A two-stage delay recovery method based on interactions between airlines and passengers

Abstract

The application discloses a two-stage delay recovery method based on interaction between airlines and passengers, including static and dynamic two stages, which can pre-specify and dynamically adjust the flight delay recovery plan. The static stage takes the aircraft flight as the recovery object, establishes a flight recovery model with the minimum sum of direct recovery cost and indirect recovery cost of the airline as the target, obtains the aircraft flight recovery plan, and publishes the scheme in the form of a recommended scheme to the passengers. The dynamic stage is based on real-time received passenger feedback, and adjusts the passenger itinerary recovery scheme in real time to minimize the passenger delay loss, and generates the final flight and passenger recovery scheme. The application can not only balance the economic cost of the airline and the delay loss of the passenger, but also consider the influence of the passenger itinerary recovery willingness on the preparation of the recovery plan, and is more in line with the actual situation of flight delay. It has wide application prospect and practical significance.

Description

Technical Field

[0001] This invention belongs to the field of flight recovery technology and relates to a two-stage delay recovery method based on the interaction between airlines and passengers. Background Technology

[0002] With the rapid development of the national economy, the rapid growth in passenger travel demand has brought enormous challenges to the air transport system. Flight delays directly lead to the cancellation or suspension of passenger trips, causing significant negative impacts on their travel plans and frequently resulting in conflicts between passengers and airlines. In most cases, the source of passengers' negative feelings is not the flight delay itself, but rather dissatisfaction with the airline's handling and support processes. Therefore, how to efficiently restore disrupted flights and passengers, and how to complete the task of passenger transport and timetable restoration in the shortest possible time by adjusting available resources, are urgent problems that airlines need to solve.

[0003] In practice, flight delay recovery is typically handled unilaterally by the airline, focusing solely on internal resource coordination and control to minimize economic losses. The final recovery plan is then released directly to passengers. This approach has the following limitations: ① There is a lack of information interaction between passengers and the airline. Passengers passively accept a single recovery plan provided by the airline, and their right to choose the recovery option is not guaranteed. ② This recovery method fails to consider the heterogeneity and randomness of passenger preferences, making it difficult to adapt to the actual delay environment. This can easily lead to passenger dissatisfaction, loss of high-quality passengers, and damage to the airline's reputation.

[0004] Advances in science and technology, along with the development of information technology in civil aviation, have made passenger travel more intelligent. Technological innovation has broadened the channels through which passengers obtain flight information and provided the hardware conditions for information exchange. Passengers can proactively collect information through platforms such as mobile apps and WeChat official accounts, and submit restoration requests through terminal systems. However, these information systems are currently only at the stage of publishing delay information, which is insufficient to fully meet passengers' needs for information on flight delays, and airlines cannot obtain real-time feedback from passengers on their willingness to participate in restoration plans.

[0005] Research on flight delay recovery is relatively mature. Relevant patents and academic journal articles can be found through publicly available information, including the invention patent "A Flight Recovery Method Considering Passenger Preferences and Bounded Rationality" (publication number CN112862258B), the invention patent "An Integrated Recovery System for Irregular Flights, Passengers and Maintenance Personnel" (publication number CN112819317B), "Aircraft and Passenger Recovery Model Based on Flight Delay" (Journal of Central China Normal University (Natural Science Edition) 2015, No. 49) by Jiang Mao et al., and "Modeling and Algorithm Research on Flight Delay Recovery" (Journal of Transportation Systems Engineering and Information 2018, No. 18) by Huang Junsheng et al.

[0006] The aforementioned research has largely solved the modeling and algorithmic problems of integrating aircraft, crew, and passengers in the recovery process. However, it lacks an effective mechanism to combine the selection of passenger itinerary recovery schemes with the formulation of flight delay recovery plans, failing to truly leverage the subjective initiative of passengers in delay recovery. Furthermore, the recovery process rarely considers the heterogeneity of passengers in different cabin classes regarding delay losses and cabin capacity, resulting in deviations from real-world delay scenarios. Currently, most algorithms involved in flight recovery problems are heuristic algorithms, which have cumbersome solution steps and often fail to yield accurate solutions, thus limiting their practical applicability and leaving considerable room for improvement in efficiency. Summary of the Invention

[0007] Purpose of the Invention: To overcome the shortcomings of existing methods for resuming flight disruptions, such as cumbersome operations, rigid processes, and an inability to balance the interests of both airlines and passengers, this invention provides a two-stage recovery method based on interaction between airlines and passengers. It aims to fully utilize information technology to dynamically formulate and adjust delay recovery plans based on thorough communication between airlines and passengers. Through a "pre-planning - feedback of intent - dynamic adjustment" mechanism, passenger participation in the delay recovery process is ensured while protecting the interests of airlines, thereby improving the efficiency and quality of delay recovery.

