Airline passenger itinerary recovery method, apparatus, device, and storage medium

Abstract

The application provides an airline passenger trip recovery method, device and equipment and a storage medium, and relates to the technical field of air operation. Flight information of an abnormal flight is acquired; based on a clustering method, neighboring airports of a flight departure airport are aggregated according to an air route space-time network to obtain a multi-airport group region; a first user trip recovery model corresponding to the abnormal flight is constructed according to the flight information and the multi-airport group region, and the first user trip recovery model is used to reflect the correlation between ticketing revenue, compensation fees and traffic costs from the flight departure airport to the neighboring airports; and flight resources are allocated for users of the abnormal flight according to the first user trip recovery model. Through the multi-airport group region, the user is considered to be rebooked to other airports in the multi-airport group region and traffic costs, and a user trip recovery model that is more in line with actual conditions can be established, and feasible user trip recovery strategies are increased.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air operation, and in particular to an airline passenger travel recovery method, device, equipment and storage medium. BACKGROUND

[0002] With the rapid development of air transportation business, when the flight plan of an airline cannot be executed according to the original plan due to weather, users, regulations, security and airline itself reasons, it is called flight plan disruption, and the flight affected by the disruption is called abnormal flight. Among them, the abnormal flight will bring inconvenience to the user, damage the interests of the user, and also damage the reputation and economic benefits of the airline. Therefore, when the disruption occurs, it is very important for the airline to recover the planned flight or user travel in a short time.

[0003] The recovery strategy for abnormal flights includes delaying or canceling flights, resource exchange, using reserved resources, empty flying or transit flying, and reallocating users, which are collectively referred to as airline disruption management (ADM). Due to the complexity of the recovery strategy and the flight recovery problem, the aircraft recovery problem, the crew recovery problem, and the user travel recovery problem, etc. Sub-problems involve complex resource requirements. The traditional research solves the above sub-problems in order, specifically: first, solve the flight recovery problem (SRP), repair the flight schedule; then solve the aircraft recovery problem (ARP), determine the delayed or canceled flights, and reallocate available aircraft to the flights; then solve the crew recovery problem (CRP), allocate the cabin and other available crews to the flights; finally, solve the passenger-travel recovery problem (PRP), and reallocate users to available flights.

[0004] In related technologies, the available flights for user reallocation are usually based on point-to-point route networks, and the allocated flights are single and lack flexibility. SUMMARY

[0005] The present application provides an airline passenger travel recovery method, device, equipment and storage medium to solve the problem of single allocation mode and lack of flexibility after flight disruption.

[0006] In a first aspect, the present application provides an airline passenger travel recovery method, comprising:

[0007] obtain flight information of the abnormal flight, the flight information including a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a ticket price, and a ticket sale quantity;

[0008] based on a clustering method, the neighboring airports of the flight departure airport are aggregated according to an air route space-time network to obtain a multi-airport group region, and the air route space-time network is a relationship network between air routes established between hub air routes and between hub air routes and non-hub air routes;

[0009] based on the flight information and the multi-airport group region, a first user itinerary recovery model corresponding to the abnormal flight is constructed, and the first user itinerary recovery model is used to reflect a correlation between ticket revenue and compensation fees and transportation costs from the flight departure airport to the neighboring airports;

[0010] based on the first user itinerary recovery model, flight resources are allocated for users of the abnormal flight.

[0011] In a possible implementation, based on the first user itinerary recovery model, flight resources are allocated for users of the abnormal flight, including: a preset algorithm is used to determine an optimal solution of the first user itinerary recovery model, the preset algorithm including a column generation algorithm and a large neighborhood search algorithm; and based on the optimal solution, flight resources are allocated for users of the abnormal flight.

[0012] In a possible implementation, the airline passenger itinerary recovery method further includes: based on the flight information, a flight type is determined, the flight type including a direct flight and a transfer flight; based on the flight information and the flight type, a second user itinerary recovery model is constructed, the second user itinerary recovery model being used to reflect a correlation between ticket revenue and compensation fees when the direct flight is converted into the transfer flight; and based on the second user itinerary recovery model, flight resources are allocated for users corresponding to the abnormal flight.

[0013] In a possible implementation, the method further includes: obtaining flight information of the target flight, the flight information of the target flight including a flight departure airport, a flight departure time, a flight arrival airport, and a flight arrival time; constructing a third user itinerary recovery model according to the flight information of the target flight and a function model, the third user itinerary recovery model being used to reflect a correlation between a profit and a ratio of flexible tickets when the target flight is an abnormal flight, the function model being constructed based on a fitting analysis method according to historical booking data and being used to describe a correlation between the ratio of flexible tickets and a flexible ticket price ratio, wherein the flexible ticket price is less than a specific ticket price, and both a delay compensation and a downgrading compensation of a user corresponding to the flexible ticket are 0, and the flexible ticket corresponding to the user is refunded when the user refuses to board; determining an influence coefficient of the ratio of flexible tickets and the flexible ticket price ratio on the recovery strategy; and determining the ratio of flexible tickets and the flexible ticket price ratio corresponding to the target flight based on the influence coefficient corresponding to different ratios of flexible ticket data and flexible ticket prices.

[0014] In a possible implementation, the method further includes: determining the influence coefficient of the ratio of flexible tickets and the flexible ticket price ratio on the recovery strategy based on a sensitivity analysis method.

[0015] In a possible implementation, the method further includes: combining at least two of the first user itinerary recovery model, the second user itinerary recovery model, and the third user itinerary recovery model to obtain a plurality of fourth user itinerary recovery models; and for each fourth user itinerary recovery model in the plurality of fourth user itinerary recovery models, allocating flight resources for users corresponding to abnormal flights according to the fourth user itinerary recovery model.

[0016] In a possible implementation, the airline passenger itinerary recovery method is constructed by: obtaining historical flight data, the historical flight data including a flight number, a tail number, a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a cabin, a number of tickets sold, and ticket price data; performing data preprocessing on the historical flight data to obtain target flight data, the data preprocessing being used to remove abnormal data, the abnormal data including flight data with a small number of flight frequencies between two airports; and constructing the airline space-time network according to the target flight data.

[0017] In a second aspect, the present application provides an airline passenger itinerary recovery device, including:

[0018] The acquisition module is configured to acquire flight information of the abnormal flight, the flight information including a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a ticket price, and a ticket sales quantity.

[0019] The aggregation module is configured to aggregate, based on a clustering method, adjacent airports of the flight departure airport according to an air route space-time network to obtain a multi-airport group region, the air route space-time network being a relationship network between air routes established between hub air routes and between hub air routes and non-hub air routes.

[0020] The construction module is configured to construct, according to the flight information and the multi-airport group region, a first user itinerary recovery model corresponding to the abnormal flight, the first user itinerary recovery model being configured to reflect an association between ticket sales revenue and compensation fees and transportation fees from the flight departure airport to the adjacent airports.

[0021] The recovery module is configured to allocate, according to the first user itinerary recovery model, flight resources for users of the abnormal flight.

[0022] In a possible implementation, the recovery module can be specifically configured to: determine an optimal solution of the first user itinerary recovery model by using a preset algorithm, the preset algorithm including a column generation algorithm and a large neighborhood search algorithm; and allocate, according to the optimal solution, flight resources for the users of the abnormal flight.

[0023] In a possible implementation, the construction module can be further configured to: determine, according to the flight information, a flight type, the flight type including a direct flight and a transfer flight; and construct, according to the flight information and the flight type, a second user itinerary recovery model, the second user itinerary recovery model being configured to reflect an association between ticket sales revenue and compensation fees when the direct flight is converted into the transfer flight. Correspondingly, the recovery module can be further configured to: allocate, according to the second user itinerary recovery model, flight resources for the users corresponding to the abnormal flight.

[0024] In a possible implementation, the constructing module can be further configured to: acquire flight information of the target flight, the flight information of the target flight including a flight departure airport, a flight departure time, a flight arrival airport, and a flight arrival time; and construct a third user itinerary recovery model according to the flight information of the target flight and a function model, the third user itinerary recovery model being used to reflect an association between a profit and a flexible ticket quantity ratio when the target flight is an abnormal flight, the function model being constructed based on a fitting analysis method according to historical booking data and being used to describe an association between the flexible ticket quantity ratio and a flexible ticket price ratio, where the flexible ticket price is less than a specific ticket price, and both a delay compensation and a downgrading compensation of a user corresponding to the flexible ticket are 0, and the flexible ticket corresponding to the user is refunded when the user refuses to board.

[0025] In a possible implementation, the determining module can be further configured to: determine the influence coefficient of the flexible ticket quantity ratio and the flexible ticket price ratio on the recovery strategy based on a sensitivity analysis method.

[0026] In a possible implementation, the constructing module can be further configured to: combine at least two of the first user itinerary recovery model, the second user itinerary recovery model, and the third user itinerary recovery model to obtain a plurality of fourth user itinerary recovery models. Correspondingly, the recovery module can be further configured to: for each fourth user itinerary recovery model in the plurality of fourth user itinerary recovery models, allocate flight resources for users corresponding to an abnormal flight according to the fourth user itinerary recovery model.

