shown, a three-dimensional bounded region model based on the air traffic network is established, and auxiliary variables and decision variables are set for the model.

[0069] The three-dimensional bounded region model of the air traffic network is as follows:

[0070]

[0071] Where Υ represents the node set, which includes the airport node set and the sector node set; L represents the set of flight routes, which includes the flight paths of all flights; L represents the set of flight altitude layers.

[0072] The expressions for the auxiliary variables of the model are:

[0073]

[0074] in, This indicates the state of flight f arriving at node k at time t with altitude layer l.

[0075] The expression for the decision variables of the model is:

[0076]

[0077] in, This indicates that flight f has arrived at node k at altitude level l before time t.

[0078] Step 12: Based on the auxiliary and decision variables of the model, construct a model for the flight control problem in the air traffic network;

[0079] The air traffic network flight control problem model is a biobjective integer programming model, and its expression is:

[0080]

[0081] Where x is the flight's scheduled takeoff time; C(x) is the operational efficiency cost; and E(x) represents the fairness of airline delay allocation.

[0082] A bi-objective integer programming model is used to obtain the scheduled takeoff times for flights.

[0083] Optionally, operational efficiency costs can be obtained based on ground waiting costs and in-flight waiting costs.

[0084] Furthermore, the ground waiting cost for flights is in, G represents the ground delay cost per unit time for flight f; f This indicates the ground delay time for flight f.

[0085] Furthermore, the expression for the ground delay time of a single flight is:

[0086]

[0087] in, Indicates the scheduled departure time of flight f; Let f represent the set of feasible flight times for flight f at node k; k = δ(f, 1) represents the departure airport of flight f. This indicates the altitude layer l of flight f at time t relative to ground level. g The state upon reaching node k; This indicates that flight f had already reached altitude layer l at ground time before time t. g The state upon reaching node k; This indicates that flight f had already reached altitude l at ground time before time t-1. g The state upon reaching node k.

[0088] It is understandable that node k includes airport nodes and sector nodes.

[0089] Specifically, the state δ(f,s) of the node where flight f is located on the flight path is:

[0090]

[0091] Among them, S f This represents the total number of nodes along the flight path of flight f.

[0092] Furthermore, the cost of waiting in the air for flights is in, This represents the cost of flight f per unit of time in-flight delay; a f This indicates the in-flight delay time for flight f.

[0093] Furthermore, the expression for the in-flight delay time of a single flight is:

[0094]

[0095] in, Indicates the scheduled landing time of flight f; This indicates that flight f is at the airport where it is landing.

[0096] Furthermore, the expression for operating efficiency cost is:

[0097]

[0098] Wherein, Ω represents the flight set.

[0099] Optionally, airline delay allocation fairness can be obtained based on the delay costs of all flights and the average delay cost.

[0100] Furthermore, belonging to airline a pThe expression for the delay cost of all flights is:

[0101]

[0102] in, Indicates belonging to airline a p All flights.

[0103] Furthermore, belonging to airline a p The expression for the average delay cost of all flights is:

[0104]

[0105] in, Indicates belonging to airline a p The number of flights.

[0106] Furthermore, the expression for the fairness of airline delay allocation is:

[0107]

[0108] Where A represents the group of airlines; and These respectively indicate that they belong to airline a p and a q The average cost of delays for all flights; A This indicates the number of airlines.

[0109] Furthermore, constraints are set for the bi-objective integer programming model; these constraints include restrictions on flight ground waiting time, flight in-flight waiting time, safe flight interval, node continuity, altitude layer uniqueness, altitude layer adjustment, altitude layer before landing, aircraft landing time, takeoff airport capacity, landing airport capacity, sector capacity, and time continuity.

[0110] Furthermore, the expression for the flight ground waiting time limit is:

[0111]

[0112] in, This indicates the maximum ground waiting time for flight f.

[0113] Understandably, ground waiting time restrictions require flights to take off within a certain range of their scheduled departure time.

