CN112949978A

CN112949978A - Emergency reserve landing field selection method based on collaborative optimization

Info

Publication number: CN112949978A
Application number: CN202110098879.8A
Authority: CN
Inventors: 赵向领; 任强; 闫凤良
Original assignee: Civil Aviation University of China
Current assignee: Civil Aviation University of China
Priority date: 2021-01-25
Filing date: 2021-01-25
Publication date: 2021-06-11
Anticipated expiration: 2041-01-25
Also published as: WO2022156797A1; CN112949978B

Abstract

An emergency landing reserve selection method based on collaborative optimization analyzes from the operation angle of an airline company and combines related actual conditions, factors such as available residual flight time of flights, expectation to a landing reserve airport, landing reserve capacity of the airport, acceptable airplane categories, flight arrival time slots allocated by the landing reserve, cruising speed of the airplane, road wind and air traffic jam and the like are comprehensively considered to design constraint conditions, and a linear planning model of a centralized landing reserve problem is established based on the idea of 'collaborative decision', is used for coordinating and allocating the flights needing emergency landing reserve to be temporarily closed at a target airport and transferred to a proper landing reserve airport, so that total flight change reserve time is reduced to the maximum extent, and the method has the characteristic of flexible centralized optimization allocation, and meets the expectation of the airline company to the landing reserve airport to the maximum extent.

Description

Emergency reserve landing field selection method based on collaborative optimization

Technical Field

The invention belongs to the technical field of flight operation scheduling, and particularly relates to a problem of changing flight reserve and selecting when a flight destination airport is closed due to uncertain factors in a flight operation process.

Background

With the rapid development of civil aviation industry, the number of flights operated by a main hub airport in peak time period is more than that of the flights operated at any time in the past, and the operation capacity of the main hub airport is in the edge of saturation for a long time. Under the influence of factors such as extreme weather, invasion of foreign objects in an airport, terrorist threats and the like, the normal operation of the airport can be interrupted, even temporary closure is caused, and a large number of affected flights must be immediately redirected to a landing reserve airport. A wide range of flight reservations not only interferes with the proper operation of the respective reserve airport, but may even interfere with the operation of the entire air traffic network. Therefore, when a terminal airport suddenly stops, it is a very critical task to deal with the process of redirecting an in-flight to an appropriate landing reserve.

There are three main types of current research that are associated with temporary closure of airports:

(1) aiming at analyzing the influence of closing of different airports, scholars provide a strategy for redirecting flights to an undisturbed terminal airport, so as to analyze the influence on the passenger of a helicopter; the goal is to properly schedule passengers for affected departure flights and to minimize the travel time of passengers and the waiting time for transit flights to the back-up airport. The influence of flight preparation and landing on the distribution of the traffic flow in the air, the load of control command and the like is ignored in the processing process, so the idealization degree is higher.

(2) A method for determining the route of a landing flight, which allocates the time for the flight to arrive at the landing airport, aims to reduce the waiting and re-flying time as much as possible and comprehensively considers the diversion time window, the limit of the residual fuel and the limit of the airport capacity. Under the condition of reasonably assuming a diversion airport and transportation capacity, the flight can be quickly backed down without reaching the fuel critical state. This approach, while reducing the total reserve flight time for flights, airlines lose the desired reserve airport in flight planning and require more time and cost to recover from the reserve.

(3) When a normal flight schedule is interrupted due to a destination airport closure, a concern is how to quickly resume airline and airport operations. The better processing effect in the existing research is a method for rearranging flight time of 'substitute pair', and the method discloses the existence of 'substitute pair' in an airport network and provides an identification method. Typically the alternate pair is a pair of airports, if one is closed, the other can take up part of its traffic load, similar to a special landing airport. Although the flight time is subdivided and an alternative pair method is proposed in the research, the research scope is strategic and the local influence of flight reserve is ignored.