[0027] In a possible implementation, the route space-time network in the airline passenger itinerary recovery apparatus is constructed in the following manner: acquiring historical flight data, the historical flight data including a flight number, a tail number, a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a cabin, a number of tickets sold, and ticket price data; performing data preprocessing on the historical flight data to obtain target flight data, the data preprocessing being used to remove abnormal data, the abnormal data including flight data with a small number of flight frequencies between two airports; and constructing the route space-time network according to the target flight data.

[0028] In a third aspect, the present application provides an electronic device, including: a memory and a processor. The memory is used to store program instructions; and the processor is used to invoke the program instructions in the memory to execute the airline passenger itinerary recovery method in the first aspect.

[0029] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed to implement the airline passenger trip recovery method in the first aspect.

[0030] In a fifth aspect, the present application provides a computer program product, wherein the computer program product contains a computer program, and the computer program is executed by a processor to implement the airline passenger trip recovery method in the first aspect.

[0031] The airline passenger trip recovery method, device, equipment and storage medium provided by the present application are provided. The flight information of an abnormal flight is obtained, the flight information includes a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a ticket price and a ticket sales quantity. Based on a clustering method, the adjacent airports of the flight departure airport are aggregated according to a route space-time network to obtain a multi-airport group region. The route space-time network is a relationship network between routes established between hub routes and between hub routes and non-hub routes. According to the flight information and the multi-airport group region, a first user trip recovery model corresponding to the abnormal flight is constructed. The first user trip recovery model is used to reflect the correlation between ticket sales revenue, compensation fees and transportation costs from the flight departure airport to the adjacent airports. According to the first user trip recovery model, flight resources are allocated for users of the abnormal flight. Through the multi-airport group region, the user can be rebooked to other airports in the multi-airport group region, and the transportation costs involved between the original flight and the new flight are considered. A more realistic user trip recovery model can be established, the airline company can more flexibly allocate flight resources, thereby increasing the feasible user trip recovery strategy, expanding the solution space of the user trip recovery problem, providing convenience for users, further improving the ability of the airline company to respond to interference, maximizing the economic benefits of the airline company, and the like. BRIEF DESCRIPTION OF DRAWINGS

[0032] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0033] Figure 1 is a flowchart of an abnormal flight sub-problem provided by an embodiment of the present application and the factors considered;

[0034] Figure 2 is an application scenario diagram provided by an embodiment of the present application;

[0035] Figure 3 is a flowchart of an airline passenger trip recovery method provided by an embodiment of the present application;

[0036] Figure 4is a schematic diagram of domestic flight take-off and landing times and normal rates from 2008 to 2018 provided by an embodiment of the present application;

[0037] Figure 5 is a flowchart of a column generation algorithm provided by an embodiment of the present application;

[0038] Figure 6 is a flowchart of a large neighborhood search algorithm provided by an embodiment of the present application;

[0039] Figure 7 is a schematic diagram of a heat map result of flight 5 being canceled provided by an embodiment of the present application;

[0040] Figure 8 is a schematic diagram of a structure of an airline passenger itinerary recovery device provided by an embodiment of the present application;

[0041] Figure 9 is a schematic diagram of a structure of an electronic device provided by an embodiment of the present application.

[0042] Through the above-mentioned drawings, the specific embodiments of the present application have been shown, and will be described in more detail hereinafter. These drawings and the written description are not intended to restrict the scope of the concept of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0043] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The following description is made with reference to the accompanying drawings in which like reference numerals represent like elements, unless the context of use indicates otherwise. The following description of exemplary embodiments is not representative of all embodiments consistent with the present application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present application as detailed in the appended claims.

[0044] The terms "first", "second", and the like in the description and the claims of the present application are used to distinguish similar objects, and do not necessarily indicate a particular order or a chronological sequence. It should be understood that the data thus used can be interchanged, where appropriate, so that the embodiments of the present application described herein can be carried out in other than the order shown or described herein. Furthermore, the terms "comprise" and "have", and any variations thereof, are intended to cover non-exclusive inclusion, for example, processes, systems, products, or devices that include a series of steps or units not necessarily limited to those clearly listed, but can include other steps or units not clearly listed or inherent to these processes, products, or devices.

[0045] First, some technical terms related to the present application are explained:

[0046] Airfly: refers to the flight of an aircraft without carrying passengers (or passengers, users).

[0047] Airfly: refers to the flight of an aircraft without carrying passengers (or passengers, users).

[0048] In existing research, the recovery problem of abnormal flights is solved in sequence or in stages, including flight recovery problem, aircraft recovery problem, crew recovery problem and passenger itinerary recovery problem, etc. sub-problems, as shown in the following formula: Figure 1 For the flight recovery problem, the flight plan or the disturbance scenario needs to be considered. Considering the flight plan, including repairing the flight schedule, such as flight time adjustment, flight exchange, etc. For example, the flight task originally performed by aircraft A is exchanged to be performed by aircraft B. Or, according to the disturbance scenario, the corresponding flight recovery is performed, such as weather reasons or air traffic control, canceling the original flight, transferring to another airport flight, etc.

[0049] For the aircraft recovery problem, the maintenance plan and airworthiness plan need to be considered to ensure that the aircraft is continuously airworthy, on time, etc. when flying. If the aircraft has a problem, the flight of the aircraft can be delayed or canceled, and the available aircraft can be allocated to the flight. For the crew recovery problem, the flight time, duty time and aircraft model of the crew need to be considered, and the flight task of the crew is reasonably arranged. For the passenger itinerary recovery problem (or passenger recovery problem), the user's itinerary needs to be modified, such as re-allocating the user to the available flight, which may exist delay, downgrading, transfer, etc.

[0050] Based on the above embodiment, it can be known that the passenger itinerary recovery method (or user itinerary recovery method) of the airline in the prior art needs to consider multiple sub-problems, each of which may affect the recovery of the user itinerary, and each sub-problem also involves complex resource requirements, and the anti-interference ability is poor.

[0051] In view of the problem that the user itinerary recovery strategy in the related art is single and the anti-interference ability is poor, the present application proposes an airline passenger itinerary recovery method, which considers a multi-airport group area to establish a more realistic passenger (or user) itinerary recovery model, which can increase the feasible user itinerary recovery strategy, so that the user can flexibly choose according to the actual situation, and improve the ability of the airline to cope with interference, and maximize the profit of the airline.

[0052] Currently, there are two main route network structures: a point-to-point route network structure and a hub-and-spoke route network structure. Among them, the point-to-point route network structure is to establish a direct route connection between any two cities, which is the most ideal user transportation method when cost is not considered, embodies the characteristics of fast air transportation, and maximizes the saving of user transit time. However, due to the limited market demand of two cities, the flight frequency and seat rate of the point-to-point route are low, which may cause waste of route resources.

[0053] The hub-and-spoke route network structure is to aggregate network traffic to hubs and axes, establish route connections between hubs and hubs, hubs and non-hubs, and there is no direct route connection between non-hubs and non-hubs, which is to reduce unit transportation cost by using scale economy effect and improve system stability. Therefore, in order to meet market demand, most airlines allocate limited route resources on the hub route network. However, due to the large flight traffic between hubs and hubs, there is no direct route connection between non-hubs, and the hub route network is very fragile even to small disturbances.

[0054] In some examples, based on the existing route network structure and the passenger trip recovery model such as the PRP model, the introduction of a multi-airport group area, the transfer of users to other flights and other airports, or the consideration of transfer airports to recover user trips can more fully utilize the route resources between hubs and non-hubs, between non-hubs and non-hubs, and improve the ability of the hub route network to face disturbances.

[0055] Figure 2 is an application scenario provided by an embodiment of the present application. As shown in Figure 2 The application scenario includes a first client 11, a server 12, a second client 13, a third client 14, and a user 15, wherein the number of the first client 11, the second client 13, and the third client 14 can be at least one.

[0056] In actual applications, such as an airport, the first client 11 can collect relevant information of a flight that has arranged a take-off task, such as a flight route, a weather condition of a flight take-off-landing city, a ticket sales number, and the like, and then send the relevant information to the server 12. The server 12 processes and monitors the relevant information periodically sent by the first client 11. If the server 12 monitors that the relevant information of the flight A is abnormal, such as a weather change, a flight route being regulated, and the like, which indicates that the flight plan of the flight A is disturbed, the server 12 can perform corresponding flight plan recovery or user itinerary recovery according to the flight information corresponding to the flight A, and feed back the recovery result to the first client 11, the second client 13, and the third client 14. The user 15 can learn the latest message of the flight in time from the second client 13 and / or the third client 14, such as whether the flight is delayed, whether the flight is rescheduled, a reason for flight delay, and the like. The first client 11 and the second client 13 can be a mobile phone, a computer, a notebook computer, a personal digital assistant (PDA), or the like, and the third client 14 can be an electronic display screen of the airport.

[0057] It should be noted that the server 12 can be replaced by a server cluster or other computing device with certain computing power. In addition, the airline passenger itinerary recovery method provided in the present application can also be used for recovery of other itineraries, such as high-speed rail itineraries, train itineraries, and the like.