[0114] Furthermore, the expression for the flight waiting time limit is:

[0115]

[0116] in, This indicates the maximum in-flight waiting time for flight f.

[0117] Furthermore, the expression for the safe flight interval limit is:

[0118]

[0119] in, This represents the shortest horizontal safe distance between any two flights; This represents the minimum vertical safe distance between any two flights; Indicates the time when flight f entered the sector; Indicates the altitude at which flight f enters the sector; This indicates that flight f has arrived at node k at altitude layer l before time t; This indicates that flight f has arrived at node k at altitude l before time t-1.

[0120] Understandably, the safe flight interval limit constrains sector nodes.

[0121] Furthermore, the expression for the node continuity constraint is:

[0122]

[0123] in, Represents the set of feasible altitude layers for flight f within node k; l * This indicates that flight f is downstream of node k. * Height level; This indicates that flight f is downstream of node k. * The set of feasible height layers within; This indicates that flight f has arrived at node k at altitude layer l before time t; This indicates that flight f had already reached altitude l before time t. * Reaching downstream node k * The state; Let f represent the set of feasible flight times for flight f within node k; δ(f,s) represents the state of flight f at the s-th node on the flight path. This represents the flight time required for flight f to traverse node k; Let f represent the set of feasible altitude layers for flight f within node k.

[0124] Understandably, the node continuity constraint restricts all flights (at all altitudes) passing through the node. (The duration must be specified.) Then all of them enter their adjacent downstream nodes (all height layers) In node continuity constraints, the constraints apply to sector nodes.

[0125] Furthermore, the expression for the uniqueness constraint of the height layer is:

[0126]

[0127] Among them, R f Let R represent the set of nodes along the flight path of flight f. f ={δ(f,s)|1≤s≤S f}

[0128] It is understandable that the uniqueness constraint of altitude layer restricts the existence of multiple feasible altitude layers within a node, but a certain flight can only fly at one altitude layer within the node at any given time.

[0129] Furthermore, the expression for the height layer adjustment limit is:

[0130]

[0131] in, This represents the maximum possible altitude adjustment range for flight f within node k.

[0132] It is understandable that the altitude layer adjustment constraint restricts any flight to fly at altitude layer l at upstream node k. After a certain duration, a feasible height layer must be reached. * Enter its adjacent downstream node k * In the height layer adjustment constraints, the sector nodes are constrained.

[0133] Furthermore, the height layer l of the upstream node and the height layer l of the downstream node * Between satisfy

[0134] Furthermore, the expression for the altitude layer limit before aircraft landing is:

[0135]

[0136] Where, k = δ(f,S) f -1) represents the last sector node before flight f lands at the airport, k = δ(f,S) f () indicates that flight f is landing at the airport.

[0137] Understandably, the altitude restrictions before landing require flights to pass through the last sector at a lower altitude during cruise to ensure that the flight can descend and land normally.

[0138] Furthermore, the expression for the aircraft landing time limit is:

[0139]

[0140] in, This represents the latest feasible flight time of flight f at node k;

[0141] Understandably, the aircraft landing time constraint requires all flights to land before the upper bound of the set of feasible times at the landing airport.

[0142] Furthermore, the expression for the takeoff airport capacity limit is:

[0143]

[0144] Where V represents the airport set, H k (t) represents the departure capacity of the takeoff airport node k at time t; T represents the set of flight times.

[0145] Understandably, when an airport is used as a departure airport, the number of flights taking off at any given time cannot exceed H. k (t).

[0146] Furthermore, the expression for the landing airport capacity limit is:

[0147]

[0148] Among them, B k (t) represents the arrival capacity of landing airport node k at time t.

[0149] Understandably, when an airport is used as a landing airport, the number of landing flights at any given time cannot exceed B. k (t).

[0150] Furthermore, the expression for the sector capacity limit is:

[0151]

[0152] Among them, Z k (t) represents the capacity of sector node k; E represents the sector set.

[0153] Understandably, the sector capacity limit restricts the number of flights within sector node k to not exceed Z at any given time. k (t).