[0008] Technical solution: To achieve the above objectives, the technical solution adopted by this invention is as follows:

[0009] A two-stage delay recovery method based on interaction between airlines and passengers, wherein the two stages are a static planning stage and a dynamic planning adjustment stage, and the static planning stage includes the following steps:

[0010] Step S1: Establish a flight information database, an airport information database, a delay scenario database, and a passenger itinerary database; obtain information data on affected flights and passengers, and load them into the flight information database and the passenger itinerary database respectively;

[0011] Step S2: Initialize the flight schedule for each aircraft to obtain the initial set of aircraft paths;

[0012] Step S3: Establish the aircraft flight recovery model minF1 for the static planning phase as follows:

[0013]

[0014] Among them: Dc f This represents the operational costs of flight f, including fuel surcharges, service fees, and parking fees; Cc f Indicates the cancellation cost of flight f; y f Indicates whether flight f has been cancelled; Dp k This represents the unit time value of passengers in class k, where k=1 represents economy class passengers, k=2 represents business class passengers, and k=3 represents first class passengers. This indicates the number of passengers of class k on flight f; This indicates the delay time of flight f operated by aircraft i; This indicates whether flight f is in path r; where A, R, F, and K represent the total number of aircraft, the total number of paths, the total number of flights, and the number of passenger categories, respectively. Defined as a decision variable, indicating whether aircraft a will follow path r;

[0015] Constrain the recovery model minF1:

[0016]

[0017]

[0018]

[0019]

[0020] Equation (1) is the aircraft coverage constraint for flight f, which means that if flight f is canceled, all paths containing the flight cannot be executed; otherwise, if flight f is not canceled, there is one and only one path containing the flight that is executed by one and only one aircraft.

[0021] Equation (2) is the unique path constraint for aircraft i, meaning that aircraft a can be assigned at most one path.

[0022] Equation (3) represents the flight delay time as the difference between the actual departure time and the scheduled departure time, where... T refers to the actual takeoff time of aircraft a when performing flight f. f This refers to the original scheduled departure time of flight f;

[0023] Equation (4) represents the value constraints of the decision variables;

[0024] Step S4: Retrieve the flight information database, solve the recovery model minF1 of aircraft and flights based on the column generation algorithm, obtain the correspondence between aircraft and flights, and generate a recovery plan;

[0025] Step S5: Identify passengers whose trips are interrupted due to flight cancellations and connection failures, and divide the passengers into z OD traffic flows based on trip attributes, trip locations, and times; match the recovery plan with the passengers on the flight, and summarize the recovery plans for passengers in each OD traffic flow; send the recovery plan to the passengers via SMS, and wait for their feedback on whether to accept, request a refund, or seek other recovery plans.

[0026] The recovery plan adjustment phase includes the following steps:

[0027] Step D1: The affected passenger responds by accepting the situation, requesting a refund, or seeking other recovery plans;

[0028] Step D2: The airline receives feedback from affected passengers in real time and completes the dynamic adjustment phase of the plan based on the feedback; the time for releasing the recovery plan information is set as t0, and an adjustment time window is set as Δt. The feedback from affected passengers is adjusted and updated in units of adjustment time window Δt; at the end of each adjustment time window Δt, the received feedback is classified and statistically analyzed. These represent the number of passengers in category k who are affected by the disruption during itinerary i within the time window t+Δt, respectively: those who accept the ticket, those who request a refund, and those who seek other recovery plans.

[0029] Step D3: For passengers affected by the disruption who accept the recovery plan, proceed to step D4; for passengers affected by the disruption who voluntarily request a refund, proceed to step D5; for passengers affected by the disruption who seek other recovery plans, proceed to step D6.

[0030] Step D4: For passengers affected by the disruption who have accepted the recovery plan, arrange their travel itineraries according to the recovery plan;

[0031] Step D5: For passengers who have voluntarily cancelled their tickets and have been disturbed, remove these passengers from the recovery plan during the static planning phase.

[0032] Step D6: Call the passenger feedback insertion algorithm to insert data from the feasible itinerary set I. fe Search the data to find feasible trips that meet the remaining capacity limit and insert the affected passenger in the OD flow; if no feasible trips can be inserted, the result is fed back to the affected passenger, and the time window is moved forward by Δt, returning to step D2, until the affected passenger accepts the recovery plan or gets a refund.

[0033] Step D7: The airline selects the final recovery plan and outputs it in the form of a set of flight sequences {f1,f2,f3,...} executed by each aircraft and a set of passenger itinerary sequences {(f1,f2),f3,...}.

[0034] Preferably, during the dynamic adjustment phase of the plan, the length of the adjustment time window Δt is set according to the following rules:

[0035] T edt This indicates the end time of the delay recovery work, set by the airline based on the scale of the delay; T cur Indicates the current time; when T edt -T cur When T ≥ 3h, Δt is set to 1h; when T edt -T cur When T ∈ (1,3), Δt is set to 30min; when T edt -T cur When ≤1h, Δt is set to 10min, and at a distance T edt After 30 minutes, the collection of feedback from disturbed passengers will cease. Passengers who do not provide feedback will be considered to have complied with the final recovery plan in step D7.

[0036] Preferably, the dynamic adjustment phase of the plan follows the following three rules:

[0037] (1) Passenger Removal Mechanism

[0038] At the start of the adjustment window, the passengers who voluntarily canceled their tickets in each OD flow in the previous time window are counted and removed from their original itineraries. The itineraries that these passengers were originally on are then selected by other passengers.

[0039] (2) Trip priority resumption mechanism

[0040] During the dynamic adjustment phase of the plan, passengers who submit feedback earlier have the right to priority in matching itineraries. Once a passenger's restored itinerary is locked, other passengers' itineraries will be restored within the subsequent adjustment window, provided that the itineraries of these passengers remain unchanged.