[0058] The airline passenger itinerary recovery method according to the exemplary embodiments of the present application will be described below in combination with an application scenario of Figure 2 , with reference to Figure 3 . It should be noted that the above-mentioned application scenario is only shown for the purpose of facilitating understanding of the spirit and principles of the present application, and the embodiments of the present application are not limited to the application scenario shown in Figure 2 .

[0059] Figure 3 is a flowchart of the airline passenger itinerary recovery method provided in an embodiment of the present application. As shown in Figure 3 , the airline passenger itinerary recovery method in the embodiment of the present application includes the following steps:

[0060] S301: Obtain flight information of an abnormal flight, the flight information including a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a ticket price, and a ticket sales number.

[0061] In actual situations, a flight affected by interference is usually referred to as an abnormal flight. The interference can be caused by two sources: an internal source and an external source. The internal source includes an aircraft or crew abnormality, and the external source includes weather and air traffic control, and the like. Typical interferences include aircraft interference, airport interference, flight delay, and flight cancellation, and the like.

[0062] Figure 4 is a schematic diagram of the number of takeoffs and landings and the normal rate of domestic flights from 2008 to 2018 provided by an embodiment of the present application. Among them, the column chart represents the number of takeoffs and landings (unit: million), and the line chart represents the normal rate of takeoffs and landings. It can be seen from Figure 4 that the number of takeoffs and landings has increased in the past ten years, but the normal rate has fluctuated between 70% and 80% without significant growth. Related investigations show that for an airline, the operating cost related to abnormal flights may cost 3% of the annual revenue. The airline with a better flight normality (or flight normal rate) has a stronger economic benefit (or profit) level, so the airline needs to further improve the flight normal rate in order to maintain a better economic benefit.

[0063] S302: Based on the clustering method, the neighboring airports of the flight departure airport are aggregated according to the route space-time network to obtain a multi-airport group region. The route space-time network is a relationship network between hub routes and between hub routes and non-hub routes.

[0064] In this step, the clustering method can use the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) method. Unlike the partition clustering method and the hierarchical clustering method, the DBSCAN clustering method defines a cluster as the maximum set of density-connected points, which can divide regions with sufficiently high density into clusters and find clusters of arbitrary shape in a noisy spatial database.

[0065] Using the DBSCAN clustering method, the neighboring airports of the flight departure airport can be aggregated to form a multi-airport group region. Based on the multi-airport group region, when the flight is disturbed, the user can be rescheduled to other airports in the multi-airport group region, which can provide more choices for the user and facilitate the rapid and efficient solution of the user's itinerary recovery problem.

[0066] S303: According to the flight information and the multi-airport group region, a first user itinerary recovery model corresponding to the abnormal flight is constructed. The first user itinerary recovery model is used to reflect the correlation between ticket sales revenue and compensation fees and transportation costs from the flight departure airport to the neighboring airport.

[0067] In some examples, the first user itinerary recovery model, i.e., the user itinerary recovery (PRP-T) model considering the transportation between multi-airport group regions, is based on the basic PRP model, considering flight information and multi-airport group regions, etc., and is improved to obtain. Among them, the basic PRP model is only used to reflect the correlation between ticket sales revenue and compensation fees when the disturbance occurs.

[0068] In addition, the difference between the PRP-T model and the basic PRP model considering the multi-airport group region is that the PRP model only allows the user to be rescheduled to other flights of the same airport, while the PRP-T model considers the case where multiple airports exist in a region, and the user can be rescheduled to a flight of the same airport or other airports in the same region. The DBSCAN clustering method can be used to cluster the adjacent airports into a multi-airport group region.

[0069] Further, if the user is rescheduled to a new flight of another airport, the airline needs to compensate the user for the transportation cost from the current airport to the new airport. By introducing the multi-airport group region, the airline can more flexibly allocate flight resources, expand the solution space of the user itinerary recovery problem, solve the problem of single flight distribution based on the point-to-point route network, and the flight frequency and seat rate of the point-to-point route are low, which may cause waste of route resources, so that the airline can obtain more economic benefits as much as possible.

[0070] In some embodiments, when improving the basic PRP model, a basic PRP model can be constructed first. In constructing the basic PRP model, some historical flight data can be obtained first, but the obtained historical flight data may not contain specific user itinerary data, such as the specific passenger capacity (or user number) of a flight during a flight process, etc. The discrete choice model (DCM) can be used to preprocess the obtained flight data, and then estimate the specific passenger capacity of each aircraft during the flight process or the user number on each user itinerary. The data preprocessing includes removing abnormal data, such as merging multiple data of shared flights when multiple airlines share a flight, etc.

[0071] Among them, common discrete choice models include binary logit model, multi-nominal logit model, conditional logit model, nested logit model, and mixed logit model, etc. There are also many software that can be used to fit discrete choice models, such as Python, Matlab, etc.

[0072] For example, possible user itineraries can be constructed first, only considering direct or one-time transfer user itineraries. Then, for the user number on each user itinerary, a discrete choice model is used for distribution, as shown in formula (1).

[0073]

[0074] where x i denotes the feature set of user itinerary (or flight) i, x j denotes the feature set of user itinerary j, u(x i ) is the utility function of user itinerary i, and u(x j ) is the utility function of user itinerary j. The utility function is related to the characteristics of the user itinerary itself, such as the departure time of the itinerary, whether the itinerary is canceled, the number of seats on the itinerary, etc. A polynomial function can be used as the utility function.

[0075] The denominator in formula (1) represents the sum of the utilities of all itineraries. When the utility function u(x i ) of a single itinerary is divided by the sum of the total utilities, the probability of user itinerary i being selected can be obtained, i.e., the proportion of users on each itinerary P(i) is obtained, so that the total number of users is distributed to each itinerary to obtain the number of users on each itinerary.

[0076] Corresponding to the above embodiment, the obtained number of users can be used as the input of each model in other embodiments to calculate the ticket sales revenue in the objective function, etc.

[0077] where in the basic PRP model, the objective function is to maximize the total profit, including ticket sales revenue and compensation costs. The first part of the objective function is the ticket sales revenue, as shown in formula (2). The second part is the compensation costs, including rebooking costs (such as delay costs, as shown in formula (3), downgrading costs, as shown in formula (4)), and denied boarding costs, as shown in formula (5).

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] x kj = 0 or 1, k e {1,..., N}, j e {0,..., N F} (15)

[0092] In the above formula (2), maximize means maximizing, ∑ means summation symbol, N means total number of tickets, k means user, means ticket price of original flight.

[0093] In the above formula (3), j means new flight number to which user is seated after disturbance, taking value range [0, N F ], N F means number of flights, x kj = 0 or 1, means whether user k is seated on flight j after disturbance, as shown in formula (15). means user delay cost generated when user k is rebooked to flight j after original flight, and how much the cost is depends on delay time, and specific calculation method is shown in formula (6), wherein, e0 means user delay boarding cost (or compensation) coefficient, Dep j means departure time of new flight after rebooking, means departure time of original flight.

[0094] In the above formula (4), means downgrading cost, used to consider the case that cabin ticket price of new flight is lower than cabin ticket price of original flight, and specific calculation method is shown in formula (7), wherein, max means taking maximum value, means ticket price of new flight after rebooking. In addition, in formula (4), x kj = 0 or 1, means whether user k is seated on flight j after disturbance, as shown in formula (15).

[0095] In the above formula (5), x k0 = 0 or 1, means whether user k is seated on flight after disturbance, as shown in formula (15). means cost generated when user is denied boarding, and the cost depends on ticket price and flight time of original flight, and specific calculation method is shown in formula (8), wherein, e1 means user denied boarding cost coefficient, means arrival time of original flight.

[0096] In addition, formula (9)-formula (15) all represent the constraint conditions (s.t.) of the basic PRP model. Specifically, formula (9) represents that the canceled flight cannot carry any user, wherein, represents the maximum number of flights that are limited to be canceled. Formula (10) represents that one user is assigned to one flight or refused to board. Formula (11) represents that the capacity of the aircraft is limited, and the maximum number of passengers is not more than C j , C j is the number of seats of flight j. Formula (12) is used to ensure that the user is rescheduled to the same pair of departure and arrival airports, represents the traffic distance of the user rescheduled to the same pair of departure and arrival airports. Formula (13) and formula (14) are both time constraints, wherein, formula (13) is used to ensure that the departure time of the new flight is not earlier than the departure time of the original flight; formula (14) defines the maximum arrival delay, and the maximum delay time is 18 hours for domestic and intracontinental flights, and the maximum delay time is 36 hours for intercontinental flights, wherein, j represents the arrival time of the new flight after rescheduling, T max represents the maximum delay time.

[0097] According to relevant regulations, the user delay boarding cost coefficient e0 can be set to 200 yuan / hour, the user refused boarding cost coefficient e1 can be set to 200 yuan / hour, the taxi is used as the transportation tool between the airports in the multi-airport group area, the taxi speed is set to 40 kilometers / hour, and the transportation cost coefficient is 2 yuan / kilometer, that is, 80 yuan / hour.

