Method, medium and equipment for flight cancellation and recovery strategy generation

Abstract

The invention provides a method, a medium and equipment for flight cancellation and recovery strategy generation. The method comprises the following steps: acquiring weather forecast information, related information of each flight and parameter configuration information; according to the advanced cancellation cost and expected recovery cost of the flight, an objective function of a flight cancellation and recovery strategy generation model is constructed, the expected recovery cost comprises delay cost, aircraft exchange cost, real-time cancellation cost and airport imbalance cost, and the expected recovery cost comprises delay cost, aircraft exchange cost, real-time cancellation cost and airport imbalance cost; the flight cancellation and recovery strategy generation model comprises a flight coverage constraint, an aircraft distribution constraint, a time period capacity constraint and an airport balance constraint; and solving an optimal solution of the target function, and determining a flight cancelling and flight recovery strategy in advance under each scene according to the optimal solution. The operation cost of an airline company can be obviously reduced, the travel experience of passengers is improved, and the on-time rate index of the airline company is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of flight cancellation and recovery, in particular to a technology for generating flight cancellation and recovery strategy. BACKGROUND

[0002] In the aviation transportation industry, flight irregularities (including delays, cancellations, etc.) are a prominent problem affecting industry operational efficiency and passenger travel experience. Weather factors are one of the main reasons for flight irregularities. According to industry data, 60.42% of flight disruption events in China in 2023 were caused by abnormal weather. In the face of such flight irregularities, airlines usually use flight delays, cancellations, use of backup aircraft and crew, etc. recovery measures.

[0003] Currently, when airlines deal with flight recovery strategies caused by weather and other sudden factors leading to flight irregularities, due to the randomness of weather, it is difficult to accurately predict, and early cancellation is usually not considered. In practice, a "wait and see" attitude is generally adopted, and the decision to cancel the flight is made only when the flight is about to take off, resulting in delayed decision-making, increased recovery costs, and decreased passenger experience, which cannot achieve the optimization of flight recovery. Although it avoids the loss of profits if the flight still takes off, it will result in more "sunk costs" when the flight is actually canceled, including the waiting cost of the crew, the cost of food and accommodation for passengers, the compensation cost of passengers, and the loss of reputation of the airline, etc. In addition, after the flight is finally canceled, the experience of passengers will be severely damaged, and the airline cannot obtain the operational statistics advantage brought by early cancellation.

[0004] Early cancellation refers to canceling a flight a few days before the flight is scheduled to be executed, rather than at the last minute before the flight is about to take off. Compared with canceling the flight at the last minute, early cancellation has many advantages: first, it can significantly improve passenger experience, as passengers have sufficient time to rearrange their itinerary; second, flights canceled early are not counted in the airline's on-time rate statistics, thereby improving the airline's on-time rate indicators (for example, according to the standards of the Civil Aviation Administration of China, flights canceled 1 calendar day in advance are not counted in the statistics); third, early cancellation provides more opportunities for crew members to rearrange their work schedules, which helps to develop more reasonable crew scheduling plans.

[0005] In addition, with increasingly stringent regulatory constraints on aviation, penalties for flight cancellation at the last minute have become increasingly severe, further highlighting the cost advantages that early flight cancellation decisions can bring.

[0006] Therefore, how to develop a flight cancellation and recovery strategy to deal with flight irregularities caused by weather, in order to obtain comprehensive cost advantages, is a technical problem that needs to be solved. SUMMARY

[0007] To solve the above technical problems, the purpose of the present application is to provide a method, system and device for flight cancellation and recovery strategy generation.

[0008] According to one aspect of the present application, a method for flight cancellation and recovery strategy generation is provided, wherein the method comprises: obtaining weather forecast information, relevant information of each flight and parameter configuration information; constructing a target function of a flight cancellation and recovery strategy generation model according to the early cancellation cost of the flight and the expected recovery cost, wherein the expected recovery cost includes a delay cost, an aircraft exchange cost, a real-time cancellation cost and an airport imbalance cost, and the target function is:

[0009] wherein L represents a flight set, S represents a scenario set, A represents an aircraft set, represents a set of flight strings executable by aircraft a, B represents a set of airports to which aircraft completing flight string execution arrive at the end of a preset flight recovery period, represents a cost generated by the early cancellation of flight l, represents a probability of scenario s occurring, represents a cost of real-time cancellation of flight l under scenario s, represents a cost of assigning aircraft a to flight string r, represents a penalty cost of airport b imbalance, is a first flight variable, which is 1 if flight l is cancelled early, and 0 otherwise, is a second flight variable, which is 1 if flight l is cancelled in real time under scenario s, and 0 otherwise, is a flight string variable, which is 1 if aircraft a flies flight string r under scenario s, and 0 otherwise, is a slack variable, which represents the number of aircraft required to be supplemented at airport b to meet the balance requirement under scenario s; wherein the flight cancellation and recovery strategy generation model includes flight coverage constraints, aircraft allocation constraints, time period capacity constraints and airport balance constraints, wherein the flight coverage constraints are:

[0010] wherein is an indication parameter, which is 1 if flight string r contains flight l, and 0 otherwise, the aircraft allocation constraints are:

[0011] the time period capacity constraints are:

[0012] in, This represents the set of flight sequences that aircraft a can execute under scenario s. The number of takeoffs and landings used by flight sequence r during time period o. This represents the maximum number of takeoffs and landings during time period o under scenario s. This represents the set of possible takeoff and landing times under scenario s. The airport balance constraint is:

[0013] in, As an indicator parameter, it is set to 1 if the last flight in flight sequence r lands normally at airport b within the preset flight recovery period, and 0 otherwise. The number of aircraft required by airport b at the end of the preset flight recovery period. Specifically, based on the weather forecast information, the relevant information of each flight, and the parameter configuration information, the parameters corresponding to the objective function, the flight coverage constraint, the diversity allocation constraint, the time period capacity constraint, and the airport balance constraint are determined. Solve for the optimal solution of the objective function, and based on the optimal solution, determine the early cancellation and flight recovery strategies for each scenario.