[0041] (3) Random trip allocation mechanism

[0042] When the capacity of feasible itineraries exceeds the total number of passengers in the OD traffic flow who need to change their tickets, only some passengers' change requests can be fulfilled. In this case, passengers who change their tickets to that itinerary will be randomly selected, and other feasible itineraries will continue to be searched for for the remaining passengers.

[0043] Preferably, when performing step D2, the objective function minF2 for stroke recovery is set as follows:

[0044] Among the decision variables The variable is 0-1, representing the number of passengers of type k who are disturbed and assigned from trip i to trip i'; This represents the total number of passengers of type k in the original itinerary i; delt i' Indicates the delay time of trip i'; D k U k C k These represent the unit time delay loss, violation of intent loss, and trip cancellation loss for passengers of type k, respectively.

[0045] Constrain the objective function min F2 for stroke recovery:

[0046]

[0047]

[0048]

[0049]

[0050] Equation (5) represents the time coherence constraint. This indicates the departure time of the subsequent flight in the connecting flight i'. t represents the arrival time of the preceding flight. MCT This represents the shortest transfer time, and the constraint states that the connection time between two flights in a connecting passenger's itinerary must not be less than the shortest transfer time; Equation (6) is the spatial connection constraint, dep i'2 arr indicates the departure airport for subsequent flights in itinerary i'. i'1 Let i' represent the arrival airport of the preceding flight in itinerary i'. This constraint indicates that the departure airport of subsequent flights in a connecting flight is the same as the arrival airport of the preceding flight; Equation (7) is the itinerary capacity constraint. This indicates the maximum passenger capacity of cabin class k for flight f. This represents the original k-type passengers in itinerary i'. The variable is 0-1, indicating whether the itinerary i' contains flight f. This constraint means that the number of passengers of class k in itinerary i' shall not exceed the minimum value of the cabin capacity of flight k in the itinerary; Equation (8) represents the value constraint of the variable.

[0051] Preferably, the passenger feedback insertion algorithm includes the following steps:

[0052] Step I-1: Retrieve information about the affected passenger's original itinerary i, including whether it is a connecting flight, and the origin and destination locations dep. i with arr i Original flight f's estimated departure time With arrival time And cabin class k; if the disturbed passenger is a direct flight passenger, follow method I-1; if the disturbed passenger is a connecting flight passenger, follow method I-2;

[0053] Step I-2: If a direct flight cannot be restored in step I-11, send an inquiry to ask if the passenger accepts the connecting flight. The response time is t2. If the affected passenger accepts, proceed with method I-2. If the affected passenger does not accept, the passenger will be considered to have received a refund.

[0054] Step I-3: Release the feasible itineraries obtained after the search to the affected travelers and solicit their feedback;

[0055] Method I-1 includes the following steps:

[0056] Step I-11: In the final recovery plan of step D7, search for the departure airport as dep. i The landing airport is ARR. i Flights;

[0057] Step I-12: Filtering The feasible flights are added to the set of feasible direct flight routes I. df ,in This indicates that the departure times of flight f are arranged in chronological order.

[0058] Step I-13: Check each item in sequence. df Check if the flight meets the flight capacity limit; if the capacity of this cabin class on the same flight is insufficient to accommodate all affected passengers, check the capacity of other cabin classes. If they meet the requirements, retain the itinerary in group I. df If the itinerary cannot accommodate all affected passengers, then at meeting point I... df Delete the trip, check the next trip, and continue searching until all flights have been searched.

[0059] Step I-14: Output the set I after the search is complete. df ;

[0060] Method I-2 includes the following steps:

[0061] Step I-21: In the final recovery plan of step D7, search for the departure airport as dep. i and The flights are designated as set F. fea1 The landing airport is ARR. i and The flights are designated as set F. fea2 Remove the intersection of sets Direct flights within the country;

[0062] Step I-22: From set F fea1The search returned the p-th flight f. p From set F fea2 The search returned the qth flight f q Check whether the two flights meet the time and space connection constraints, determine the found connecting flights as a feasible itinerary i, complete one search and loop, let the set of feasible connecting itineraries be I. af ;

[0063] Step I-23: Check each item sequentially. af Whether the connecting flight meets the capacity limit conditions: If the connecting flight includes two segments, the number of passengers affected by changes in their itinerary must not exceed the smaller of the capacities of the two flight segments, i.e., ∑ i,i'∈I t ii' +Psg i' ≤min{Cap f1 Cap f2 If the capacity of the flight's cabin class is insufficient to accommodate all affected passengers, then the capacity of other cabin classes will be checked; if the itinerary still cannot accommodate all affected passengers, then the passengers will be accommodated at meeting point I. af Delete the trip and check the next connecting trip until all flights in the set have been searched.

[0064] Step I-24: Output I after the search is complete af gather.