[0098] Compared with the basic PRP model, the PRP-T model is an extension of the basic PRP model, which considers the transportation cost (as shown in formula (16)) and transportation time (as shown in formula (19)) from the original airport to the new airport, so as to construct a user trip recovery model that is more in line with the actual situation.

[0099]

[0100]

[0101]

[0102]

[0103] Specifically, the transportation cost is added in the objective function in the above embodiment to consider the cost of transporting the user from the original airport to another airport. The specific transportation cost is calculated in the manner shown in formula (17), wherein e2 represents the transportation cost coefficient.

[0104] Formula (18) relaxes the constraint condition of formula (12), allowing the user to be rebooked to other flights of the same flight departure and landing city pair, that is, the new flight is the same as the original flight in terms of origin-destination (OD) city pair. Wherein, represents the traffic distance of the user being rebooked to other flights of the same flight departure and landing city pair as the original flight.

[0105] In addition, formula (19) additionally considers the traffic time Tra kj .

[0106] Based on the above embodiments, it can be seen that the introduction of the multi-airport group region and the transfer can establish a more realistic user itinerary recovery model, can make the airline more flexible to allocate flight resources, increase the feasible user itinerary recovery strategy, expand the solution space of the user itinerary recovery problem, and further improve the ability of the airline to cope with interference, and maximize the economic benefit of the airline.

[0107] S304: According to the first user itinerary recovery model, allocate flight resources for users of abnormal flights.

[0108] Based on the first user itinerary recovery model, that is, the PRP-T model, obtained based on the above embodiments, some algorithms such as column generation algorithm or large neighborhood search algorithm can be used to solve the model to obtain the user itinerary recovery strategy provided by the airline, to ensure that after the interference occurs, the user can still be provided with convenience as much as possible, such as flight delay, user ticket rebooking, or rebooking to other flights of the same OD city pair, etc. The user can also select a suitable user itinerary recovery strategy according to the actual situation.

[0109] Compared with the related art, the airline passenger itinerary recovery method provided by the present application models the delay cost, the downgrading cost, and the denied boarding cost involved in the user itinerary recovery problem in detail, and can better balance the related recovery strategies of user reassignment.

[0110] The airline passenger itinerary recovery method provided by the embodiment of the present application can establish a more realistic user itinerary recovery model by considering rebooking the user to other airports in the multi-airport group region, and considering the traffic cost involved between the original flight and the new flight, etc. The airline can more flexibly allocate flight resources, thereby increasing the feasible user itinerary recovery strategy, expanding the solution space of the user itinerary recovery problem, providing convenience for the user, further improving the ability of the airline to cope with interference, maximizing the economic benefit of the airline, etc.

[0111] In some examples, the flight resources are allocated for the user of the abnormal flight according to the first user itinerary recovery model, including: determining an optimal solution of the first user itinerary recovery model by using a preset algorithm, the preset algorithm including a column generation algorithm and a large neighborhood search algorithm; and allocating the flight resources for the user of the abnormal flight according to the optimal solution. In view of the large amount of flight operation data in the country, the column generation algorithm is used to accurately solve small-scale examples of each model, and the large neighborhood search algorithm is used to approximately solve large-scale examples of each model, so that the calculation efficiency can be greatly improved and the operation time can be saved. In addition, the algorithm design research can also be migrated to the solution of other combination optimization problems of airline operation management.

[0112] Specifically, the column generation algorithm is an efficient exact algorithm for solving large-scale linear programming problems, which accelerates the solving speed by iteratively solving sub-problems with fewer variables than the original problem. First, the original problem is converted into smaller sub-problems, and then the sub-problems are solved. Second, the sub-problems are used to determine whether there is a variable that makes the reduced cost (RC) less than 0 among the variables not considered, and if so, the variable is added to the sub-problems and the sub-problems are re-solved. The above two steps are repeatedly performed until no new variable is added to the sub-problems, and then the optimal solution of the original problem is found.

[0113] In addition, the target function solving maximum value problem of the present application can be converted into a minimum value problem, and when the optimization target function is a minimum value function, the reduced cost can be used to determine whether the new variable introduced can further reduce the value of the target function. When the reduced cost is negative, the introduced variable can make the value of the target function smaller, and the variable is added to the sub-problems; when the reduced cost is positive, the introduced variable can make the value of the target function larger, and the variable is not added to the sub-problems.

[0114] Figure 5 is a flowchart of the column generation algorithm provided by an embodiment of the present application. As shown in Figure 5 , an initial solution is found according to the first user itinerary recovery model, and a restricted master problem (RMP) is constructed, that is, a sub-problem with fewer variables than the original problem is constructed. By calculating the sub-problems, a number of new variables with RC<0 can be generated, and it is determined whether the newly generated variables are the same as the existing variables, that is, whether no new variable is generated. If so, the optimal solution is obtained; if not, the new variable is added to the RMP for solving.

[0115] The large neighborhood search algorithm is a heuristic algorithm for solving large-scale optimization problems. The algorithm improves the solution quality step by step by alternately using the two methods of destruction and repair, and constantly approaches the optimal solution. First, in the construction stage, for abnormal flights, an initial solution can be constructed by delaying or canceling flights. Second, in the repair stage (or recovery stage), the flights that do not satisfy the airport capacity constraints in the initial solution, the canceled flights, and the user's reallocation part are adjusted to obtain a feasible solution. Third, in the improvement stage (or promotion stage), the flights and users are fine-tuned by a local search method to obtain a better solution. The second and third steps are repeatedly performed until the iteration time reaches the set upper limit, and the calculation is stopped. Although the result obtained by the large neighborhood search algorithm cannot guarantee the optimality, it can quickly obtain the optimization result of a large-scale problem.

[0116] Figure 6 is a flowchart of the large neighborhood search algorithm provided by an embodiment of the present application. As shown in Figure 6 , it includes a construction stage, a recovery stage, and a promotion stage. In the construction stage, an initial solution can be constructed by delaying or canceling flights according to the first user itinerary recovery model for abnormal flights. In the recovery stage, the initial solution constructed is adjusted to obtain a feasible solution. In the promotion stage, the feasible solution is fine-tuned by a local search method to obtain a better solution.

[0117] In addition, the large neighborhood search algorithm includes two loops in Figure 6 , wherein the small loop includes "construction-recovery-construction-recovery…", which is repeatedly performed until the central processing unit (CPU) time reaches the set time upper limit or a certain number of iterations without improvement, and then jumps to the "promotion stage". After the promotion stage ends, it returns to the construction stage. The large loop includes "construction-recovery-(reaching the CPU time upper limit or the iteration number upper limit without improvement)-promotion-construction-…", which can repeatedly iterate the construction, recovery, and promotion of the obtained solution until the set CPU time upper limit is reached. The CPU time, i.e., the code running time, can be set according to the actual situation, and is not limited here.

[0118] In some embodiments, the airline passenger itinerary recovery method further comprises: determining a flight type according to the flight information, the flight type comprising a direct flight and a transfer flight; constructing a second user itinerary recovery model according to the flight information and the flight type, the second user itinerary recovery model being used to reflect the association between ticketing revenue and compensation cost when converting a direct flight into a transfer flight; and allocating flight resources for the user corresponding to the abnormal flight according to the second user itinerary recovery model. The second user itinerary recovery model is a user itinerary recovery (PRP-I) model considering transfer flights (or transfer airports), and the PRP-I model is based on a basic PRP model and considers the case of splicing at least two flights into one direct flight.

[0119] For example, according to the departure airport / city and the arrival airport / city of the flight in the flight information, the flight can be further divided into a direct flight and a transfer flight. Further, an intermediate airport can be considered to be introduced to increase the solution space of the user itinerary recovery strategy, and the airline company can further reduce the compensation cost.

[0120] Specifically, when the abnormal flight is a direct flight from region A to region B, other flights from the departure airport / city of the flight can be queried according to the flight information and the multi-airport group region. If there is a flight 1 from region A to region C and a flight 2 from region C to region B, the existing transfer flights, i.e., the flight 1 and the flight 2, can be spliced at the beginning and the end to be considered as a feasible direct flight, and the departure and arrival airports / cities of the direct flight are the same.

[0121] In another example, the flights of the neighboring airports / cities of the departure airport / city of the flight can also be queried according to the flight information and the multi-airport group region. Then, there will be transportation costs and transportation times, etc. The transportation costs and the transportation times can exist when going from the airport / city of the abnormal flight to the neighboring airport / city, or when converting the flight 1 to the flight 2, etc. The airline company will give corresponding compensation, etc. for the existing transportation costs. It can be understood that in this embodiment, the transfer airport and the multi-airport group region are considered, i.e., the PRP-I model and the PRP-T model are combined, and the column generation algorithm or the large neighborhood search algorithm can also be used for solving. The specific solving process can be referred to the above embodiments, and will not be described here.

[0122] In addition, after combining the PRP-I model and the PRP-T model, more user itinerary recovery strategies can be obtained, and the solution space of the user itinerary recovery strategy is further improved.