[0154] Furthermore, the expression for the time continuity constraint is:

[0155]

[0156] It is understandable that if flight f has already reached upstream node k at altitude l before time t-1, then it must have reached upstream node k at altitude l before time t.

[0157] Furthermore, the expression for the constraint on the decision variables is:

[0158]

[0159] Understandably, the decision variables are binary variables.

[0160] Step 2: Design a regularized penalty control strategy for the air traffic network;

[0161] Step 21: Quantify the causal relationship of delays for any airport using transitive entropy, and establish a weighted causal relationship network for delays, expressed as:

[0162] G * =(V,E) * ,W) (24)

[0163] Where V represents the set of airport nodes; E * Let W represent the set of flight routes; let W represent the set of m-order transit entropy between any two airports.

[0164] Specifically, the set of m-order transit entropies between any two airports is obtained from the set of m-order transit entropies between all any two airports, expressed as:

[0165]

[0166] in, Airport v i airport v j m-order transfer entropy; Airport v i The historical delay value at time t; Airport v j The historical delay value at time t+1; Airport v j The sequence of historical delay values ​​of length m; H(·) represents the information entropy operator; I(·) is the conditional mutual information operator; x1, x2 and x3 respectively represent Elements in;

[0167] Step 22: Perform a random walk in the delay causal weighted association network to obtain the delay penalty coefficient between any two airports;

[0168] Random walks are performed in the delayed causal weighted association network to obtain a delayed propagation path of length Q.

[0169] By adjusting the number of airports affected by delay propagation in a random walk of length Q, the spatiotemporal network of delay propagation is obtained, expressed as:

[0170] G cau =(V,E) cau ,D); (26)

[0171] Among them, E cau Let represent the set of propagation paths of delays; D represents the set of propagation delay values ​​for airports.

[0172] Specifically, the propagation delay values ​​of all airports constitute the set of airport propagation delay values, and the expression for the propagation delay value of each airport is as follows:

[0173]

[0174] in, Airport v i The propagation delay value, specifically, represents the delay propagation path of airport v. i upstream airport v i-1 The ripple effects of its delays Indicates upstream airport v i-1 The propagation delay value, Airport v i The absorption delay value, Indicates upstream airport v i-1 airport v i The m-th order transitive entropy.

[0175] The delay penalty coefficient between any two airports can be obtained based on the delay propagation spatiotemporal network.

[0176]

[0177] in, Airport v i With airport v j Delay penalty coefficient between Airport v i With airport v j The set of all delayed propagation paths between them; Indicates upstream airport v i-1 With airport v i The shortest distance between them.

[0178] It is understandable that the delay penalty coefficient between any two airports is the regularized penalty control strategy of the air traffic network.

[0179] Step 3: Adjust the flight path based on the regularization penalty strategy. The specific steps are as follows:

[0180] Step 31: Based on the regularized penalty control strategy of the air traffic network in Step 2 (i.e., the delay penalty coefficient between any two airports), generate a set of feasible takeoff times for flights.

[0181] Based on the delay penalty coefficient between any two airports in step 22, the feasible departure time for each flight is obtained, expressed as follows:

[0182]

[0183] Step 32: Sort the feasible departure times of each flight in ascending order to obtain a set of feasible departure times.

[0184] Step 33: Arrange all flights in the feasible departure time set according to their control order, encode the chromosomes of the flights, and obtain the initial population. The expression is:

[0185] Step 34: Based on the bi-objective integer programming model in Step 1, use a greedy strategy to process each chromosome in the initial population in turn to obtain feasible solutions for the chromosomes; put the feasible solutions for the chromosomes into the pre-trained encoding library.

[0186] It is understandable that the feasible solution for each chromosome is the efficiency and fairness target value of the combination of control order for each flight.

[0187] Step 35: Using a hash table, generate the optimal feasible solution for each chromosome by determining the search efficiency and fairness target values ​​for each chromosome in the pre-trained encoding library; based on the optimal feasible solution for each chromosome, obtain the dominance level and crowding degree of each chromosome.