Disclosure of Invention

The invention overcomes the defects of the existing research, establishes a linear programming model of the concentrated landing reserve problem based on the idea of 'cooperative decision', is used for coordinating and distributing the flights needing emergency landing reserve temporarily closed at a target airport, transfers the flights to a proper landing reserve airport, reduces the total flight change landing reserve time to the maximum extent, has the characteristic of flexible concentrated optimization distribution, and meets the expectation of an airline company on the landing reserve airport to the maximum extent.

The invention analyzes from the operation angle of an airline company and combines with relevant actual conditions, comprehensively considers the available residual flight time of flights, the expectation of a landing reserve airport, the landing reserve capacity of the airport, the acceptable airplane category, the flight arrival time slot allocated to the landing reserve airport, the cruising speed of the airplane, the road wind, the air traffic jam and other factors and designs constraint conditions, and provides the following technical scheme:

the emergency landing preparation field selection method based on the collaborative optimization comprises the following steps:

(1) after the destination airport is closed, collecting flight plan information of all flights affected by the closing of the destination airport at the first time, and determining the number of affected flights, the number of available landing reserve airports and expected landing reserve airports of all the flights;

(2) determining the real-time position, speed and airway wind speed of the affected flight, and the airplane category of the affected flight;

(3) determining the residual aircraft oil amount of each affected flight to obtain available residual flight time;

(4) determining time slots which can be used for standby landing and airport capacity of each standby landing airport according to the standby landing airports involved in the flight plan, and allocating landing time for the affected flights;

(5) establishing an integer programming model of flight reserve for reducing total flight reserve time to the maximum extent and meeting the expectation of an airline company on a reserve airport;

(6) solving the integer programming model of the flight reserve landing to obtain a flight reserve landing distribution result;

(7) and issuing a standby landing scheme according to the distribution result, and enabling each affected flight to go to a corresponding standby landing airport for standby landing according to the standby landing scheme.

The integer planning model for flight preparation for landing in the step (5) specifically comprises:

an objective function:

six constraint conditions:

(21) each flight is allocated to at most one time slot of an airport, and each time slot can be allocated to at most one flight, as follows:

(22) the time slot s time allocated by the standby landing must be less than the remaining flight time T of the flight_iExpressed as:

(23) the flight i has a flight path flight time from the diversion point to the standby airport a that is earlier than the allocated time slot s of the airport a, and is expressed as:

(24) the total number of flights to the reserve airport a should not exceed its acceptable total number of flights, i.e. capacity Q_aExpressed as:

(25) airplane category restrictions for airports, expressed as:

wherein min represents the minimum value, beta is a weighting coefficient, beta is more than or equal to 0 and less than or equal to 1, i represents a flight index, N represents the total flight quantity needing to be spared, a represents an available sparing yard index, M represents the number of available sparing airports, S represents an optional sparing time slot provided by the airport, and S represents the number of available sparing airports_aIndicating the amount of time slots available for reserve flights in airport a in the future, ST_asShowing the time of the s-th time slot of the a airport, BT showing the time of the sudden closing of the target airport, T_iResidual flight time, x, obtained from residual fuel for flight i_iasFor decision variables, x is the time slot s of the airport a when flight i is allocated_iasThe value is 1, the other values are 0, P_iaThe expectation value of a reserve airport representing that the airline selects a as flight i, BM being much greater than T_iNumerical value of (D)_iaDistance of flight i to airport a, V_iSelected cruise speed, Wind, for flight i_iaMean wind speed, λ, representing the flight course of flight i to the airport a_iaAs a correction factor, C_iIs the model code classification of flight i, C_aIs the airplane model code with the largest acceptable weight for the airport a.

Compared with the prior art, the invention has the advantages that:

(1) the optimization method is integrated into an air traffic flow management system, works when the primary hub airport is suddenly shut down, and then provides the best diversion solution to the airline operators, pilots, air traffic controllers and airport managers.

(2) The method can provide better reference for the recovery of the air transportation network and provide effective decision support for centralized flight diversion management.

(3) The model related to the invention considers the closed flight of the destination airport in emergency, and reduces the landing cost of the airline company to a certain extent by reducing the time for changing the flight and landing and meeting the expectation of the airline company on the landing airport to the greatest extent.