[0123] Further, the existing research on irregular flight management does not consider the problem from the perspective of revenue management, and lacks a redundant user trip recovery strategy before the disturbance occurs; and most revenue management only studies the static seat allocation without disturbance. The present application considers the combination of user trip recovery problem and revenue management (RM) research, wherein the revenue management mainly considers the introduction of flexible products (Flexible Product), such as flexible tickets. The introduction of flexible tickets in the user trip recovery problem can be used as a proactive prevention strategy for disturbances, thereby obtaining a flight recovery strategy that is beneficial to both users and airlines.

[0124] Among them, the flexible product refers to multiple selectable products serving the same market, the price of which is lower than that of the specific product (Specific Product), and the specific product is not specified when it is sold, but is specified by the supplier later. Flexible products have been widely studied in the field of revenue management, and have two major advantages: risk pooling and demand induction. Risk pooling refers to the fact that the designation of flexible products is after specific products, thereby allowing the supplier to solve the imbalance between demand and capacity; demand induction refers to the fact that due to the lower price, flexible products may attract customers who would not have purchased specific products, thereby increasing product demand.

[0125] Based on the above embodiments, the airline passenger trip recovery method provided by the present application also introduces flexible tickets, which are usually not specified to a specific flight until a later time.

[0126] In some embodiments, the airline passenger itinerary recovery method further comprises: obtaining flight information of a target flight, the flight information of the target flight comprising a flight departure airport, a flight departure time, a flight arrival airport, and a flight arrival time; constructing a third user itinerary recovery model according to the flight information of the target flight and a function model, the third user itinerary recovery model being used to reflect an associated relationship between a profit and a flexible ticket quantity ratio and a flexible ticket price ratio when the target flight is an irregular flight, the function model being constructed based on a fitting analysis method according to historical booking data and being used to describe the associated relationship between the flexible ticket quantity ratio and the flexible ticket price ratio, wherein the flexible ticket price is less than a specific ticket price, and both a delay compensation and a downgrading compensation of a user corresponding to the flexible ticket are 0, and the user corresponding to the flexible ticket is refunded the flexible ticket fee when refusing to board; determining an influence coefficient of the flexible ticket quantity ratio and the flexible ticket price ratio on the recovery strategy; and determining the flexible ticket quantity ratio and the flexible ticket price ratio corresponding to the target flight based on the influence coefficient corresponding to different flexible ticket data ratios and flexible ticket price ratios. The third user itinerary recovery model is a user itinerary recovery model considering flexible tickets (PRP-F) model.

[0127] For example, a conventional cabin control technique assumes that an airline only provides specific tickets (or determined tickets) with fixed departure and arrival times to users, and when a disturbance occurs, the passengers are given corresponding compensation. In addition, the price of a specific ticket is generally fixed and higher than that of a flexible ticket. However, because a flexible ticket is sold at a lower price, a user who purchases a flexible ticket will not obtain a delay compensation or a downgrading compensation when changing a ticket.

[0128] Further, according to the type of the purchased ticket, a user can be divided into a deterministic user and a flexible user.

[0129] It can be understood that the introduction of a flexible ticket has two opposite effects on the overall profit of an airline. On the one hand, the price of a flexible ticket is lower than that of a specific ticket, resulting in a decrease in ticketing revenue. On the other hand, a user who purchases a flexible ticket will not incur any fees when changing a ticket, thereby reducing the total user change fee. Because the overall profit of an airline is the sum of the revenue and various compensation fees, different parameter configurations, i.e., different flexible ticket ratios and prices, can have different effects on the overall profit.

[0130] Compared with the basic PRP model, the PRP-F model is also an extension of the basic PRP model, and a part of specific tickets is sold at a reduced price as flexible tickets. Therefore, when the total number of users is a constant (const), the number of flexible tickets N flex and the number of specific tickets N spec sum up to the constant, and the constant is equal to the total number of tickets N, as shown in equation (20).

[0131] N flex +N spec =N=const (20)

[0132] r d =A×1 / (1+exp(∈×(r p -0.25))) (21)

[0133] N flex =min{r l ,r d}×N (22)

[0134]

[0135]

[0136]

[0137] Optionally, the demand r d of flexible tickets by users can be assumed to be a function of the ratio r p of the price of flexible tickets, according to actual conditions (such as a Sigmoid function, specifically as shown in equation (21)). It can be understood that the lower the price of flexible tickets, the higher the demand for flexible tickets, that is, the more flexible tickets are sold, and the less specific tickets are sold.

[0138] Through sensitivity analysis, it can be known that when the ratio r p of the price of flexible tickets changes in the range of [0.00, 0.50], ε determines the change speed of the demand r d of flexible tickets, and A controls the maximum possible demand of flexible tickets. For example, according to historical data of users ordering flexible tickets, the values of ε and A can be estimated. Similarly, according to historical data of users ordering flexible tickets, or the sales situation of current tickets, the ratio r p of the price of flexible tickets can also be determined, and then the price p flex of flexible tickets can be determined by p flex =r p ×max{p spec}, where the value range of r p is [0.0, 1.0], p spec represents the price of specific tickets, and max represents the maximum value of the price of specific tickets.

[0139] In addition, the actual number N flex of flexible tickets sold is the upper limit r l of the number of flexible tickets and the demand r dThe minimum value (min) is multiplied by the total number of tickets N, as shown in formula (22). l With r d The value range is [0.0, 1.0], and the maximum number of flexible tickets r is... l It can be determined based on the airline's historical data or current ticket sales.

[0140] Furthermore, after setting the quantity and price of flexible tickets, users can freely choose to order specific tickets or flexible tickets. However, if a user who purchased a flexible ticket is delayed or downgraded after an disruption occurs, they will not receive compensation, as shown in formulas (23) and (24). In addition, if a user who purchased a flexible ticket is denied boarding, they will only be refunded the cost of the flexible ticket, as shown in formula (25).

[0141] For example, based on the influence coefficients corresponding to different flexible ticket data proportions and flexible ticket price ratios, after determining the proportion of flexible ticket quantity and flexible ticket price ratio for the target flight, the PRP-F model can be solved using a column generation algorithm or a large neighborhood search algorithm. The specific solution process can be referred to the above embodiments, and will not be repeated here.

[0142] Based on the above embodiments, it can be seen that the airline passenger itinerary recovery method provided in this application, while offering two passive response strategies (introducing multi-airport clusters and transit airports) to cope with flight schedule interference, also provides a proactive prevention strategy, namely, introducing flexible tickets, to proactively address flight schedule interference. Introducing multi-airport clusters involves integrating clustered airports; introducing transit airports involves classifying flights into direct flights and connecting flights; introducing flexible tickets involves converting a portion of the original fixed tickets into flexible tickets, resulting in a new number of fixed tickets and a new number of flexible tickets, thereby identifying users with fixed tickets and users with flexible tickets. Furthermore, the study of flexible tickets can use a customer choice model, i.e., a discrete choice model; however, due to the complexity of the problem, only a few flight data points are analyzed and calculated to obtain the flight number, ticket price, departure time, etc., for a specific user.

[0143] In some embodiments, determining the impact coefficients of the proportion of flexible ticket quantities and flexible ticket prices on the recovery strategy includes: determining the impact coefficients of the proportion of flexible ticket quantities and flexible ticket prices on the recovery strategy based on sensitivity analysis. Sensitivity analysis (or sensitivity analysis experiments) is an uncertainty analysis method that can identify sensitive factors that have a significant impact on the project's economic benefit indicators from numerous uncertain factors, and analyze and calculate the degree of influence and sensitivity of these sensitive factors on the project's economic benefit indicators, thereby assessing the project's risk-bearing capacity.

[0144] Corresponding to the above embodiment, in the sensitivity analysis experiment, the upper limit r l of the number of flexible tickets is increased from 0.0 (i.e., no flexible tickets are introduced) to 1.0 (i.e., all are flexible tickets) with a step of 0.1; the price ratio r p of the flexible tickets is increased from 0.00 to 0.50 with a step of 0.05. For example, each set of data in [r l , r p ] is substituted into the PRP-F model for experiment, and the column generation algorithm or the large neighborhood search algorithm is used to calculate the interference after-profits corresponding to each set of data.

[0145] Based on the above embodiment, the introduction of flexible tickets can be controlled by setting the price ratio r p and the upper limit r l of the number of flexible tickets, and the demand r d of flexible tickets can be obtained by formula (21), where the parameters A and ε can be estimated according to historical user subscription data of flexible tickets. Alternatively, if historical user subscription data of flexible tickets cannot be obtained, A = 1 and ε = 20 can be simply set, i.e., when r p decreases to 0.00, the demand r d approaches 1.0; when r p increases to 0.50, the demand r d drops to 0.0.

[0146] In some examples, the upper limit r l of the number of flexible tickets can be set to 0.0 and 0.1 (i.e., flexible tickets are introduced), and the price ratio r p of the flexible tickets is set to 0.45, and the experimental results when flights are canceled due to disturbances are shown in Table 1. Table 1 includes the results (Canx) when flights 1 to 8 are canceled, the number of specific tickets N spec , the number of flexible tickets N flex , the pre-disturbance profit P be , the post-disturbance profit P af , the number of users not affected by the disturbance N same , the number of delayed specific users D spec corresponding to specific tickets, the number of delayed flexible users D flex corresponding to flexible tickets, the number of deplaned users N down , and the number of refused boarding users N deny . Among them, the delayed flexible users corresponding to the flexible tickets will not be compensated due to the delay; the number of deplaned users N down includes the specific users corresponding to the specific tickets and the flexible users corresponding to the flexible tickets; the number of refused boarding users N denyThe table 1 includes the user's refusal to board corresponding to the specific ticket and the user's refusal to board corresponding to the flexible ticket. The number of users in each row of the table 1 satisfies N same +D spec +D flex +N deny =N=4102.