[0014] Optionally, the step of solving for the optimal solution of the objective function includes: Set unexpected flight constraint variables for each scenario, and unexpected flight constraint variables for flight l under scenario s. The following constraints must be satisfied:

[0015] The flight coverage constraint has been changed to:

[0016] The first flight variable in the objective function By replacing the flight non-compliance constraint variable under each scenario, we obtain the first objective function, which is:

[0017] Find the optimal integer solution of the first objective function, and take the optimal integer solution of the first objective function as the optimal solution of the objective function.

[0018] Optionally, the step of solving the optimal integer solution of the first objective function includes: To introduce Lagrange multipliers for unexpected constraints, the first objective function is decomposed into independent second objective functions for each scenario. The second objective function for scenario s is:

[0019] The column generation method is used to solve the optimal solution of the second objective function under each scenario. Based on the optimal solution of the second objective function under each scenario, the linear relaxation solution of the first objective function is determined. Based on the linear relaxation solution of the first objective function, the dual price corresponding to the unexpected constraint is determined and assigned to the Lagrange multiplier to update the second objective function. The solution is iterated until the iteration termination condition is met, and the optimal linear solution of the unexpected constraint variable of the first objective function is obtained. If there are non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function, a divergence heuristic algorithm is used to iteratively adjust the non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function and the column generation method is used to solve the optimal solution of the unexpected constraint variables of the second objective function in each scenario until there are no non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function, and the optimal integer solution of the unexpected constraint variables of the first objective function is obtained. Based on the optimal integer solution of the unexpected constraint variables of the first objective function, the optimal integer solution of the second objective function under each scenario is obtained. Based on the optimal integer solution of the second objective function under each scenario, the optimal integer solution of the first objective function is obtained.

[0020] Optionally, the method further includes, before solving for the optimal solution of the objective function: Construct an information complete model for each scenario corresponding to the flight cancellation and recovery strategy generation model, solve the dual feasible solution of the objective function of the information complete model for each scenario, and determine the flights to be cancelled in advance based on the dual feasible solution of the objective function of the information complete model for each scenario.

[0021] Optionally, the step of solving the dual feasible solution of the objective function of the information completeness model for each scenario, and determining the early cancellation of flights based on the dual feasible solution of the objective function of the information completeness model for each scenario, includes: Based on the information completeness model of each scenario, its dual form information completeness model is obtained, and the objective function of the dual form information completeness model of each scenario is solved to obtain the corresponding dual feasible solution. Based on each dual feasible solution, flights that are canceled in advance or in real time in at least one dual feasible solution are identified as flights to be identified; The first dual cost sum is calculated for each of the following scenarios: the first dual cost sum is calculated when the flight to be identified is cancelled in advance in each scenario, and the second dual cost sum is calculated when cancellation is not allowed. The results are then compared. If the first dual cost sum is less than the second dual cost sum, the flight to be identified is determined to be a flight cancelled in advance.

[0022] Optionally, where the expected recovery cost also includes unit connection cost, the objective function is:

[0023] Where D represents the set of flight connection pairs, To operate connecting flights and The costs incurred due to unit connection interruptions As a variable for crew connection, if connecting flights are operated and If the connection of the unit is interrupted under scenario s, the value is 1; otherwise, it is 0. The flight cancellation and resumption strategy generation model also includes crew connection constraints, wherein the crew connection constraints are:

[0024] Where M is a preset positive integer, As an indicator parameter, if flight string r contains connecting flights and If it is true, then it is 1; otherwise, it is 0. This represents the flights in the flight string r. Actual takeoff time; This represents the flights in the flight string r. Actual landing time; Indicates connecting flights and The minimum ground rest time required by the crew.

[0025] Optionally, the expected recovery cost also includes passenger recovery costs. The objective function is then:

[0026] Where I represents the original set of passenger itineraries, and J represents the candidate set of passenger itineraries to be reset. This represents the set of flights included in the original passenger itinerary i. Let i represent the set of candidate reset passenger itineraries provided by the original passenger itinerary. This represents the unit cost of refunds or rebooking to other airlines for passengers whose original travel itinerary was interrupted. This represents the number of passengers in scenario s whose original travel itinerary i requires a refund or rebooking to another airline. Let $\frac{ ... This represents the number of passengers who are reassigned from their original travel itinerary i to a candidate reset travel itinerary j under scenario s. The flight cancellation and resumption strategy generation model also includes passenger connection constraints, passenger flow conservation constraints, feasibility constraints for resetting passenger itineraries, and flight capacity constraints. The passenger connection constraints are as follows:

[0027]

[0028] Where E represents the set of all flight connection pairs in the original passenger itinerary. As an indicator parameter, if resetting passenger itinerary j includes connecting flights and If the result is 1, then it equals 1; otherwise, it equals 0. For passenger connection variables, if passengers cannot connect to flights under scenario s. and If it is true, then it is 1; otherwise, it is 0. Indicates connecting flights and The shortest connection time required for passengers between the two routes. The passenger flow conservation constraint is:

[0029] in, Let i represent the set of candidate reset passenger itineraries provided by the original passenger itinerary. Let i be the number of passengers in the original passenger itinerary i. The feasibility constraint for resetting passenger itineraries is:

[0030] The flight capacity constraint is as follows:

[0031] in, As an indicator parameter, it equals 1 if resetting passenger itinerary j includes flight copy t, and 0 otherwise. Indicates airplane Seating capacity, This represents the delayed repetition set of flight l obtained from flight string r.

[0032] According to another aspect of this application, a computer-readable medium is provided, wherein computer-readable instructions are stored on the medium, which are executed by a processor to implement part or all of any of the above methods.

[0033] According to another aspect of this application, an apparatus for generating flight cancellation and recovery strategies is provided, wherein the apparatus includes: One or more processors; and a memory storing computer-readable instructions that, when executed, cause the processor to perform some or all of the operations described above.