(4) The constraints involved in the model are: the remaining flight time and landing time slot, the capacity of the landing reserve airport, the influence of wind and air traffic delay, the acceptable airplane type of the airport and the like meet most requirements in actual operation.

(5) The time required in the model solving process is less than one second, and the emergency decision process can be met.

(6) The method combines the aim of saving the total time of changing the flight and landing reserve with the expectation of an airline company on a landing reserve airport, always has the functions of flexibly carrying out central control and optimization, can meet the requirements of the airline company, and reflects the CDM concept.

Drawings

Fig. 1 is a problem description diagram.

FIG. 2 is a graph of required time of flight versus available remaining time of flight. The numbers "1-50" of the outermost circles in the figure represent flight indices, the numerical values "0-180" of the radii of the concentric circles represent flight times, and larger radii represent longer flight times.

Fig. 3 shows the number of flights in different time windows. Including time to flight (ST)_asBT), flight time (using time), and air latency (Holding time), wherein the switch-off time comprises flight time and air latency.

Fig. 4 shows the assigned landing preparation expectation values and the maximum and minimum expectation values for different flights. The numbers "1-50" on the outermost circle in the figure represent flight index numbers, and the numerical values "0-1" on the radii of the concentric circles represent the flight's expected value for different landfills.

Fig. 5 is a diagram of meeting the expected number of different flight deplot. For 8 landing case, according to the difference of the expectation degree of each flight to each landing, each flight has an expectation sequence for 8 landing, the sequence is "1" to represent the largest expected landing, and so on, "8" to represent the smallest expected landing. In the solution of fig. 5 in this case, the data in the sector, such as "1, 23, 46%" indicates that 23 flights are allocated to the respective maximum expected drop-off, and the 23 flights account for 46% of the total flight number, i.e. 46% of the flights can be dropped off their maximum expected drop-off. )

FIG. 6 is a comparison graph of the method for minimizing time to take off and take off (50 flight cases) with the time to take off and take off of the present strategy.

Fig. 7 is a comparison graph of the method for minimizing the time to take the flight away and the expected value of the take-off site of the strategy (50 flight cases).

Fig. 8 shows the average expected descent value and the average time to change flight for different weight coefficients β.

Fig. 9 shows the algorithm execution time (51 test calculation times were counted in seconds).

Detailed Description

Before the flight runs, a flight plan must be submitted to an air traffic control unit, and in the flight plan, an airline dispatcher selects an appropriate landing reserve airport for the flight to deal with the situation that a destination airport cannot land. Under normal conditions, the flight can operate according to the flight plan content, and can normally land at the destination airport. However, when the destination airport is closed urgently, all flights to the destination must be re-routed to the destination airport for the destination airport. Under some practical constraints, it is necessary to assign N sets of flights to M (M < N) landed airports, as shown in fig. 1.

According to the invention, the allocation and selection of the reserve landing airport is carried out according to the following steps

(1) The flight plan information for all affected flights is collected the first time after the destination airport is closed, thereby determining the number of affected flights, the number of available drop-off airports and the expected drop-off airport for each flight.

(2) Determining real-time position parameters, speed, air speed of the aircraft and the category of the affected aircraft according to the existing air traffic control automation equipment.

(3) The pilot reports the remaining aircraft oil to obtain the remaining flight time available for each affected flight, the amount of which largely affects the allocation of the take-off airport.

(4) And determining available time slots for standby landing and airport capacity through coordination communication according to the standby landing airports involved in the flight plan, and allocating landing time for the flight.

(6) taking the flight and reserve landing place information collected in the steps (1) to (4) as model parameter data, solving the model by using the integer programming model established in the step (5) and adopting general integer programming model solving software (such as a CPLEX solver) to obtain a flight reserve landing distribution result;

(7) and the control unit issues a preparation landing scheme according to the distribution result, and each flight goes to a corresponding preparation landing place for preparation landing according to the preparation landing scheme.