[0065] Compared with existing delay recovery techniques, the advantages of this invention are as follows:

[0066] On the one hand, existing flight delay recovery methods can be mainly divided into two categories: phased recovery and integrated recovery. Phased recovery restores the aircraft, flights, and passengers separately in sequence, with each part independently searching for the optimal solution. This may result in suboptimal solutions or even infeasible recovery plans, leading to a failure to match passenger itineraries with flight schedules. Integrated recovery involves cumbersome steps, a large problem scale, and high requirements for intelligent algorithms, making it impractical in real-world applications. The two-phase recovery method based on inter-agent interaction proposed in this invention is not limited by the limitations of manual scheduling. Through an information technology platform, it enhances the interaction between aircraft recovery and passenger itineraries, resulting in stronger real-time performance in actual delay handling scenarios. This method will facilitate the implementation of global recovery that balances the interests of multiple parties, providing tactical decision support for airlines' delay recovery efforts.

[0067] On the other hand, existing recovery methods only consider the interests of airlines and do not pay enough attention to the wishes of passengers. Passengers can only passively receive flight arrangements after delays are resolved, which easily breeds passenger dissatisfaction and negatively impacts the long-term profitability and sustainable development of airlines. As the main source of profit in the aviation business, the travel experience of passengers should be given full attention by airlines. Therefore, this invention enables passengers to participate in the flight delay recovery process based on their actual needs through information exchange and feedback between airlines and passengers. This is conducive to the flight recovery plan fully reflecting the interests of both airlines and passengers. In addition, this invention dynamically adjusts and corrects the recovery plan through real-time passenger feedback, updates flight capacity in a timely manner, and provides limited seats to passengers with genuine needs, significantly reducing resource waste during the recovery process. Attached Figure Description

[0068] Figure 1 Here is the main flowchart of the two-stage delay recovery method proposed in this invention;

[0069] Figure 2 This is a flowchart of the solution process for the first-stage flight recovery model in this invention;

[0070] Figure 3 This is a flowchart of the passenger preference insertion algorithm in this invention. Detailed Implementation Plan

[0071] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0072] A two-stage delay recovery method based on airline-passenger interaction, such as Figure 1 As shown, the process includes static and dynamic phases. First, in the static phase, the basic delay situation is assessed, and information on the number and timing of aircraft, flights, and passengers is compiled to prepare data for subsequent delay recovery. Next, flight recovery is implemented with the objective of minimizing airline delay losses, resulting in a flight recovery plan and a corresponding preliminary passenger itinerary recovery plan, which is then released to passengers. In the dynamic phase, adjustment time windows are set, passenger feedback is categorized and statistically analyzed, and passenger itineraries are dynamically adjusted based on this feedback, with the goal of minimizing passenger delay losses. Finally, a flight and passenger itinerary recovery plan is generated.

[0073] Specifically, the static design phase includes the following steps:

[0074] S1. Establish a flight information database, an airport information database, a delay scenario database, and a passenger itinerary database. Obtain information data on disrupted flights and passengers, load them into the corresponding databases, and configure passenger parameters and model parameters.

[0075] S3. Initialize the flight schedule of each aircraft to define the flight path ROF (RouteofFlight) as a path consisting of one or more flights executed by the same aircraft that meet the time and space connection conditions.

[0076] S4. Based on the objective function of minimizing the sum of direct and indirect economic losses for airlines, and with airport capacity limitations and node aircraft flow balance constraints as constraints, a first-stage aircraft and flight recovery model is established based on the minimum-cost multi-goods network flow problem. The first stage is aircraft and flight recovery, which essentially involves assigning aircraft to flights that meet operational constraints, minimizing the sum of direct and indirect economic losses to airlines caused by delays, i.e., minimizing the total cost of the aircraft path set.

[0077] The model follows the following assumptions:

[0078] (1) The recovery method is applicable to large-scale flight delay scenarios, including flight delays caused by temporary airport closures and the cessation of use of some aircraft.

[0079] (2) No priority is given to flight recovery, and no distinction is made between domestic and international flights during the recovery process.

[0080] (3) The delay time shall not exceed the maximum allowable delay time of 3 hours. If the estimated delay time of a flight exceeds 3 hours, the flight shall be cancelled directly.

[0081] (4) Consider the capacity limitations caused by different aircraft types for different flights, and consider the aircraft type matching limitations when exchanging aircraft.

[0082] The objective function of the model is as follows:

[0083]

[0084] The decision variables include: Indicates whether path r is executed by aircraft i; Indicates whether flight f is in path r; y f Indicates whether flight f has been cancelled; where Dc f Table Cc shows the operational costs of flight f (including fuel surcharges, service fees, and parking fees, which vary depending on the aircraft type); f Dp represents the cancellation cost of flight f (related to passenger numbers and ticket price); k This represents the unit time value of passengers in class k (k=1 for economy class passengers; k=2 for business class passengers; k=3 for first class passengers). This indicates the delay time of flight f operated by aircraft i. This indicates the number of passengers in category k on flight f; the first phase of recovery, from the airline's perspective, involves rescheduling flights and re-arranging travel for passengers whose journeys have been disrupted, with the goal of minimizing costs associated with flight delays, rebookings, and cancellations.

[0085]

[0086]

[0087]

[0088]

[0089] The model contains three constraints: Equation (1) is the aircraft coverage constraint, which means that if a flight is canceled, all paths containing that flight cannot be executed; otherwise, if the flight is not canceled, there is one and only one path containing that flight that is executed by one and only one aircraft; Equation (2) is the uniqueness constraint of the aircraft path, which means that each aircraft can be assigned at most one path, and there can be aircraft that are not assigned a path; Equation (3) means that the flight delay time is the difference between the actual departure time and the planned departure time; Equation (4) means the value constraint of the decision variable.