[0034] Compared with existing technologies, this application provides a method, medium, and device for generating flight cancellation and recovery strategies. The method includes: acquiring weather forecast information, relevant information for each flight, and parameter configuration information; constructing an objective function for a flight cancellation and recovery strategy generation model based on the early cancellation cost and expected recovery cost of the flight, wherein the expected recovery cost includes delay cost, aircraft swap cost, real-time cancellation cost, and airport imbalance cost; and wherein the flight cancellation and recovery strategy generation model includes flight coverage constraints, aircraft allocation constraints, time slot capacity constraints, and airport balance constraints; solving for the optimal solution of the objective function, and determining the early cancellation flight and flight recovery strategy for each scenario based on the optimal solution. This application can assess flight recovery strategies caused by future abnormal weather in advance, considering not only expected flight recovery costs but also early cancellation costs, and evaluating flight recovery strategies under possible scenarios based on comprehensive costs for decision-making. It can significantly reduce airline operating costs, improve passenger travel experience, and enhance airline on-time performance, thereby bringing significant commercial and social value. Attached Figure Description

[0035] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A schematic diagram is shown of a method for generating flight cancellation and recovery strategies according to one aspect of this application; Figure 2 This diagram illustrates a system for generating flight cancellation and recovery strategies according to one aspect of this application. Figure 3 A schematic diagram illustrating the solution process of the first objective function of a NAM model according to an optional embodiment of this application is shown.

[0036] The same or similar reference numerals in the accompanying drawings represent the same or similar parts. Detailed Implementation

[0037] The present application will now be described in further detail with reference to the accompanying drawings.

[0038] In a typical configuration of various embodiments of this application, the method execution entity, each trusted party of the system, and / or each module of the device may include one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0039] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0040] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0041] This application provides a technical solution for generating flight cancellation and recovery strategies. It assesses flight recovery strategies in advance for scenarios caused by future abnormal weather, considering not only the anticipated recovery costs but also the costs of early flight cancellations. Based on a comprehensive cost analysis, it evaluates flight recovery strategies under various possible scenarios. Practical application and comparative verification have shown that it can save an average of 17% in operating expenses. It also allows for advance notification of passengers affected by early flight cancellations, avoiding passive waiting upon arrival at the airport and significantly improving the service experience. Furthermore, it helps airlines maintain key operational indicators such as on-time performance and provides more time for crew resource reallocation. Therefore, this application offers significant commercial and social value.

[0042] To further illustrate the technical means adopted and the effects achieved in this application, the technical solution of this application will be clearly and completely described below in conjunction with the accompanying drawings and preferred embodiments.

[0043] Figure 1 The diagram illustrates a method for generating flight cancellation and recovery strategies according to one aspect of this application, wherein one embodiment of the method includes: S101 obtains weather forecast information, relevant information for each flight, and parameter configuration information; S102 constructs an objective function for a flight cancellation and recovery strategy generation model based on the early cancellation cost and expected recovery cost of flights. The expected recovery cost includes delay cost, aircraft swap cost, real-time cancellation cost, and airport imbalance cost. The objective function is:

[0044] Where L represents the set of flights, S represents the set of scenarios, and A represents the set of aircraft. Let B represent the set of flight sequences that aircraft a can execute, and let B represent the set of airports that the aircraft that complete the flight sequence at the end of the preset flight recovery period will arrive at. This refers to the cost incurred due to the early cancellation of flight l. This represents the probability of scenario s occurring. This represents the cost of canceling flight l in real time under scenario s. This represents the cost of assigning aircraft a to flight string r. This indicates that airport B has an unbalanced penalty cost. For the first flight variable, the value is 1 if flight l is cancelled in advance, and 0 otherwise. For the second flight variable, it is 1 if flight l is canceled in real time under scenario s, and 0 otherwise. Let be the flight string variable. If aircraft 'a' is flying flight string 'r' under scenario 's', then the value is 1; otherwise, it is 0. Let be a slack variable, representing the number of aircraft that airport b needs to add to meet the balance requirements under scenario s; The flight cancellation and recovery strategy generation model includes flight coverage constraints, aircraft allocation constraints, time slot capacity constraints, and airport balance constraints. The flight coverage constraint is as follows:

[0045] in, This is an indicator parameter; it is 1 if flight string r contains flight l, and 0 otherwise. The aircraft allocation constraints are as follows:

[0046] The time period capacity constraint is:

[0047] in, This represents the set of flight sequences that aircraft a can execute under scenario s. The number of takeoffs and landings used by flight sequence r during time period o. This represents the maximum number of takeoffs and landings during time period o under scenario s. This represents the set of possible takeoff and landing times under scenario s. The airport balance constraint is:

[0048] in, As an indicator parameter, it is set to 1 if the last flight in flight sequence r lands normally at airport b within the preset flight recovery period, and 0 otherwise. The number of aircraft required by airport b at the end of the preset flight recovery period. Specifically, based on the weather forecast information, the relevant information of each flight, and the parameter configuration information, the parameters corresponding to the objective function, the flight coverage constraint, the diversity allocation constraint, the time period capacity constraint, and the airport balance constraint are determined. S104 solves for the optimal solution of the objective function, and determines the early cancellation and flight recovery strategies for each scenario based on the optimal solution.

[0049] In the application scenario of this application, when severe weather may affect a large number of flights, airlines can consider the costs caused by flight disruptions in two stages according to the technical solution of this application: First, assess and decide in advance which flights should be canceled before the scheduled flight execution (usually at least one day in advance), rather than canceling them just before the scheduled flight (about to take off); second, provide the optimal flight recovery plan under various scenarios to minimize the overall cost. The method for generating flight cancellation and recovery strategies provided in this application is implemented through system 100. System 100 includes computer equipment and / or a cloud with the necessary hardware and software environment deployed. The computer equipment includes, but is not limited to, personal computers, laptops, industrial computers, servers, network hosts, single network servers, or network server clusters. The cloud consists of a large number of computers or network servers based on cloud computing, where cloud computing is a type of distributed computing, consisting of a virtual supercomputer composed of a group of loosely coupled computers. The computer equipment and / or cloud described herein are merely examples. Other existing or future equipment and / or resource sharing platforms that are applicable to this application should also be included within the scope of protection of this application, and are hereby incorporated by reference.

[0050] like Figure 2As shown, system 100 includes input module 110, core algorithm module 120 and output module 130.

[0051] In this embodiment of the method, in step S101, the system 100 can obtain weather forecast information, relevant information of each flight, and parameter configuration information through the input module 110.