[0147] Table 1

[0148] Canx <![CDATA[N spec ]]> <![CDATA[N flex ]]> P be ]]> P af ]]> <![CDATA[N same ]]> <![CDATA[D spec ]]>

[00007] D flex ]] <![CDATA[N down ]]> N deny <!-- 14 -->]]> 1 4102 0 2832365.00 2621352.00 3777 142 0 0 183 1 4034 68 2824240.00 2619758.33 3760 159 18 0 165 2 4102 0 2832365.00 2558434.33 3817 232 0 208 53 2 4034 68 2824240.00 2571487.67 3773 267 34 229 28 3 4102 0 2832365.00 2673397.00 3840 134 0 18 128 3 4034 68 2824240.00 2676302.67 3830 144 14 23 114 4 4102 0 2832365.00 2620941.67 3704 306 0 0 92 4 4034 68 2824240.00 2629495.00 3689 310 32 0 71 5 4102 0 2832365.00 2547877.00 3845 73 0 73 184 5 4034 68 2824240.00 2549768.00 3838 77 11 77 176 6 4102 0 2832365.00 2604731.67 3769 204 0 0 129 6 4034 68 2824240.00 2611390.00 3747 217 26 0 112 7 4102 0 2832365.00 2543903.67 3832 87 0 0 183 7 4034 68 2824240.00 2545370.00 3805 101 26 0 170 8 4102 0 2832365.00 2539989.00 3875 0 0 0 227 8 4034 68 2824240.00 2537016.00 3868 0 11 0 223

[0149] As can be seen from the table 1, for the same flight, the introduction of the flexible ticket basically leads to a lower pre-interference profit P be , but leads to a higher post-interference profit P l when r p = 0.1 and r af = 0.45, which shows that the introduction of the flexible ticket has opposite effects on the revenue loss and the cost saving.

[0150] In the table 1, when the upper limit of the number of the flexible ticket r l = 0.1 and r p = 0.45, setting A = 1 and ε = 20, substituting into the formula (21) can obtain r d = 0.017, and then substituting r d = 0.017 and r l = 0.1 into the formula (22) can obtain N flex is about 73, that is, the number of the flexible ticket is 73, and then according to the actual situation, the number of the flexible ticket is fine-tuned, and the number of the flexible ticket in the table 1 is set to 68.

[0151] On the basis of the above embodiment, the sensitivity analysis is performed on the case that the flight 5 is cancelled. Figure 7 is a schematic diagram of the heat map result of the case that the flight 5 is cancelled provided by an embodiment of the present application, wherein the upper limit of the number of the flexible ticket r l increases from 0.0 to 1.0 with a step of 0.1, and the price ratio of the flexible ticket increases from 0.00 to 0.50 with a step of 0.05. When the price of the flexible ticket exceeds 0.50, the price of the flexible ticket will be too high, and the introduction of the flexible ticket will even bring income gain, therefore, only the case of r p ≤ 0.50 is considered. In addition, when substituting into the PRP-F model, the delay and the refusal to board cost coefficients e0 = e1 = 200 yuan / hour can be set.

[0152] Figure 7 Each heat map value in the table 2 represents the relative profit after the interference, wherein the calculation formula of the relative profit is (P af [r p ,r l ]-Paf [0,0]) / P af [0,0]. Therefore, in Figure 7 r l = 0.0 row, all values are 0, i.e., the relative profit before and after the disruption is 0 when no flexible tickets are introduced. When r p is less than 0.45, all relative profits are negative, indicating that the revenue loss from introducing flexible tickets is higher than the cost saving. When r p is greater than 0.45, the relative profit is positive, indicating that the cost saving from introducing flexible tickets is higher than the revenue loss. As can be seen from Figure 7 , the maximum profit is achieved at r p = 0.45, r l = 0.1, and is affected by the user booking behavior.

[0153] Still taking Figure 7 as an example, it can be seen that in each column, it can be observed that as r l decreases or increases, the final profit will remain unchanged, because at the corresponding price of each column, the demand for flexible tickets r d is lower than the upper limit of the number of flexible tickets r l , i.e., the number of flexible tickets sold is fixed, and further increasing the upper limit of the number of flexible tickets r l will only waste flight capacity and have no contribution to the increase of the overall profit. As shown in equation (21), as the price ratio of flexible tickets r p increases, the demand for flexible tickets r d decreases.

[0154] Based on the above examples, introducing flexible tickets will result in both revenue loss and cost saving, which have opposite effects on the profit after the disruption. In addition, introducing an appropriate number and price of flexible tickets can also lead to higher profits, but the number and price of flexible tickets need to be strictly set according to the user booking behavior. Finally, the induction function, i.e., equation (21), simulates the negative correlation between the demand for flexible tickets and the price in the real world.

[0155] In some embodiments, the airline passenger trip recovery method further includes: combining at least two of the first user trip recovery model, the second user trip recovery model, and the third user trip recovery model to obtain a plurality of fourth user trip recovery models; for each fourth user trip recovery model in the plurality of fourth user trip recovery models, allocating flight resources for users corresponding to the abnormal flight according to the fourth user trip recovery model.

[0156] It can be understood that the various cases considered in the above embodiments, such as the introduction of flexible tickets, transfer airports and multi-airport area, can be considered for introduction alone, or multiple cases can be considered for introduction after, that is, the PRP-F model, the PRP-I model and the PRP-T model can be freely combined to build different composite models, such as the PRP-TF model, the PRP-TI model and the PRP-TFI model.

[0157] For example, Table 2 is the profit increase percentage of different models relative to the basic PRP model, including the PRP-T model, the PRP-F model, the PRP-TF model, the PRP-I model, the PRP-TI model and the PRP-TFI model. Among them, the definition of the profit increase percentage is the difference between the profit of each model and the profit of the PRP model divided by the profit of the PRP model.

[0158] Table 2

[0159] Canx PRP PRP-T PRP-F PRP-TF PRP-I PRP-TI PRP-TFI 1 2621352.00 0.39％ -0.06％ 0.37％ 0.06％ 0.45％ 0.43％ 2 2558434.33 1.98％ 0.51％ 2.08％ 0.00％ 2.94％ 2.88％ 3 2673397.00 1.91％ 0.11％ 2.20％ 2.83％ 2.87％ 2.71％ 4 2620941.67 2.54％ 0.33％ 3.00％ 0.00％ 4.20％ 4.26％ 5 2547877.00 1.17％ 0.07％ 1.36％ 7.37％ 8.20％ 8.09％ 6 2604731.67 0.00％ 0.26％ 0.26％ 0.00％ 0.00％ 0.26％ 7 2543903.67 0.00％ 0.06％ 0.06％ 0.00％ 0.00％ 0.06％ 8 2539989.00 1.36％ -0.12％ 1.39％ 3.16％ 4.53％ 4.55％ 9 2490090.33 0.00％ 0.15％ 0.15％ 0.00％ 0.00％ 0.15％ 10 2501222.33 0.00％ 0.03％ 0.11％ 0.00％ 0.00％ 0.11％ 11 2679840.00 0.00％ -0.16％ 0.07％ 0.00％ 0.00％ 0.07％ 12 2736922.33 0.04％ -0.22％ -0.01％ 0.00％ 0.04％ -0.01％ 13 2694429.00 0.08％ -0.24％ -0.07％ 0.00％ 0.08％ -0.07％ 14 2577965.00 0.16％ -0.24％ -0.05％ 0.00％ 0.16％ -0.05％ 15 2581829.00 0.00％ -0.25％ -0.25％ 0.00％ 0.00％ -0.25％ 16 2670162.33 0.75％ 0.01％ 0.83％ 2.85％ 3.12％ 2.94％

[0160] From Table 2, first, except for some models considering flexible tickets, such as the PRP-F model, the PRP-TF model and the PRP-TFI model, the profit increase percentages calculated by the PRP-I model and the PRP-TI model are all non-negative, because only one specific flexible ticket setting (i.e., r l = 0.1, r p = 0.45) is used in this embodiment.

[0161] Second, when flights 9-15 are canceled, the values of the PRP-I model are all 0.00%, and the values of the PRP-TFI model are equal to the values of the PRP-TF model, because the departure time of flight 9-15 is later than the departure time of all transfer flights, resulting in that the transfer airport is not used. By constructing more transfer flights or transfer airports, these 0.00% values can become positive.

[0162] Third, when flights 2 and 4 are canceled, although the value of the PRP-I model is 0.00%, the value of the PRP-TI model is greater than the value of the PRP-T model. This result shows that the transfer airport has a positive impact on the PRP-TI model, because the introduction of the multi-airport area increases the possibility of using the transfer airport.