Wherein the detailed model in (5) is as follows:

first, the decision variables are established

Wherein:

i represents a flight index;

n represents the total number of flights that need to be spared;

a represents the available touchdown field index;

m represents the number of available airports to be landed;

s represents an optional backup timeslot provided by the airport;

S_arepresenting a future period of airportThe amount of time slots available in the bay for the reserve flight.

Second, establishing an objective function

Target 1:

target 2:

where min represents x for the decision variable_iasIs calculated as a minimum value, max represents x for the decision variable_iasThe function of (2) is maximized. Equation (2) represents minimizing the total turn-by-turn landing time. BT denotes the moment when the target airport suddenly closes, i.e. the moment when all flights needing to take hold start to perform a hold-down. ST (ST)_asIndicating the time of the s-th slot of the a-field. Thus, if flight i selects airport a and is waiting at that airport's time, x_iasIf 1, the flight flies to the airport, waits overhead, and is at ST_asThe time required for landing is ST_asBT, including air flight time and air latency. Equation (3) represents the overall desire to maximize airline operational flexibility. P_iaIs the expected value of a reserve landing airport with an airport a selected by an airline company as a flight i, and P is more than or equal to 0_iaNot more than 1, and P_iaThe larger the value, the more desirable the airline is to move to the airport for landing.

And thirdly, weighting the two objective functions by a weighting coefficient beta (beta is more than or equal to 0 and less than or equal to 1) to obtain a formula (4).

Min represents the minimum value calculation.

The formula is used for ensuring that the total time for changing the aircraft to a landing is minimum, and simultaneously ensuring that the total expectation of selecting the landing site by an airline company is maximum. Wherein, T_iObtaining a remaining flight time for flight i from the remaining fuel;

for ensuring that the order of magnitude is the same in both objective functions;

a penalty function indicating that if flight i does not select any alternate landing airport alternate landing. If flight i is assigned to a landing airport, the function value is 0, otherwise the value is BM, which is a value far greater than T_iThe numerical value of (c). The penalty function can enable the model to solve the flight landing preparation situation in an emergency situation, namely, if the re-sailing flight cannot follow the given constraint, the model can still give a landing preparation scheme;

the fourth step, establish mathematical constraint

(1) Each flight is allocated to at most one time slot of an airport, and each time slot can be allocated to at most one flight, i.e. inequality constraints (5) and (6).

(2) Time slot window constraint

Since the remaining fuel in the tank of flight i is limited, the time required for changing flight and landing is from the point of change to landing airport a and waiting at the airport to land at the time of the allocated time slot s, therefore, the time slot s allocated for landing must be less than the remaining flight time T_iI.e. inequality (7).

Meanwhile, the time slot allocated by the standby landing place a for the flight i is later than the time when the airplane flies to the airport, otherwise, the time slot needs to be allocated again after the airplane flies to the airport. That is, the flight i has a route flight time from the re-routing point to the standby airport a that is earlier than the allocated time slot s of the airport a, as shown in equation (8).

Wherein

D_iaThe distance from flight i to the depot a;

V_iis the selected cruising speed;

Wind_iathe average wind speed representing the flight course of flight i to airport a, which can be predicted using the high altitude wind temperature map;

λ_iafor correcting factors, the method is used for considering factors which can cause flight delay such as external air temperature, unpredictable wind direction change, navigation error, air traffic jam and the like, wherein 1 is less than or equal to lambda_ij≤1.1；

(3) Capacity limitation at landing reserve airport

The total number of flights to the reserve airport a should not exceed its acceptable total number of flights, i.e. capacity Q_a。

(4) Airplane class restriction at an airport

According to the airport operation guarantee, airplane models can be classified into 6 types, namely { A, B, C, D, E and F }. Generally, the category 6 is larger and heavier from left to right, and if an airport is suitable for heavy-duty airplane operation, the airplane with smaller weight can be accepted, for example, if the airport can receive the D type airplane at maximum, the accepted airplane type is four categories { A, B, C and D }. For ease of comparison, this set of reference codes may be mapped to a set of

numbers

1, 2, 3, 4, 5, 6. C_iIs a classification code C of flight i_i∈{1、2、3、4、5、6}。C_aIs the aircraft model code with the maximum weight acceptable for airport a, C_aE {1, 2, 3, 4, 5, 6 }; thus, if flight i decides to prepare for landing at airport a, it must beSatisfies C_i≤C_aI.e. inequality (10).