[0052] The system can obtain weather forecasts and historical weather information for the next few days (e.g., 1-3 days) from external or public data sources (such as commercial meteorological service agencies, the National Meteorological Administration, and air traffic control departments) via input module 110. It can also obtain information related to each flight from airlines, airports, and air traffic control departments, such as historical and future aircraft information (e.g., aircraft number, aircraft type, seat capacity, base airport), flight information (e.g., planned departure and arrival airports, planned departure and arrival times, operating aircraft, flight type (domestic / international), number of passengers, ticket price, etc.), passenger itinerary information, airport information (e.g., airport code, airport name, airport type (domestic / international), unit capacity (e.g., number of takeoffs and landings per hour, minimum turnaround time requirement, etc.), and time-segment capacity information (sc can be used to determine scenario time-segment information). Airport closure scenarios (ENAs) are generated based on uncertain weather forecasts and historical data (weather, flight, etc.). Each scenario corresponds to a set of airport capacity information and its probability of occurrence. For example, based on weather forecasts and historical flight data, 10 scenarios could be generated where airports might close, leading to flight disruptions, along with their probabilities. Two of these scenarios are: Scenario 1, "Airport A will be closed for 19 hours, from 8:00 PM on September 15th to 3:00 PM the next day," with a probability of 19%; and Scenario 2, "Airport A will be closed for 18 hours, from 8:00 PM on September 15th to 2:00 PM the next day," with a probability of 20%. The scenario time period information can be understood as the limit on the number of takeoffs and landings available at a particular airport within a specific time period (e.g., per hour) under a given scenario. Continuing with the above example, the scenario time period information for Scenario 1 could be: "Airport A, from 20:00 on September 15th to 15:00 the next day, the airport is closed, and the takeoff and landing capacity is 0; from 15:00 to 16:00 on September 16th, the takeoff and landing capacity recovers to 30% of normal." Furthermore, relevant operational constraints and industry standards known in the civil aviation industry (such as the minimum turntime for aircraft, the minimum ground rest time for crew members, the minimum connection time required for passenger transfers, and airport takeoff and landing capacity limits during specific time periods) are obtained as boundary conditions that the core algorithm of this application needs to satisfy.

[0053] Through input module 110, relevant parameter configuration information input by airline operators can also be obtained, including but not limited to: cost parameters such as flight cancellation cost, aircraft type change cost, delay cost, flight delay compensation cost, passenger cancellation cost, passenger rebooking cost, etc., as well as constraint parameters such as the number of aircraft required at the airport at the end of the preset flight recovery period, crew connection time, passenger transfer time, and airport capacity restrictions related to the airline.

[0054] In this application, considering the randomness and uncertainty of weather forecasts, multiple scenarios that may lead to flight disruptions in the next few days can be determined based on weather forecast information and historical flight-related information (that is, the uncertain weather forecast is transformed into multiple quantifiable scenarios that may lead to flight disruptions), and the optimal flight cancellation and recovery strategies under each scenario are given so that the decision-making scheme with the lowest overall cost can be selected.

[0055] Continuing in this embodiment, in step S102, the core algorithm module 120 of system 100 performs data mining and processing on the acquired weather forecast information, relevant information of each flight, and parameter configuration information. This allows for the determination of the early cancellation cost and expected recovery cost for each flight. The expected recovery cost includes delay costs, aircraft swap costs, real-time cancellation costs, and airport imbalance costs. Furthermore, it can determine the flight set, scenario set, aircraft set, set of flight sequences that can be executed by aircraft, the set of airports where the aircraft that complete the flight sequence will arrive at the end of the preset flight recovery period (e.g., one calendar day after the end of severe weather; by the end of the flight recovery period, airline resources can usually be ensured to return to their original normal balance), the cost incurred by early flight cancellation, the probability of each scenario occurring, the cost of real-time flight cancellation under a certain scenario, the cost of allocating aircraft to flight sequences, the penalty cost for airport imbalance, the set of flight sequences that a certain aircraft can execute under a certain scenario, the number of takeoffs and landings used by a flight sequence in a certain time period, the maximum number of takeoffs and landings in a certain time period under a certain scenario, the set of available takeoff and landing time periods under a certain scenario, the number of aircraft required by the airport at the end of the preset flight recovery period, and so on.

[0056] In this application, the total cost of flight cancellation and recovery may include two main parts: the cost of early flight cancellation in the first phase and the expected recovery cost under each possible scenario in the second phase, wherein the expected recovery cost may include: delay cost, aircraft swap cost, real-time cancellation cost and airport imbalance cost.

[0057] Through the core algorithm module 120, the early cancellation cost and expected recovery cost of each flight can be comprehensively considered. A constrained mathematical optimization model (or linear programming model) is used to construct a flight cancellation and recovery strategy generation model. The objective function of this model is:

[0058] Where L represents the set of all affected flights, S represents the set of all possible scenarios that could lead to flight disruptions due to severe weather, and A represents the set of all affected flights operated by the same aircraft. Let B represent the set of all flight sequences that aircraft a can execute, and let B represent the set of airports that all aircraft that have completed their flight sequences will arrive at at the end of the preset flight recovery period. This refers to the cost incurred due to the early cancellation of flight l. This represents the probability of scenario s occurring. This represents the cost of canceling flight l in real time under scenario s. This represents the cost of assigning aircraft a to flight string r. This indicates that airport B has an unbalanced penalty cost.

[0059] in, Let be the first flight variable. It is a binary variable and its value is as follows: if flight 1 is canceled in advance, it is 1; otherwise, it is 0. Let be the second flight variable. It is a binary variable and its value is as follows: if flight l is canceled in real time under scenario s, it is 1; otherwise, it is 0. Let be the flight string variable, which is a binary variable. Its value is as follows: if aircraft a flies flight string r under scenario s, then it is 1; otherwise, it is 0. Let be a slack variable, representing the number of aircraft that airport b needs to add to meet the balance requirements under scenario s.

[0060] The constraints of the flight cancellation and resumption strategy generation model include: flight coverage constraints, aircraft allocation constraints, time slot capacity constraints, and airport balance constraints. The flight coverage constraint is:

[0061] in, This is an indicator parameter; it is 1 if flight string r contains flight l, and 0 otherwise.