[0188] Step 36: Based on the dominance level and crowding of all chromosomes, apply an elite strategy to obtain the parent population.

[0189] Specifically, all chromosomes are sorted by dominance level and crowding degree, respectively. If the total number of chromosomes is less than [a certain value], [the following is considered]. Select all chromosomes as the parent population if the total number of chromosomes is greater than 1. Sort by crowding level, and remove individuals with high crowding levels in turn, then select... There are 3 chromosomes, of which N p The total size of the parent generation.

[0190] Step 37: Perform selection and gene neighbor shift genetic operations on the parent population to generate the offspring population; merge and iterate the parent and offspring populations to generate a new generation population with a size of N. p .

[0191] Step 38: Determine whether the generated new generation population meets the convergence condition. If the convergence condition is met, obtain the optimal control takeoff time for each flight and proceed to step 39. If not, return to step 35.

[0192] Step 39: Obtain the four-dimensional control trajectory based on the optimal control takeoff time for each flight.

[0193] Among them, the four dimensions refer to longitude, latitude, altitude, and time.

[0194] The method of this invention, given the air traffic network topology and the time period to be regulated, and considering factors such as flight safety intervals and airspace unit capacity constraints, is based on controlled sectors. By adopting air traffic flow management procedures, it optimizes the takeoff time, sector crossing time, landing time, and cruising altitude of all flights within the time period to be regulated, thereby obtaining the optimal spatiotemporal distribution of any aircraft in the air traffic network. This ensures that air traffic flow flows into or through the corresponding airspace in the best possible way, achieving a match between airspace spatiotemporal resources and flight flow demand, promoting a balance between traffic demand and airspace capacity, and thus ensuring the safe and efficient operation of the air traffic network.

[0195] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Any person skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features within the scope of the technology disclosed in the present invention; and such modifications or substitutions should all be covered within the protection scope of the present invention.

Claims

1. A flight four-dimensional trajectory intelligent regulation method based on a regular penalty strategy, characterized in that, Specifically comprising the following steps: Step 1, establishing an air traffic network flight regulation problem model; Step 2, designing a regular penalty regulation strategy for the air traffic network, the specific steps being: Step 21, using the transfer entropy to quantify the delay causal correlation between any two airports, and establishing a delay causal weighting correlation network; Step 22, performing random walk in the delay causal weighting correlation network to obtain the delay penalty coefficient between any two airports; Step 3, regulating the four-dimensional flight path of the flight based on the air traffic network flight regulation problem model of step 1 and the regular penalty strategy of step 2; In step 21, the specific steps for using the transfer entropy to quantify the delay causal correlation between any two airports and establishing a delay causal weighting correlation network are: establishing the mutual transfer entropy between any two airports m the expression for the transfer entropy is in, Airport airport of m Order of propagation entropy; Airport In the Historical delay values ​​at any given moment; Airport In the Historical delay values ​​at any given moment; Airport The length is Historical delay numerical sequence; Represents the information entropy operator; For conditional mutual information operators; , and They represent , , Elements in; all pairs of airports m a set of order transfer entropies between all pairs of airports m a set of order transfer entropies between all pairs of airports Using the transfer entropy to quantify the delay causal correlation between any two airports and establishing a delay causal weighting correlation network, the expression being: wherein, V denotes a set of airport nodes; denotes a set of routes; W denotes a set of m orders of transfer entropy sets; In step 22, the specific steps for performing random walk in the delay causal weighting correlation network to obtain the delay penalty coefficient between any two airports are: In the random walk of the delay causal empowerment association network, the length of the random walk is obtained Q delay propagation path; The length of the random walk is adjusted as Q The number of airports affected by the delay propagation in the delay propagation path of the flight is obtained, and the delay propagation space-time network is obtained, and the expression is as follows: ； wherein, denotes a set of propagation delay paths; D denotes a set of propagation delay values for the airport; The propagation delay values of all airports form a set of airport propagation delay values, and the expression of the propagation delay value of each airport is: wherein, denotes the propagation delay value of the airport ; denotes the propagation delay value of the upstream airport , denotes the absorption delay value of the airport ; denotes the absorption delay value of the upstream airport to the airport ; m denotes the k-th transfer entropy of the airport According to the delay propagation spatiotemporal network, the delay penalty coefficient between any two airports is obtained; wherein, denotes the airport with the airport a delay penalty coefficient between, denotes the airport with the airport a set of all delay propagation paths between; denotes the shortest distance between the upstream airport with the airport . 2.The method according to claim 1, wherein, The specific steps for establishing the air traffic network flight regulation problem model in step 1 are: Step 11, based on the three-dimensional bounded region model of the air traffic network, setting the auxiliary variables and decision variables of the model; The expression of the auxiliary variable of the model is: wherein, representing a flight at time at altitude arrival node k the state; The expression of the decision variable of the model is: wherein, representing a flight at time has reached a level of the node k ; Step 12, based on the auxiliary variables and decision variables of the model, constructing the air traffic network flight regulation problem model, the expression being: wherein, is a flight's assigned control departure time; is a running efficiency cost; is a fairness of delay assignment to an airline. 3.The method according to claim 2, wherein, Based on the flight ground waiting cost and the flight air waiting cost, the operation efficiency cost is obtained. 4.The method of claim 2, wherein, Based on the delay cost of all flights and the average delay cost, the airline delay allocation fairness is obtained.