[0090] S5. Retrieve flight information data, solve the aircraft flight recovery model based on the column generation algorithm, and generate corresponding passenger itinerary recovery arrangements. This includes the following steps:

[0091] S5-1. Set dominance rules to perform preliminary screening of In(R) and reduce model size. Rule 1: The number of aircraft exchanges in ROF should be minimized. Rule 2: Considering that delay time should be as short as possible, ROFs whose starting point is between the starting points of two existing flights are eliminated;

[0092] S5-2. Based on the constrained master problem (RMP), solve the RMP to generate an initial set of aircraft paths and add it to Fn(R). Record the minimum cost at this point as the initial objective value.

[0093] S5-3. Solve the dual subproblem to obtain the reduction in total cost of ROF in In(R). Compare the total cost value when a feasible solution is obtained with the initial target value. If it is less than the target value, add the obtained feasible ROF to Fn(R).

[0094] S5-4. Iterate through the subproblems to generate the final Fn(R).

[0095] S5-5. Generate a new correspondence between aircraft and flights, i.e., a flight recovery plan, and generate a corresponding passenger travel recovery scheme based on the plan.

[0096] S6. Identify passengers whose trips are interrupted due to flight cancellations and connection failures, and divide the passengers into z OD flows based on trip attributes, trip locations, and times; match the flight recovery plan with the passengers on the flight, and summarize the preliminary plans for trip recovery for passengers in each OD flow; send the plan to the affected passengers via SMS, and wait for the passengers' feedback on their willingness to accept, refund, or seek other recovery plans;

[0097] The following three rules shall be followed during the adjustment phase of the recovery plan:

[0098] (1) Passenger Removal Mechanism

[0099] At the start of the adjustment window, passengers who voluntarily cancelled their tickets in each OD flow within the previous time window are counted and removed from their original itineraries. Their original itineraries can then be selected by other passengers; for example, in the... At the start time of the adjustment time window, the statistics are as follows: For passengers who choose to cancel their tickets within the time window, if their original itinerary has been cancelled, only cabin class and fare information will be recorded; if it has not been cancelled, the corresponding units of the itinerary capacity will be released, starting from the [date missing]. Starting from this time window, other passengers can be assigned to this itinerary;

[0100] (2) Trip priority resumption mechanism

[0101] When dynamically adjusting passenger travel itineraries, passengers who submitted feedback earlier have priority in itinerary matching. Once a passenger's travel itinerary is locked, other passengers' travel itineraries must be restored within subsequent adjustment windows, ensuring that their itineraries remain unchanged. For example, in the [missing information - likely a date or timeframe]... If, within the adjustment time window, itinerary i1 has already been occupied by passengers who have changed their tickets, and there is no remaining capacity, then... Passengers within the adjusted time window may not be reassigned to i1 itineraries, even if the itinerary meets their rebooking requirements.

[0102] (3) Random trip allocation mechanism

[0103] When the available itinerary capacity exceeds the total number of passengers in the OD flow with rebooking requests, only some passengers' rebooking needs can be met. In this case, passengers rebooking to that itinerary will be randomly selected, and other feasible itineraries will continue to be searched for the remaining passengers. For example, if the total number of passengers in OD flow z is n... z The remaining capacity of feasible path i' is n. i n i <n z Randomly select n i One passenger was assigned to itinerary i', and the number of passengers in z was reduced to n. z -n i name.

[0104] Specifically, the dynamic adjustment phase includes the following steps:

[0105] D1. Passengers provide feedback on the recommendations based on their own preferences and acceptance of the recommended options;

[0106] D2. The airline receives passenger feedback in real time and adjusts the second phase plan based on the feedback.

[0107] The recommended travel plan information will be released at time t0, and an adjustment time window will be set as Δt. The travel plan will be adjusted and updated in units of the adjustment time window. Considering that passengers check flight delay information more frequently closer to the estimated departure time, and their response speed is faster, the length of Δt will be the same as T. cur Distance from estimated departure time T edt Establish a mapping relationship over time. When T edt -T cur When T ≥ 3h, Δt is set to 1h; when T edt -T cur When T ∈ (1,3), Δt is set to 30min; when T edt -T cur When ≤1h, Δt is set to 10min, and at a distance T edt Passenger feedback collection will cease after 30 minutes. Passengers who do not provide feedback will be considered to be complying with the flight resumption plan and will be able to resume their travel accordingly.

[0108] At the end of each time window, the received passenger feedback is categorized and statistically analyzed. These represent the number of passengers in category k within the time window t+Δt who accept arrangements, voluntarily cancel tickets, or seek other rescheduling options for itinerary i.

[0109] D3. For passengers who accept the recovery plan, proceed to step D4; for passengers who voluntarily request a refund, proceed to step D5; for passengers seeking other solutions, proceed to step D6.

[0110] D4. Arrange passenger itineraries according to the pre-planned schedule, restore the fixed itineraries for this group of passengers, and keep the remaining capacity of the itinerary unchanged;

[0111] D5. Remove these passengers from the recovery itineraries allocated in Phase 1, and release the remaining K cabin capacity for those itineraries. Units;

[0112] D6. Establish a passenger itinerary recovery model with the objective function of minimizing passenger delay losses. Invoke the passenger intention insertion algorithm, and based on the departure time and connecting flight information of the passengers' pre-planned restored itineraries, extract information from the feasible itinerary set I. feSearch the data to find feasible passenger trips that meet the remaining capacity limit and can be inserted into the OD flow; if no feasible trips can be inserted, the result is fed back to the passenger, the time window is moved forward by Δt, and the process returns to step D2 until the passenger accepts the allocation plan or gets a refund.