[0062] The aircraft allocation constraints are as follows:

[0063] The capacity constraint for this period is:

[0064] in, Let S represent the set of flight strings that aircraft a can execute under scenario s. This refers to the number of takeoffs and landings used by flight series r during time period o. This represents the maximum number of takeoffs and landings during time period o under scenario s. This represents the set of all possible takeoff and landing times under scenario s.

[0065] The airport's equilibrium constraints are:

[0066] in, As an indicator parameter, it is 1 if the last flight in flight string r lands normally at airport b within the preset flight recovery period, and 0 otherwise. The airport balance constraint specifies the number of aircraft required for airport b to achieve balance at the end of the preset flight recovery period. This constraint stipulates that the actual number of aircraft arriving at airport b plus the number of aircraft to be added must not be less than the number of aircraft required for airport b to achieve balance.

[0067] The system can perform data mining and processing on the weather forecast information, relevant information for each flight, and parameter configuration information obtained through the input module 110 to determine the aforementioned parameters.

[0068] Among them, the variables that minimize the objective function value under the above constraints can be calculated. , , , The optimal solution.

[0069] Continuing in this embodiment, in step S104, the optimal solution of the objective function is solved by the output module 130 of the system 100, and the early cancellation and flight recovery strategies under each scenario are determined based on the optimal solution obtained.

[0070] The constructed flight cancellation and recovery strategy generation model is large in scale, resulting in a high computational load and low efficiency due to its randomness. Moreover, the costs associated with early cancellation in this model have no constraints themselves, yet they are embedded in the expected recovery costs. If the costs associated with early cancellation can be decomposed, the flight cancellation and recovery strategy generation model can be decomposed into multiple independent sub-models, each corresponding to a scenario and containing only a small number of variables and constraints, thereby greatly simplifying the solution process.

[0071] Optionally, in step S104, solving for the optimal solution of the objective function includes: Set unexpected flight constraint variables for each scenario, and unexpected flight constraint variables for flight l under scenario s. The following constraints must be satisfied:

[0072] The flight coverage constraint has been changed to:

[0073] The first flight variable in the objective function By replacing the flight non-compliance constraint variable under each scenario, we obtain the first objective function, which is:

[0074] Find the optimal integer solution of the first objective function, and take the optimal integer solution of the first objective function as the optimal solution of the objective function.

[0075] The flight cancellation and recovery strategy generation model constructed in this application is a large-scale stochastic programming model. Due to the inclusion of multiple scenarios, its objective function has low solution efficiency.

[0076] In this optional embodiment, the flight cancellation and recovery strategy generation model can be considered as the base model (BM). The BM model has a block diagonal matrix structure. Based on this structural characteristic, it can be first converted into a nonanticipativity model (NAM). Then, the NAM model can be decomposed into multiple sub-models with the same number of scenarios as mentioned above.

[0077] In the objective function of the BM model, the first variable related to early cancellation is... Regardless of the scenario, firstly, based on the decomposition requirements, a variable related to the first variable can be introduced for any scenario s in the scenario set S. Corresponding unexpected constraint variables To ensure that the variables related to early cancellation costs are consistent across all scenarios (i.e., each unexpected constraint variable), (The values ​​are all equal), in addition to satisfying the aforementioned constraints, the following unexpected constraints should also be satisfied:

[0078] That is, the unexpected constraint variables related to the cost of early cancellation of flight l under any scenario s. It must be equal to the expected value of the unexpected constraint variable for flight l being canceled early in all scenarios, thus ensuring that in the feasible solution, all unexpected constraint variables are equal. For the same flight, l is always equal.

[0079] Secondly, the flight constraints in the BM model that link early cancellation costs and expected recovery costs should be modified accordingly, with the first variable being changed. Replace with the corresponding unexpected constraint variables The revised flight constraints are as follows:

[0080] Then, the objective function of the above BM model is modified accordingly, and the first flight variable is changed. Use the flight unexpected constraint variable under each scenario The BM model is replaced and transformed into a NAM model, resulting in the first objective function of the NAM model, where the first objective function is:

[0081] As can be seen from the first objective function, different scenarios in the NAM model are only related through unexpected constraints. Based on this structure, the Lagrange dual decomposition method can be used to decompose the NAM model into multiple independent sub-models (i.e., Lagrange sub-models), with the same number of scenarios as mentioned above. Therefore, solving for the optimal solution of the objective function of the BM model can be transformed into solving for the optimal integer solution of the first objective function of the NAM model (i.e., solving multiple sub-models), thereby reducing the computational load and improving computational efficiency. The optimal integer solution of the first objective function can be used as the optimal solution of the objective function of the BM model.

[0082] Optionally, the step of solving the optimal integer solution of the first objective function includes: To introduce Lagrange multipliers for unexpected constraints, the first objective function is decomposed into independent second objective functions for each scenario. The second objective function for scenario s is:

[0083] The column generation method is used to solve the optimal solution of the second objective function under each scenario. Based on the optimal solution of the second objective function under each scenario, the linear relaxation solution of the first objective function is determined. Based on the linear relaxation solution of the first objective function, the dual price corresponding to the unexpected constraint is determined and assigned to the Lagrange multiplier to update the second objective function. The solution is iterated until the iteration termination condition is met, and the optimal linear solution of the unexpected constraint variable of the first objective function is obtained. If there are non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function, a divergence heuristic algorithm is used to iteratively adjust the non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function and the column generation method is used to solve the optimal solution of the unexpected constraint variables of the second objective function in each scenario until there are no non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function, and the optimal integer solution of the unexpected constraint variables of the first objective function is obtained. Based on the optimal integer solution of the unexpected constraint variables of the first objective function, the optimal integer solution of the second objective function under each scenario is obtained. Based on the optimal integer solution of the second objective function under each scenario, the optimal integer solution of the first objective function is obtained.

[0084] The relationships between the scenarios in the above NAM model are only unexpected constraints, and are the only "coupling constraints" that prevent the NAM model from being decomposed into multiple sub-models according to scenarios.