[0163] Fourth, when flight 8 is canceled, although the value of the PRP-F model is negative, the values of the PRP-TFI model and the PRP-TI model are positive, and the value of the PRP-TFI model is greater than the value of the PRP-TI model, indicating that some economic losses caused by the introduction of flexible tickets can be improved when the transfer airport and / or the multi-airport area are introduced.

[0164] The above results show that there is a nonlinear relationship between different models, i.e., the profit of the composite model is not equal to the sum of each component.

[0165] Based on the above embodiments, the airline passenger itinerary recovery method provided by the present application can provide a variety of different user itinerary recovery strategies for users to choose from, and the airline will have different economic benefits when considering different situations.

[0166] In some embodiments, the airline passenger itinerary recovery method is constructed by the following method: obtaining historical flight data, the historical flight data including flight number, tail number, flight departure airport, flight departure time, flight arrival airport, flight arrival time, cabin, ticket sales and ticket price data; performing data preprocessing on the historical flight data to obtain target flight data, the data preprocessing being used to remove abnormal data, the abnormal data including flight data with less flight frequency between two airports; and constructing the route space-time network according to the target flight data. The historical flight data can be provided by the airline or obtained by online inquiry, but the data obtained by online inquiry may not be comprehensive, such as lacking specific ticket sales, user number, etc.

[0167] For example, if the actual booking data, i.e., the number of users, of each flight cannot be obtained, it can be assumed that the number of booking users is randomly and uniformly distributed on the aircraft capacity in the interval [3 / 4, 1], i.e., if the aircraft A has a capacity of 100, it is assumed that the aircraft A has a random number of passengers between 75 and 100 each time; if the aircraft B has a capacity of 200, it is assumed that the aircraft B has a random number of passengers between 150 and 200 each time.

[0168] In some examples, the data preprocessing of the historical flight data obtained from the Internet includes merging multiple data of shared flights, or removing flight data between some airports with too few flights, which lacks researchability, etc.

[0169] Based on the above embodiments, the airline passenger itinerary recovery method for abnormal flight management provided by the present application not only focuses on passive recovery strategies after the occurrence of interference, such as the PRP-I model considering the transfer airport and the PRP-T model considering the multi-airport group region, but also focuses on active recovery strategies before the occurrence of interference, such as the PRP-F model introducing flexible tickets, which can solve the problem of existing researches only focusing on passive recovery strategies after the occurrence of interference, and existing lag recovery problems, etc.

[0170] In summary, the present application has at least the following advantages:

[0171] 1. For the issue of user trip recovery, detailed models are developed for the delay fees, downgrade fees, and denial-of-board fees involved in the problem, such as the PRP-T model, PRP-F model, and PRP-I model. This makes the resulting model closer to the real-world scenario and allows for a better balance of recovery strategies related to user reallocation.

[0172] 2. Construct a multi-airport cluster area, consider the transportation costs involved in rebooking users to other airports in the multi-airport cluster area, and transfer between two airports, so as to more accurately compare the costs of different recovery strategies and provide users with more options for trip recovery.

[0173] 3. Combining irregular flight management with revenue management research and introducing flexible ticketing can help further improve the overall economic benefits of user trip recovery strategies and aviation formulas; moreover, the introduction of flexible ticketing provides a certain degree of redundancy, which will be more robust before the disruption occurs and can reduce the impact of irregular flights on user trips.

[0174] 4. Based on the fitting analysis method, a large amount of historical ticketing data is fitted and user booking behavior is analyzed to obtain a better model of the relationship between the number of flexible air tickets sold and the flexible air ticket price.

[0175] 5. In view of the large amount of flight operation data across the country, the present invention uses column generation algorithm and large neighborhood search algorithm to significantly improve computational efficiency and save computation time; and the design and research of this algorithm can also be transferred to the solution of other combinatorial optimization problems in airline operation management.

[0176] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0177] Figure 8 This is a schematic diagram of an airline passenger itinerary recovery device according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. Figure 8 As shown, the airline passenger itinerary recovery device 80 includes: an acquisition module 81, an aggregation module 82, a construction module 83, and a recovery module 84. Wherein:

[0178] The acquisition module 81 is used to acquire flight information of irregular flights. The flight information includes the flight departure airport, flight departure time, flight arrival airport, flight arrival time, ticket price and number of tickets sold.

[0179] The aggregation module 82 is configured to aggregate, based on a clustering method, the adjacent airports of the flight departure airport according to the route space-time network to obtain a multi-airport group region, and the route space-time network is a relationship network between hub routes and between hub routes and non-hub routes.

[0180] The construction module 83 is configured to construct a first user itinerary recovery model corresponding to the abnormal flight according to the flight information and the multi-airport group region, and the first user itinerary recovery model is used to reflect the association between the ticketing revenue and the compensation fee and the transportation fee from the flight departure airport to the adjacent airport.

[0181] The recovery module 84 is configured to allocate flight resources for the user of the abnormal flight according to the first user itinerary recovery model.

[0182] In a possible implementation, the recovery module 84 can be specifically configured to: determine an optimal solution of the first user itinerary recovery model by using a preset algorithm, the preset algorithm including a column generation algorithm and a large neighborhood search algorithm; and allocate flight resources for the user of the abnormal flight according to the optimal solution.

[0183] In a possible implementation, the construction module 83 can be further configured to: determine a flight type according to the flight information, the flight type including a direct flight and a transfer flight; and construct a second user itinerary recovery model according to the flight information and the flight type, the second user itinerary recovery model being used to reflect the association between the ticketing revenue and the compensation fee when the direct flight is converted into the transfer flight. Correspondingly, the recovery module 84 can be further configured to: allocate flight resources for the user corresponding to the abnormal flight according to the second user itinerary recovery model.

[0184] In a possible implementation, the constructing module 83 can further be configured to: obtain flight information of the target flight, the flight information of the target flight comprising a flight departure airport, a flight departure time, a flight arrival airport, and a flight arrival time; and construct a third user itinerary recovery model according to the flight information of the target flight and a function model, the third user itinerary recovery model being used to reflect an association between a profit and a proportion of flexible tickets when the target flight is an abnormal flight, the function model being constructed based on a fitting analysis method according to historical booking data and being used to describe an association between the proportion of flexible tickets and a flexible ticket price ratio, wherein the flexible ticket price is less than a specific ticket price, and both a delay compensation and a downgrading compensation of a user corresponding to the flexible ticket are 0, and the flexible ticket corresponding to the user is refunded when the user refuses to board. Correspondingly, the airline passenger itinerary recovery apparatus further comprises a determining module (not shown) configured to: determine an influence coefficient of the proportion of flexible tickets and the flexible ticket price ratio on the recovery strategy; and determine the proportion of flexible tickets and the flexible ticket price ratio corresponding to the target flight based on the influence coefficient corresponding to different proportions of flexible ticket data and flexible ticket price ratios.

[0185] In a possible implementation, the determining module can be specifically configured to: determine the influence coefficient of the proportion of flexible tickets and the flexible ticket price ratio on the recovery strategy based on a sensitivity analysis method.

[0186] In a possible implementation, the constructing module 83 can be specifically configured to: combine at least two of the first user itinerary recovery model, the second user itinerary recovery model, and the third user itinerary recovery model to obtain a plurality of fourth user itinerary recovery models. Correspondingly, the recovery module 84 can be further configured to: for each fourth user itinerary recovery model in the plurality of fourth user itinerary recovery models, allocate flight resources for users corresponding to an abnormal flight according to the fourth user itinerary recovery model.

[0187] In a possible implementation, the route space-time network is constructed by: obtaining historical flight data, the historical flight data comprising a flight number, a tail number, a flight departure airport, a flight departure time, a flight arrival airport, a flight arrival time, a cabin, a number of tickets sold, and ticket price data; performing data preprocessing on the historical flight data to obtain target flight data, the data preprocessing being used to remove abnormal data, the abnormal data comprising flight data with a small number of flight frequencies between two airports; and constructing the route space-time network according to the target flight data

[0188] The airline passenger itinerary recovery apparatus provided in the embodiments of the present application has similar implementation principles and technical effects to those of the above-described embodiments, and details can be referred to the above-described embodiments, which will not be described herein again.

[0189] Figure 9is a structural schematic diagram of an electronic device provided by an embodiment of the present application. As shown in Figure 9 The electronic device 900 includes:

[0190] The processor 901, the memory 902, the communication interface 903, and the system bus 904.

[0191] The memory 902 and the communication interface 903 are connected with the processor 901 through the system bus 904 and complete communication with each other. The memory 902 is configured to store computer-executable instructions. The communication interface 903 is configured to communicate with other devices. The processor 901 is configured to execute the computer-executable instructions to perform the scheme of the airline passenger trip recovery method as in the method embodiments.