Examples

Assume that 50 flights need to be redirected to 8 landed airports, with the specific case parameters as follows.

The test file information is as follows: the file name read is: 3d _50_8_20_3DJun 262019 result txt

Number of flights to be refunded: 50

The number of optional reserve landing fields: 8

The number of slots per settled field is: 20

It is assumed that flight reserve time is as important as the reserve ground expects two targets, i.e., in equation (4), the two targets are weighted equally (β is 1- β, so β is 0.5). And (3) typing the model in a Cplex solver, importing case parameters, and solving through the software to obtain an allocation result.

The test resulted in an average take-off time of 59.36 minutes for 50 flights, with a route flight time and latency of 53.53 and 5.83 minutes, respectively.

FIG. 2 shows the time required to fly for each flight and the remaining flight time T available in the distribution results_iComparison of (1). From FIG. 2, T can be seen_iAll in excess of 120 minutes, while the time required for the assigned flight is almost all less than 120 minutes (only flights 24 and 40 are exceeded), and most less than 60 minutes.

FIG. 3 shows the number of flights for different flight time intervalsCounting the time ST required for changing flight to reserve flight_as-BT less than 60 minutes with 33 racks (15+18 ═ 33), accounting for 66% of total flights; the flight volume requiring less than 120 minutes of flight change has 48 frames (15+18+7+4+4+0 is 48), accounting for 98% of the total number of flights. The 50 flights had 37 flights (12+ 25: 37) with a flight time of less than 60 minutes, accounting for 72% of the total flight volume. The flight has 40 air waiting times less than 10 minutes (29+11 equals 40), accounting for 80% of the total flight volume. According to the current operating regulation (CCAR-121R5), the last fuel reserve of the flight waiting in the air of the landing depot can ensure that the flight waits for 30 minutes. The total waiting time in the distribution results is less than 30 minutes and 80% of the flight waiting time is less than 10 minutes, so that the regulatory requirements are met and the air waiting time and fuel consumption can be saved.

FIG. 4 shows the expected value of the flight drop-off airport and the maximum expected value and the minimum expected value of the corresponding drop-off airport of the flight in the distribution result. The average value of 50 flights in the test is 0.888, and the expected value of the back-off airport distributed by most flights is close to the maximum expected value as can be seen from figure 4.

Fig. 5 shows the ordering of flight expectations for flights flying to all of the landed airports. Flight number assigned to the 1 st expected drop-off yard has 23, accounting for 46% of the total flight number ("1, 23, 46%" in the figure); the flight number of the first three expected landing airports is allocated to 39, which accounts for 78% of the total flight number (in the figure, "1, 23, 46%", "2, 9, 18%", "3, 7, 14%"), so that most flights can obtain the expected landing airports.

In the formula (4), two targets (hereinafter referred to as "dual target models") that the total time for flight change and standby landing is shortest and the expectation value of the airline to the standby landing site is maximum are considered as β ═ 0.5. If only the minimum line-changing standby time is considered and the desired target of the standby space is not considered, β in the formula (4) may be 1. The different results of the two cases of comparing β -0.5 with β -1 are as follows.

After comparing the dual-target model (β ═ 0.5) with the method targeting the shortest turn-off standby time (β ═ 1), it is found that the dual-target model consumes a little more flight time, but can largely meet the expectation of the airline company for the standby airport. In one aspect, the average switch preparation fall time increased from 53.64 minutes to 59.36 minutes, which increased by 10.66% (as shown in the boxplot of FIG. 6). The total reshuffling time for 19 (38%) flights in 0.5 is greater than 1, with an average increase in flight time of 5.72 minutes per flight, but the maximum flight time is the same (time per flight vs. bar chart of fig. 6). On the other hand, the expected average increases from 0.524 to 0.704 by 34.35% (as shown in the boxplot of FIG. 7), with an expected increase for 18 flights. For β -1, there are only 16 flights expected to be greater than 0.8, 26 flights expected to be greater than 0.6, and 28 and 39 flights, respectively, for β -0.5. Therefore, most flights can be redirected to their intended take-off airport, which will bring much convenience to the airline.