[0085] In this optional embodiment, the Lagrange relaxation method can be used to "relax" the constraints. First, a Lagrange multiplier can be introduced for any unexpected constraint in scenario s. Construct the corresponding penalty terms, where, For unsigned real numbers, the NAM model is transformed into a Lagrangian Dual Model (LDM), relaxing the original coupling constraints. After relaxation, solving the objective function of the NAM model becomes solving its Lagrangian dual problem, i.e., maximizing the dual function of the Lagrange multipliers. The objective function of the LDM model can be:

[0086] Since the unexpected constraints of coupling have been moved into the objective function of the LDM model, the LDM model can be perfectly decomposed into several independent sub-models, the same number as the number of scenarios, with each sub-model corresponding to one scenario. These independent sub-models can be called Lagrangian Submodels (LSMs), and the second objective function of each LSM model contains only variables and constraints related to a single scenario. Through this transformation, the original solution to the objective function of a large-scale BM model can be transformed into the parallel solution to the second objective function of multiple much smaller LSM models, which can greatly improve computational efficiency.

[0087] The second objective function of the LSM model under any scenario s can be:

[0088] According to Lagrange duality theory, operations research, and optimization theory, the optimal value of the dual problem is equal to the optimal value of the linear relaxation of the primal problem (i.e., strong duality). Through iterative calculation, the optimal dual price corresponding to the unexpected constraint can be found and assigned to the Lagrange multiplier to calculate the optimal objective function value of the LDM model. The optimal dual can then be used as the optimal solution to the LDM model.

[0089] Specifically, the Lagrange multipliers in the second objective function of the LSM model corresponding to each scenario are first initialized (initial values ​​can be set to 0). Column generation (CG) can be used to solve for the optimal solution of the second objective function of the LSM model corresponding to each scenario. Column generation is an algorithm that finds the optimal path solution without enumerating all possible paths. In this optional embodiment, referring to column generation, the LSM model corresponding to each scenario can be used as the master problem, and the flight string variables are restricted to a preset subset, thereby constructing the corresponding restricted master problem (RMP), which includes two main parts: the RMP model and the flight string generation model. The RMP model is used to find the optimal solution and provide dual information under the current flight string subset (initially the preset flight string subset), while the flight string generation model uses the dual information to automatically generate the optimal potential flight string through a multi-label shortest path algorithm. The solution process may include the following steps: Step 1: Based on the relevant information of each flight, the original flight set and the set of aircraft to be executed can be obtained, and then the initial flight string can be obtained, which is added to the restricted master problem RMP.

[0090] Step 2: Iteratively solve the RMP. In each iteration, obtain the optimal solution of the current RMP (as the solution of the LSM in the current round) and the dual solution. Then pass the dual solution to the aircraft flight string generation module.

[0091] Step 3: Based on the dual solution, identify whether there are flight strings with negative cost reduction, indicating that adding these flight strings to the LSM could potentially bring additional profits.

[0092] Step 4: If it exists, immediately add it to RMP, then return to Step 2 to solve RMP again and start a new round of iterations; if it does not exist, the optimal linear programming (LP) solution has been obtained, which is the optimal solution of the second objective function of the LSM model.

[0093] Based on the optimal solution of the second objective function of the LSM model for each scenario, the optimal solutions of the flight string variables (i.e., a potential set of recoverable flight strings) are summarized and applied to the first objective function of the NAM model. This allows us to obtain the linear relaxation solutions for each scenario of the first objective function of the NAM model, as well as the solutions for the unexpected constraint variables in the linear relaxation solutions for each scenario of the first objective function. Based on these solutions, the dual price corresponding to the unexpected constraint of any scenario s can be determined. And assign it to the Lagrange multiplier of scenario s. The second objective function of the LSM model corresponding to any scenario s is updated iteratively until the iteration termination condition is met, yielding the optimal linear solution of the unexpected constraint variables of the first objective function of the NAM model. The iteration termination condition can be determined by whether the value of the first objective function of the NAM model converges (i.e., no longer changes significantly) in each iteration. If the value of the first objective function has not converged, it means that new and more valuable flight sequences can be generated. The dual price can be determined based on the solution of the unexpected constraint variables of the first objective function in this iteration, and assigned to the Lagrange multiplier to update the Lagrange multiplier, then a new round of iteration begins. If the value of the first objective function converges, it means that no flight sequence that can further improve costs can be found, and the solution of the unexpected constraint variables of the first objective function obtained at this time is the optimal linear solution.

[0094] The optimal linear solution of the unexpected constraint variables of the first objective function obtained above may be a non-integer value (i.e., a decimal or fraction), which needs to be further processed to obtain the optimal integer solution of the unexpected constraint variables of the first objective function.

[0095] Continuing in this optional embodiment, if there are non-integers in the optimal linear solution of the unexpected constraint variables of the first objective function, a diving heuristic algorithm can be used. First, non-integers are identified in the optimal linear solution of the unexpected constraint variables for each flight in each scenario of the first objective function. For each non-integer, if its value is close to 1 (e.g., greater than a preset threshold, such as 0.9), it is directly fixed to 1; if its value is close to 0 (e.g., less than a preset threshold, such as 0.1), it is directly fixed to 0. Then, among the remaining non-integers, the one with the highest value is selected and fixed to 1 or 0 according to preset conditions. For example, suppose the preset condition is: for flight l, compare the corresponding early cancellation cost with the expected real-time cancellation cost. If the early cancellation cost is lower than the expected real-time cancellation cost, it is fixed to 1; otherwise, it is fixed to 0. After fixing a portion of the unexpected constraint variables of the first objective function, the column generation method is used again to solve for the optimal solution of the second objective function under each scenario, thereby obtaining the optimal linear solution of the unexpected constraint variables for each scenario of the first objective function of the NAM model. The iteration is fixed and the solution is obtained until the optimal linear solution of the unexpected constraint variables corresponding to each scenario of the first objective function of the NAM model is an integer (1 or 0). That is, the optimal integer solution of the unexpected constraint variables of the first objective function of the NAM model is obtained, which is the unexpected constraint variable corresponding to any flight l under any scenario s. A set of.