[0113] D7. The airline selects the final recovery plan and outputs it in the form of a set of flight sequences {f1,f2,f3,...} executed by each aircraft and a set of passenger itinerary sequences {(f1,f2),f3,...}.

[0114] The passenger itinerary recovery model in step D5 follows the following assumptions:

[0115] (1) Disturbed passengers are defined as connecting passengers whose itinerary connection fails due to flight delays, and passengers whose flights are cancelled during their itinerary.

[0116] (2) Without considering the price difference when passengers purchase tickets, it is assumed that passengers of the same class of the same flight have the same ticket purchase cost; since the decision to upgrade or downgrade to resume the trip is based on the passenger's own wishes, the loss caused by changing the class of the flight is not reflected in the objective function.

[0117] (3) For passengers who only consider direct flights and connecting flights that only include one transfer, connecting flights will not be arranged for direct flight passengers during the first phase of recovery. Direct flight flights can be arranged for connecting flight passengers, and passengers will not need to refund or pay the price difference.

[0118] The objective function of the passenger travel recovery model is as follows:

[0119]

[0120] From the passenger's perspective, three types of recovery costs are considered: losses due to delays, losses due to breach of intent, and losses due to trip cancellation. Among these, the decision variables... The variable is 0-1, representing the number of passengers of type k who are disturbed and assigned from trip i to trip i'; This represents the total number of passengers of type k in the original itinerary i; delt i' Indicates the delay time of trip i'; D k U k C k These represent the unit time delay loss, violation of intent loss, and trip cancellation loss for passengers of type k, respectively.

[0121] The model includes four constraints:

[0122]

[0123]

[0124]

[0125]

[0126] Equation (5) represents the time connection constraint, indicating that the difference between the departure time of a subsequent flight and the arrival time of a preceding flight in a connecting passenger's itinerary must not be less than the passenger's shortest transfer time; Equation (6) represents the spatial connection constraint, indicating that the departure airport of a subsequent flight in a connecting passenger's itinerary is the same as the arrival airport of a preceding flight; Equation (7) represents the itinerary capacity constraint. This indicates the maximum passenger capacity of cabin class k for flight f. This represents the original k-type passengers in itinerary i'. The variable is 0-1, indicating whether the itinerary i' contains flight f. This constraint means that the number of passengers of class k in itinerary i' shall not exceed the minimum value of the cabin capacity of flight k in the itinerary; Equation (8) represents the value constraint of the variable.

[0127] The passenger preference insertion algorithm in step D5 includes the following steps:

[0128] Step I1: Retrieve information about the original itinerary i for this group of passengers, including whether it is a connecting flight and the origin and destination of the itinerary. i with arr i Original flight f's estimated departure time With arrival time And cabin class k. If the passenger is a direct flight passenger, execute algorithm I-1; if the passenger is a connecting flight passenger, execute algorithm I-2.

[0129] Step I2: If I-11 cannot find a direct flight to resume, send an inquiry to ask if the passenger accepts the connecting flight. The response time is t2. If the passenger accepts, execute algorithm I-12. If the passenger does not accept, it is considered that the passenger has refunded the ticket.

[0130] Step I3: Release the feasible itineraries obtained after the search to travelers and solicit their feedback.

[0131] The direct flight range search algorithm I-1 process is as follows:

[0132] Step I-11: Search for the departure airport as dep in the flight recovery plan table. i The landing airport is ARR. i Flights;

[0133] Step I-12: Filtering The feasible flights are added to the set of feasible direct flight routes I. df Arranged in chronological order;

[0134] Step I-13: Check them one by one in order Check if the flight meets the flight capacity restrictions. If the capacity of this cabin class on the same flight is insufficient to accommodate all passengers, check the capacity of other cabin classes. If they meet the requirements, the itinerary will be reserved for flight I. df If the itinerary cannot accommodate all passengers, then in I df Delete the trip, check the next trip, and continue searching until all flights have been searched.

[0135] Step I-14: Output I after the search is complete df gather.

[0136] The process of the connecting trip search algorithm I-2 is as follows:

[0137] Step I-21: Search for the departure airport as dep in the flight recovery plan table. i and The flights are designated as set F. fea1 The landing airport is ARR. i and The flights are designated as set F. fea2 Remove the intersection of sets Direct flights within the country;

[0138] Step I-22: From F fea1 The search returned the p-th flight f. p From F fea2 The search returned the qth flight f q Check if the two flights meet the time and space connectivity constraints, identify the found connecting flights as a feasible itinerary i, complete one search, and repeat. Let the set of feasible connecting itineraries be I. af ;

[0139] Step I-23: Check each item sequentially. af Does the connecting flight meet the capacity limit requirements? For connecting flights involving two segments, the number of passengers changing their itinerary must not exceed the smaller of the capacities of the two flight segments, i.e., ∑ i,i'∈I t ii' +Psg i' ≤min{Cap f1 Cap f2 If the capacity of that cabin class is insufficient to accommodate all passengers, then the capacity of other cabin classes will be checked; if the itinerary still cannot accommodate all passengers, then in section I... af Delete the trip and check the next connecting trip until all flights in the set have been searched.