[0192] Specifically, the processor 901 can include one or more processing units. For example, the processor 901 can be a CPU, a digital signal processor (DSP), an application specific integrated circuit (ASIC), or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the application can be directly embodied as execution completed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0193] The memory 902 can be configured to store program instructions. The memory 902 can include a program storage area and a data storage area. The program storage area can store an operating system, at least one application required by a function (such as an algorithm solving function), and the like. The data storage area can store data created during use of the electronic device 900 (such as flight data of related airports in a multi-airport group area), and the like. In addition, the memory 902 can include a high-speed random access memory, and can also include a non-volatile memory such as at least one magnetic disk storage device, a flash memory device, a universal flash storage (UFS), or the like. The processor 901 executes various functional applications and data processing of the electronic device 900 by running the program instructions stored in the memory 902.

[0194] The communication interface 903 can provide a solution including 2G / 3G / 4G / 16G, etc. wireless communication applied on the electronic device 900. The communication interface 903 can receive electromagnetic waves by an antenna, and perform filtering, amplification, etc. processing on the received electromagnetic waves, and transmit to a modem processor for demodulation. The communication interface 903 can also amplify the signal modulated by the modem processor, and convert to electromagnetic wave radiation by the antenna. In some embodiments, at least part of the function modules of the communication interface 903 can be arranged in the processor 901. In some embodiments, at least part of the function modules of the communication interface 903 can be arranged in the same device as at least part of the modules of the processor 901.

[0195] The system bus 904 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus 904 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is shown in the figure, but it does not mean that there is only one bus or only one type of bus.

[0196] It should be noted that the number of the memory 902 and the processor 901 is not limited in the embodiments of the present application, and each of them can be one or more. Figure 9 Take one as an example for illustration; the memory 902 and the processor 901 can be connected by various wired or wireless ways, for example, connected by a bus. In actual application, the electronic device 900 can be various forms of computers or mobile terminals. Among them, the computer is, for example, a laptop computer, a desktop computer, a workbench, a server, a blade server, a mainframe computer, etc.; the mobile terminal is, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device and other similar computing devices.

[0197] Those skilled in the art can understand that, Figure 9 The electronic device shown does not constitute a limitation on the electronic device, and can include more or fewer components than shown, or combine certain components, or different component arrangements.

[0198] The embodiments of the present application also provide a computer readable storage medium, which stores computer execution instructions, and when the computer execution instructions are executed, the airline passenger itinerary recovery method is realized.

[0199] The embodiments of the present application also provide a computer program product, which includes a computer program, and when the computer program is executed, the airline passenger itinerary recovery method is realized.

[0200] The embodiment of the present application also provides a chip for running instructions, which is used for executing the airline passenger trip recovery method according to any one of the method embodiments.

[0201] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of the related data need to comply with the relevant laws, regulations and standards of the country and region, and provide corresponding operation entrances for the user to choose authorization or refusal.

[0202] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the application cover any and all variations of the application that come within the scope of the general concept of the application and that the claims be interpreted not to be limited to the specific examples described herein. The specification and examples are to be considered exemplary only, with the true scope and spirit of the application indicated by the following claims.

[0203] It should be understood that the application is not limited to the precise construction that has been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application. The scope of the application is limited only by the claims that follow.

Claims

1. A method for restoring passenger travel itineraries on an airline, characterized in that, include: Obtain flight information for irregular flights, including the flight's departure airport, departure time, arrival airport, arrival time, ticket price, and number of tickets sold. Based on clustering methods, the neighboring airports of the flight departure airport are aggregated according to the route spatiotemporal network to obtain a multi-airport cluster region. The route spatiotemporal network is a network of relationships between hub routes and between hub routes and non-hub routes. Based on the flight information and the multi-airport cluster area, a first user trip recovery model is constructed for the abnormal flight. The first user trip recovery model reflects the relationship between ticket revenue, compensation fees, and transportation costs from the flight's departure airport to a nearby airport. The first user trip recovery model has an objective function and constraints. The objective function is to maximize total profit, and its calculation includes ticket revenue, compensation fees, and transportation costs. The compensation fees include rebooking fees and denied boarding fees, and the transportation costs are the costs of transporting the user from the original airport to another airport. The constraints are as follows: a cancelled flight cannot carry any users; a user cannot be assigned to a flight or denied boarding; aircraft capacity is limited; the origin-destination city pair of the new flight is the same as that of the original flight; the departure time of the new flight is ensured to be no earlier than the departure time of the original flight, and when ensuring that the departure time of the new flight is no earlier than the departure time of the original flight, the travel time between the two airports must be taken into account; and a maximum arrival delay is defined. Based on the first user trip recovery model, flight resources are allocated to users with abnormal flights.

2. The method for restoring airline passenger travel itineraries according to claim 1, characterized in that, The step of allocating flight resources to users with abnormal flights based on the first user trip recovery model includes: The optimal solution of the first user trip recovery model is determined by a preset algorithm, which includes a column generation algorithm and a large neighborhood search algorithm. Based on the optimal solution, flight resources are allocated to users of the abnormal flights.

3. The method for restoring airline passenger travel itineraries according to claim 1 or 2, characterized in that, Also includes: Based on the flight information, the flight type is determined, including direct flights and connecting flights; Based on the flight information and the flight type, a second user trip recovery model is constructed. The second user trip recovery model is used to reflect the relationship between ticket revenue and compensation costs when a direct flight is converted into a connecting flight. Based on the second user trip recovery model, flight resources are allocated to the users corresponding to the abnormal flights.

4. The method for restoring airline passenger travel itineraries according to claim 3, characterized in that, Also includes: Obtain the flight information of the target flight, which includes the flight's departure airport, departure time, arrival airport, and arrival time. Based on the flight information and function model of the target flight, a third user trip recovery model is constructed. The third user trip recovery model is used to reflect the relationship between the profit and the proportion of flexible tickets and the proportion of flexible ticket prices when the target flight is an irregular flight. The function model is constructed based on historical booking data using fitting analysis method and is used to describe the relationship between the proportion of flexible tickets and the proportion of flexible ticket prices. In this model, the flexible ticket price is less than the specific ticket price, and the delay compensation and downgrade compensation for the user corresponding to the flexible ticket are both 0. The user corresponding to the flexible ticket will receive a refund of the flexible ticket fee when boarding is refused. Determine the impact coefficients of the proportion of flexible air tickets and the proportion of flexible air ticket prices on the recovery strategy; Based on the influence coefficients corresponding to different flexible ticket quantity ratios and flexible ticket price ratios, the flexible ticket quantity ratio and flexible ticket price ratio for the target flight are determined.

5. The method for restoring airline passenger travel itineraries according to claim 4, characterized in that, The determination of the impact coefficients of the proportion of flexible ticket quantity and the proportion of flexible ticket price on the recovery strategy includes: Based on sensitivity analysis, the impact coefficients of the proportion of flexible ticket quantity and the proportion of flexible ticket price on the recovery strategy are determined.

6. The method for restoring airline passenger travel itineraries according to claim 4, characterized in that, Also includes: At least two of the first user trip recovery model, the second user trip recovery model, and the third user trip recovery model are combined to obtain multiple fourth user trip recovery models; For each of the multiple fourth-user trip recovery models, flight resources are allocated to the user corresponding to the abnormal flight according to the fourth-user trip recovery model.

7. The method for restoring airline passenger travel itineraries according to claim 1 or 2, characterized in that, The route spatiotemporal network is constructed in the following way: Obtain historical flight data, which includes flight number, tail number, departure airport, departure time, arrival airport, arrival time, cabin class, number of tickets sold, and ticket price data; The historical flight data is preprocessed to obtain the target flight data. The data preprocessing is used to remove abnormal data, including flight data with a low number of flights between the two airports. Based on the target flight data, construct a spatiotemporal network of flight routes.

8. An airline passenger itinerary recovery device, characterized in that, include: The acquisition module is used to acquire flight information of irregular flights, including the flight departure airport, flight departure time, flight arrival airport, flight arrival time, ticket price, and number of tickets sold. The aggregation module is used to aggregate the neighboring airports of the flight departure airport based on the clustering method and the route spatiotemporal network to obtain a multi-airport cluster area. The route spatiotemporal network is a relationship network between hub routes and between hub routes and non-hub routes. A construction module is used to construct a first user trip recovery model corresponding to the abnormal flight based on the flight information and the multi-airport cluster area. The first user trip recovery model reflects the correlation between ticket revenue, compensation fees, and transportation costs from the flight's departure airport to a nearby airport. The first user trip recovery model has an objective function and constraints. The objective function is to maximize total profit and includes ticket revenue, compensation fees, and transportation costs in its calculation. The compensation fees include rebooking fees and denial-of-boarding fees, and the transportation costs are the costs of transporting the user from the original airport to another airport. The constraints are as follows: a cancelled flight cannot carry any users; a user cannot be assigned to a flight or denied boarding; aircraft capacity is limited; the origin-destination city pair of the new flight is the same as that of the original flight; the departure time of the new flight is ensured to be no earlier than the departure time of the original flight, and when ensuring that the departure time of the new flight is no earlier than the departure time of the original flight, the travel time between the two airports must be taken into account; and a maximum arrival delay is defined. The recovery module is used to allocate flight resources to users with abnormal flights based on the first user trip recovery model.

9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store program instructions; The processor is configured to invoke the program instructions to execute the airline passenger itinerary recovery method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed, are used to implement the airline passenger itinerary recovery method as described in any one of claims 1 to 7.

Images