[0096] Finally, based on the optimal integer solution of the unexpected constraint variables of the first objective function of the NAM model, the optimal integer solution of the second objective function of the LSM model under each scenario is obtained. Based on the optimal integer solution of the second objective function of the LSM model under each scenario, the optimal integer solutions of each variable of the first objective function of the NAM model are obtained, including the unexpected constraint variables corresponding to any flight l under any given scenario s. The set of variables, and the second flight variable corresponding to any flight l under any scenario s. , the flight string variable corresponding to any aircraft a under any scenario s And the slack variables corresponding to any airport b under any scenario s. The set of values. Based on the optimal integer solution of the first objective function, the optimal solution of the objective function of the BM model can be determined, and then output through the output module 130 of the system 100.

[0097] For example, suppose that after data mining and processing of weather forecast information, relevant information of each flight, and parameter configuration information by the core algorithm module 120 of system 100, the resulting scenario set S includes N possible scenarios. The solution process of the first objective function of the NAM model obtained by the BM model transformation is as follows:Figure 3 As shown.

[0098] Before solving the objective function of the BM model, flights that obviously do not need to be cancelled in advance can be identified and excluded in advance. This reduces the search space of feasible solutions during the solution process, thereby reducing the amount of computation and improving computational efficiency.

[0099] Optionally, before step S104, the method further includes: S103 Constructs an information complete model for each scenario corresponding to the flight cancellation and recovery strategy generation model, solves the dual feasible solution of the objective function of the information complete model for each scenario, and determines the flights to be cancelled in advance based on the dual feasible solution of the objective function of the information complete model for each scenario.

[0100] In this optional embodiment, the core algorithm module 120 of system 100 can also construct a Model with Perfect Information (MPI) model for each scenario corresponding to the flight cancellation and recovery strategy generation model. The MPI model is an idealized model that assumes future scenarios are known. It assumes that the decision-maker knows exactly which specific scenario will occur before making the decision to cancel a flight early. In this case, the decision is made independently for a specific scenario. Generally, the cost of early cancellation is lower than the cost of real-time cancellation. From a cost perspective, any flight that needs to be cancelled will choose the lower-cost early cancellation method. Therefore, for the MPI model corresponding to a single scenario, its objective function will not include the real-time cancellation cost term, but only the early cancellation cost, delay cost, aircraft swap cost, and airport imbalance cost. The objective function of the MPI model corresponding to any scenario s is shown below:

[0101] The constraints of the objective function of the MPI model are the same as those of the first objective function of the NAM model.

[0102] The dual feasible solution of the objective function of the MPI model for each scenario can be solved through the core algorithm module 120. Based on the dual feasible solution of the objective function of the MPI model for each scenario, the flights to be cancelled in advance can be determined, thereby reducing the search space when solving the flight cancellation and recovery strategy generation model and improving computational efficiency.

[0103] Optionally, the step of solving the dual feasible solution of the objective function of the information completeness model for each scenario, and determining the early cancellation of flights based on the dual feasible solution of the objective function of the information completeness model for each scenario, includes: Based on the information completeness model of each scenario, its dual form information completeness model is obtained, and the objective function of the dual form information completeness model of each scenario is solved to obtain the corresponding dual feasible solution. Based on each dual feasible solution, flights that are canceled in advance or in real time in at least one dual feasible solution are identified as flights to be identified; The first dual cost sum is calculated for each of the following scenarios: the first dual cost sum is calculated when the flight to be identified is cancelled in advance in each scenario, and the second dual cost sum is calculated when cancellation is not allowed. The results are then compared. If the first dual cost sum is less than the second dual cost sum, the flight to be identified is determined to be a flight cancelled in advance.

[0104] In this optional embodiment, the core algorithm module 120 can first obtain the dual form of the MPI model corresponding to each scenario, the DMPI (Dual Model with Perfect Information) model. The objective function of the DMPI model can be expressed as follows:

[0105] Among them, dual variables Corresponding to the flight coverage constraint in the MPI model, it can reflect the shadow price that satisfies the coverage requirement (i.e., is operated normally, canceled in advance, or canceled in real time) for a certain flight l under the corresponding scenario s; dual variable Corresponding to the aircraft allocation constraint in the MPI model, it can be reflected in the shadow price of using aircraft a (i.e., allocating flight strings to aircraft a) under the corresponding scenario s; dual variable Corresponding to the time-slot capacity constraint in the MPI model, it can reflect the shadow price of the capacity of time slot o under the corresponding scenario s; dual variable Corresponding to the airport balance constraint in the MPI model, this reflects the shadow price of the number of aircraft required for airport b under the corresponding scenario s. Specifically, the sum of the shadow prices of all flights in a flight string should not exceed the adjusted cost of that flight string, while the shadow price of a single flight should not exceed its early cancellation cost. The constraints of the DMPI model include:

[0106]

[0107]

[0108]

[0109]

[0110] Next, the optimal solution of the objective function of the DMPI model corresponding to each scenario is obtained. This solution serves as the dual feasible solution of the objective function of the information completeness model for each scenario, and is also a feasible linear solution of the dual base model (DBM) of the BM model. The optimal function value of the objective function of the MPI model corresponding to each scenario is a lower bound of the optimal function value of the NAM model (DMPI and DBM have the same objective function, and the feasible region of DMPI is no greater than that of DBM. The constraints of DMPI are no weaker than those of DBM. Therefore, any solution that satisfies the constraints of DMPI must also satisfy the constraints of DBM. Thus, the feasible linear solution of DMPI must also be a feasible linear solution of DBM. The optimal value of DMPI must not be greater than the optimal value of DBM. Correspondingly, the optimal value of MPI is not greater than the optimal value of NAM. That is, the optimal value of MPI is a lower bound of the optimal value of NAM). Therefore, based on the dual feasible solution of the objective function of the information completeness model for each scenario, it is possible to determine which flights can be cancelled in advance.