To determine the influence of the weighting factor β, the two target weighting factors β are changed from 0 to 1 in steps of 0.02, and the correspondence between the total switch-off time and the desired value can be obtained, as shown in fig. 8. In the figure, the minimum average switch-off time is 53.64 minutes, where β is 1 and the average expected value is only 0.524. If the average switch-off time is 60 minutes, the average expected value will increase to more than 0.7 and 0.704, where β is 0.5. Thus, with an extended 6.35 minute flight time, an average increment of expected value of 0.18 can be achieved, an increase of 34.35%, which would provide great convenience to airlines, and most flights would have the opportunity to reroute to their intended take-off airport. If the average switch-back-off time exceeds 60 minutes, the expected value increases slowly, and the expected increase in benefit is small.

Fig. 9 counts the time required for the model to run, where a total of 51 trials were performed, the average test elapsed time was 0.456 seconds, the standard deviation was 0.024, and the decision time requirement was met in the case of the diversion decision.

The case study on the situations of 50 flights and 8 landing preparation airports shows that the landing preparation can be completed in the remaining flight time of each flight, and the calculation time of the landing preparation optimization scheme is not more than 1 second; in addition, the time required for preparing for landing of 33 flights (66%) is less than 60 minutes, and the time required for preparing for landing of 48 flights (66%) is less than 120 minutes, so that the flexibility is high, and the requirement of an airline company is met.

Those skilled in the art will appreciate that those matters not described in detail in the present specification are well known in the art.

Claims

1. The emergency descent control field selection method based on collaborative optimization is characterized by comprising the following steps of:

2. The collaborative optimization-based emergency drop-off yard selection method according to claim 1, wherein the integer planning model for flight drop-off in step (5) is specifically:

an objective function:

six constraint conditions:

(25) airplane category restrictions for airports, expressed as:

wherein min represents the minimum value, beta is a weighting coefficient, beta is more than or equal to 0 and less than or equal to 1, and i represents navigationClass index, N represents the total number of flights requiring a reserve, a represents the available reserve index, M represents the number of reserve airports available, S represents the optional reserve time slot provided by the airport, S_aIndicating the amount of time slots available for reserve flights in airport a in the future, ST_asShowing the time of the s-th time slot of the a airport, BT showing the time of the sudden closing of the target airport, T_iResidual flight time, x, obtained from residual fuel for flight i_iasFor decision variables, x is the time slot s of the airport a when flight i is allocated_iasThe value is 1, the other values are 0, P_iaThe expectation value of a reserve airport representing that the airline selects a as flight i, BM being much greater than T_iNumerical value of (D)_iaDistance of flight i to airport a, V_iSelected cruise speed, Wind, for flight i_iaMean wind speed, λ, representing the flight course of flight i to the airport a_iaAs a correction factor, C_iIs the model code classification of flight i, C_aIs the airplane model code with the largest acceptable weight for the airport a.

3. The method for selecting emergency landing place for emergency based on collaborative optimization according to claim 2, wherein the number M of landing places and the total number N of flights to be landed satisfy M < N.

4. The collaborative optimization-based emergency hedging yard selection method according to claim 2, wherein P is greater than or equal to 0_iaNot more than 1, and P_iaThe larger the value, the more desirable the airline is to move to the airport for landing.

5. The collaborative optimization-based emergency hedging yard selection method according to claim 2, wherein λ is 1 ≦ λ_ia≤1.1。

6. The method for selecting emergency hedging ground based on cooperative optimization according to claim 2, wherein β is 0.5.