[0140] Step I-24: Output I after the search is complete af gather.

[0141] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The above descriptions are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A two-stage delay recovery method based on interaction between airlines and passengers, characterized in that, The two stages are the static planning stage and the dynamic planning adjustment stage, wherein the static planning stage includes the following steps: Step S1: Establish a flight information database, an airport information database, a delay scenario database, and a passenger itinerary database; obtain information data on affected flights and passengers, and load them into the flight information database and the passenger itinerary database respectively; Step S2: Initialize the flight schedule for each aircraft to obtain the initial set of aircraft paths; Step S3: Establish the aircraft flight recovery model minF1 for the static planning phase as follows: Among them: Dc f This represents the operational costs of flight f, including fuel surcharges, service fees, and parking fees; Cc f Indicates the cancellation cost of flight f; y f Indicates whether flight f has been cancelled; Dp k This represents the unit time value of passengers in class k, where k=1 represents economy class passengers, k=2 represents business class passengers, and k=3 represents first class passengers. This indicates the number of passengers of class k on flight f; This indicates the delay time of flight f operated by aircraft i; This indicates whether flight f is in path r; where A, R, F, and K represent the total number of aircraft, the total number of paths, the total number of flights, and the number of passenger categories, respectively. Defined as a decision variable, indicating whether aircraft a will follow path r; Constrain the recovery model minF1: Equation (1) is the aircraft coverage constraint for flight f, which means that if flight f is canceled, all paths containing the flight cannot be executed; otherwise, if flight f is not canceled, there is one and only one path containing the flight that is executed by one and only one aircraft. Equation (2) is the unique path constraint for aircraft i, meaning that aircraft a can be assigned at most one path. Equation (3) represents the flight delay time as the difference between the actual departure time and the scheduled departure time, where... T refers to the actual takeoff time of aircraft a when performing flight f. f This refers to the original scheduled departure time of flight f; Equation (4) represents the value constraints of the decision variables; Step S4: Retrieve the flight information database, solve the recovery model minF1 of aircraft and flights based on the column generation algorithm, obtain the correspondence between aircraft and flights, and generate a recovery plan; Step S5: Identify passengers whose trips are interrupted due to flight cancellations and connection failures, and divide the passengers into z OD traffic flows based on trip attributes, trip locations, and times; match the recovery plan with the passengers on the flight, and summarize the recovery plans for passengers in each OD traffic flow; send the recovery plan to the passengers via SMS, and wait for their feedback on whether to accept, request a refund, or seek other recovery plans. The recovery plan adjustment phase includes the following steps: Step D1: The affected passenger responds by accepting the situation, requesting a refund, or seeking other recovery plans; Step D2: The airline receives feedback from affected passengers in real time and completes the dynamic adjustment phase of the plan based on the feedback; the time for releasing the recovery plan information is set as t0, and an adjustment time window is set as Δt. The feedback from affected passengers is adjusted and updated in units of adjustment time window Δt; at the end of each adjustment time window Δt, the received feedback is classified and statistically analyzed. These represent the number of passengers in category k who are affected by the disruption during itinerary i within the time window t+Δt, respectively: those who accept the ticket, those who request a refund, and those who seek other recovery plans. Step D3: For passengers affected by the disruption who accept the recovery plan, proceed to step D4; for passengers affected by the disruption who voluntarily request a refund, proceed to step D5; for passengers affected by the disruption who seek other recovery plans, proceed to step D6. Step D4: For passengers affected by the disruption who have accepted the recovery plan, arrange their travel itineraries according to the recovery plan; Step D5: For passengers who have voluntarily cancelled their tickets and have been disturbed, remove these passengers from the recovery plan during the static planning phase. Step D6: Call the passenger feedback insertion algorithm to insert data from the feasible itinerary set I. fe Search the data to find feasible trips that meet the remaining capacity limit and insert the affected passenger for that OD traffic flow; if no feasible trips can be inserted, the result is fed back to the affected passenger, and the time window is moved backward by Δt, returning to step D2, until the affected passenger accepts the recovery plan or gets a refund. Step D7: The airline selects the final recovery plan and outputs it in the form of a set of flight sequences {f1,f2,f3,...} executed by each aircraft and a set of passenger itinerary sequences {(f1,f2),f3,...}.

2. The two-stage delay recovery method based on airline-passenger interaction as described in claim 1, characterized in that: During the dynamic adjustment phase of the plan, the length of the adjustment time window Δt is set according to the following rules: T edt This indicates the end time of the delay recovery work, set by the airline based on the scale of the delay; T cur Indicates the current time; when T edt -T cur When T ≥ 3h, Δt is set to 1h; when T edt -T cur When T ∈ (1,3), Δt is set to 30min; when T edt -T cur When ≤1h, Δt is set to 10min, and at a distance T edt After 30 minutes, the collection of feedback from disturbed passengers will cease. Passengers who do not provide feedback will be considered to have complied with the final recovery plan in step D7.