5. The method of claim 2, wherein, In step 1, when establishing the air traffic network flight regulation problem model, the constraint conditions of the air traffic network flight regulation problem model are set; the constraint conditions include flight ground waiting time limit, flight air waiting time limit, safety flight interval limit, node continuity limit, height layer uniqueness limit, height layer adjustment limit, height layer limit before aircraft landing, aircraft landing time limit, takeoff airport capacity limit, landing airport capacity limit, sector capacity limit, and time continuity limit. 6.The method of claim 1, wherein, In step 3, the specific steps for regulating the four-dimensional flight path of the flight based on the air traffic network flight regulation problem model of step 1 and the regular penalty strategy involved in step 2 are: Step 31, based on the regular penalty regulation strategy of the air traffic network in step 2, generating a feasible takeoff time set of the flight; Step 32, arranging the flights in the order of regulation of all flights in the feasible takeoff time set to perform chromosome coding and obtain an initial population; Step 33, based on the double-objective integer programming model in step 1, using a greedy strategy to process each chromosome in the initial population in turn to obtain a feasible solution of the chromosome; Placing the feasible solution of the chromosome into a pre-trained code library; Step 34, find the efficiency and fairness target value of each chromosome in the pre-training encoding library using the hash table, generate the optimal feasible solution of each chromosome; obtain the dominance level and crowding degree of each chromosome based on the optimal feasible solution of each chromosome; Step 35, based on the dominance level and crowding degree of all chromosomes, apply the elite strategy to obtain the parent population; Step 36, implement genetic operations of selection and gene neighborhood movement on the parent population respectively to generate the offspring population; merge the iteration parent population and the offspring population to generate the new generation population; Step 37, judge whether the generated new generation population meets the convergence condition, if it meets the convergence condition, obtain the optimal control takeoff time of each flight, enter step 38, if it does not meet, return to step 34; Step 38, obtain the four-dimensional control flight path according to the optimal control takeoff time of each flight.

7. The method of claim 6, wherein, The feasible solution of each chromosome is the efficiency and fairness target value of the control order combination of each flight. 8.The method of claim 6 or 7, wherein, Based on the delay penalty coefficient between any two airports in step 22, obtain the feasible takeoff time of each flight, the expression is: wherein, represents a planned departure time of a flight; represents a maximum ground waiting duration of a flight; represents a set of feasible flight times of a flight at a node; represents a set of flights; represents a delay penalty coefficient between an airport and an airport; represents a delay penalty coefficient between an airport and an airport;​​​​​ Sort the feasible takeoff time of each flight in the order from short to long to obtain the feasible takeoff time set.

Images