3. The two-stage delay recovery method based on interaction between airlines and passengers as described in claim 2, characterized in that: The dynamic adjustment phase of the plan shall follow the following three rules: (1) Passenger Removal Mechanism At the start of the adjustment window, passengers who voluntarily canceled their tickets in each OD traffic flow within the previous time window are counted and removed from their original itineraries. The itineraries that these passengers were originally on are then selected by other passengers. (2) Trip priority resumption mechanism During the dynamic adjustment phase of the plan, passengers who submit feedback earlier have the right to priority in matching itineraries. Once a passenger's restored itinerary is locked, other passengers' itineraries will be restored within the subsequent adjustment window, provided that the itineraries of these passengers remain unchanged. (3) Random trip allocation mechanism When the capacity of feasible itineraries exceeds the total number of passengers in the OD traffic flow who need to change their tickets, only some passengers' change requests can be fulfilled. In this case, passengers who change their tickets to that itinerary will be randomly selected, and other feasible itineraries will continue to be searched for for the remaining passengers.

4. The two-stage delay recovery method based on interaction between airlines and passengers as described in claim 3, characterized in that: When performing step D2, the objective function minF2 for stroke recovery is set as follows: Among the decision variables The variable is 0-1, representing the number of passengers of type k who are disturbed and assigned from trip i to trip i'; This represents the total number of passengers of type k in the original itinerary i; delt i' Indicates the delay time of trip i'; D k U k C k These represent the unit time delay loss, violation of intent loss, and trip cancellation loss for passengers of type k, respectively. Constrain the objective function minF2 for stroke recovery: Equation (5) represents the time coherence constraint. This indicates the departure time of the subsequent flight in the connecting flight i'. t represents the arrival time of the preceding flight. MCT This represents the shortest transfer time, and the constraint states that the connection time between two flights in a connecting passenger's itinerary must not be less than the shortest transfer time; Equation (6) is the spatial connection constraint, dep i'2 arr indicates the departure airport for subsequent flights in itinerary i'. i'1 Let i' represent the arrival airport of the preceding flight in itinerary i'. This constraint indicates that the departure airport of subsequent flights in a connecting flight is the same as the arrival airport of the preceding flight; Equation (7) is the itinerary capacity constraint. This indicates the maximum passenger capacity of cabin class k for flight f. This represents the original k-type passengers in itinerary i'. The variable is 0-1, indicating whether the itinerary i' contains flight f. This constraint means that the number of passengers of class k in itinerary i' shall not exceed the minimum value of the cabin capacity of flight k in the itinerary; Equation (8) represents the value constraint of the variable.

5. A two-stage delay recovery method based on airline-passenger interaction as described in claim 4, characterized in that, The passenger feedback insertion algorithm includes the following steps: Step I-1: Retrieve information about the affected passenger's original itinerary i, including whether it is a connecting flight, and the origin and destination locations dep. i with arr i Original flight f's estimated departure time With arrival time And cabin class k; if the disturbed passenger is a direct flight passenger, follow method I-1; if the disturbed passenger is a connecting flight passenger, follow method I-2; Step I-2: If a direct flight cannot be restored in step I-11, send an inquiry to ask if the passenger accepts the connecting flight. The response time is t2. If the affected passenger accepts, proceed with method I-2. If the affected passenger does not accept, the passenger will be considered to have received a refund. Step I-3: Release the feasible itineraries obtained after the search to the affected travelers and solicit their feedback; Method I-1 includes the following steps: Step I-11: In the final recovery plan of step D7, search for the departure airport as dep. i The landing airport is ARR. i Flights; Step I-12: Filtering The feasible flights are added to the set of feasible direct flight routes I. df ,in This indicates that the departure times of flight f are arranged in chronological order. Step I-13: Check each item in sequence. df Check if the flight meets the flight capacity limit; if the capacity of this cabin class on the same flight is insufficient to accommodate all affected passengers, check the capacity of other cabin classes. If they meet the requirements, retain the itinerary in group I. df If the itinerary cannot accommodate all affected passengers, then at meeting point I... df Delete the trip, check the next trip, and continue searching until all flights have been searched. Step I-14: Output the set I after the search is complete. df ; Method I-2 includes the following steps: Step I-21: In the final recovery plan of step D7, search for the departure airport as dep. i and The flights are designated as set F. fea1 The landing airport is ARR. i and The flights are designated as set F. fea2 Remove the intersection of sets Direct flights within the country; Step I-22: From set F fea1 The search returned the p-th flight f. p From set F fea2 The search returned the qth flight f q Check whether the two flights meet the time and space connection constraints, determine the found connecting flights as a feasible itinerary i, complete one search and loop, let the set of feasible connecting itineraries be I. af ; Step I-23: Check each item sequentially. af Whether the connecting flight meets the capacity limit conditions: If the connecting flight includes two segments, the number of passengers affected by changes in their itinerary must not exceed the smaller of the capacities of the two flight segments, i.e., ∑ i,i'∈I t ii' +Psg i' ≤min{Cap f1 Cap f2 If the capacity of the flight's cabin class is insufficient to accommodate all affected passengers, then the capacity of other cabin classes will be checked; if the itinerary still cannot accommodate all affected passengers, then the passengers will be accommodated at meeting point I. af Delete the trip and check the next connecting trip until all flights in the set have been searched. Step I-24: Output I after the search is complete af gather.

Images