[0111] Specifically, the optimal solution of the objective function of the DMPI model corresponding to the MPI model for each scenario can be solved in parallel. This optimal solution serves as the dual feasible solution of the objective function of the MPI model for each scenario. Iterating through all scenarios, based on the dual feasible solution of the objective function of the MPI model for each scenario, for any flight l, it is determined whether the flight operates normally in all scenarios. If so, it can be considered that the flight is unlikely to be cancelled in advance or in real-time. For flights that are cancelled in advance in at least one scenario, two cases can be considered: Scenario 1: Assuming the flight is cancelled in advance in all scenarios, then according to the relevant constraints, the sum of the dual costs of the flight in all scenarios should equal its early cancellation cost.

[0112] Scenario 2: Assuming the flight cannot be cancelled in advance, if it is cancelled under a certain scenario, it should be considered an immediate cancellation. Therefore, according to the relevant constraints, the dual cost of the flight under that scenario equals the immediate cancellation cost. If the flight is operating normally under a certain scenario, its dual cost under that scenario can be considered equal to the corresponding shadow price. By iterating through all scenarios and adding the dual costs of the flight under all scenarios, we obtain the total dual cost.

[0113] Next, compare the sum of the dual costs in the two scenarios. If the sum of the dual costs in scenario one is greater than the sum of the dual costs in scenario two, it means that the cost of forced early cancellation is higher. This indicates that canceling the flight early is not the lowest-cost option, and therefore, the flight can be excluded from the list of candidates for early cancellation (i.e., the first flight variable corresponding to this flight is forcibly set to 0). If the sum of the dual costs in scenario one is less than the sum of the dual costs in scenario two, then the decision option of early cancellation for this flight is retained.

[0114] In the above embodiments and / or optional embodiments, when constructing the flight cancellation and recovery strategy generation model, the expected recovery cost only considers the flight delay cost, aircraft exchange cost, real-time cancellation cost and airport imbalance cost, without considering the cost caused by crew interruption.

[0115] In actual operations, the turnaround time requirements for aircraft and crews at airports differ. After completing a flight mission, a crew needs a minimum ground rest time before flying the next connecting flight mission, and this minimum ground rest time is usually longer than the aircraft's minimum turnaround time. Therefore, a flight resumption strategy that is feasible for the aircraft may lead to crew duty interruptions due to insufficient crew turnaround time, resulting in additional costs.

[0116] Therefore, when building a model for generating flight cancellation and recovery strategies, the costs associated with potential crew disruptions can also be taken into account.

[0117] Optionally, where the expected recovery cost also includes unit connection cost, the objective function is:

[0118] Where D represents the set of flight connection pairs, To operate connecting flights and The costs incurred due to unit connection interruptions As a variable for crew connection, if connecting flights are operated and If the connection of the unit is interrupted under scenario s, the value is 1; otherwise, it is 0. The flight cancellation and resumption strategy generation model also includes crew connection constraints, wherein the crew connection constraints are:

[0119] Where M is a preset positive integer, As an indicator parameter, if flight string r contains connecting flights and If it is true, then it is 1; otherwise, it is 0. This represents the flights in the flight string r. Actual takeoff time; This represents the flights in the flight string r. Actual landing time; Indicates connecting flights and The minimum ground rest time required by the crew.

[0120] In this optional embodiment, in step S102, the core algorithm module 120 of system 100 performs data mining and processing on the acquired weather forecast information, relevant information of each flight and parameter configuration information. It can also determine the crew connection cost, the flight connection pair set, the cost incurred if the crew of the connecting flight is interrupted, the actual take-off and landing time of the flight series, the minimum ground rest time required by the crew of the connecting flight, and so on.

[0121] Through the core algorithm module 120, the early cancellation cost, crew connection cost, and expected recovery cost of each flight can be comprehensively considered. A constrained mathematical optimization model is used to construct a flight cancellation and recovery strategy generation model, where the objective function of this model is:

[0122] Where D represents the set of all connecting flight pairs that have a connection relationship. For connecting flights ( and connecting flights and The costs incurred due to unit connection interruptions in, For the connecting flight, which is a binary variable, its values ​​are as follows: If the connecting flight is operated... and If the connection of the unit is interrupted under scenario s, the value is 1; otherwise, it is 0.

[0123] The constraints of the flight cancellation and resumption strategy generation model also include: crew connection constraints, which are as follows:

[0124] Where M is a pre-defined, sufficiently large positive integer in minutes, and M should be greater than the maximum possible time difference. With minimum rest time The difference, for example, M can take the value 99999, which ensures that it is greater than any possible time difference in the model. As an indicator parameter, if a flight string r contains connecting flights and If the result is positive, the value is 1; otherwise, it is 0. This represents the flights in the flight string r. Actual takeoff time. This represents the flights in the flight string r. Actual landing time. Indicates connecting flights and The minimum ground rest time required for the unit.

[0125] Similarly, the method described in the above-mentioned optional embodiments can be used to solve the optimal solution of the objective function in this optional embodiment, thereby determining the early cancellation and flight recovery strategies for each scenario.

[0126] In actual operations, flight cancellations and resumptions not only affect passengers on individual flights but also disrupt the entire passenger itinerary. For example, when a flight is canceled or delayed, passengers on that flight who have booked connecting flights may miss their connecting flights. Whether the airline reschedules the passengers' connecting flights or the passengers have to cancel their connecting flights, it will result in losses.

[0127] Therefore, when building a model for generating flight cancellation and recovery strategies, the costs incurred by passenger travel disruptions and the relocation of passengers can also be taken into account.

[0128] Optionally, where the expected recovery cost also includes passenger recovery cost, the objective function is:

[0129] Where I represents the original set of passenger itineraries, and J represents the candidate set of passenger itineraries to be reset. This represents the set of flights included in the original passenger itinerary i. Let i represent the set of candidate reset passenger itineraries provided by the original passenger itinerary. This represents the unit cost of refunds or rebooking to other airlines for passengers whose original travel itinerary was interrupted. This represents the number of passengers in scenario s whose original travel itinerary i requires a refund or rebooking to another airline. Let $\frac{ ... This represents the number of passengers who are reassigned from their original travel itinerary i to a candidate reset travel itinerary j under scenario s. The flight cancellation and resumption strategy generation model also includes passenger connection constraints, passenger flow conservation constraints, feasibility constraints for resetting passenger itineraries, and flight capacity constraints. The passenger connection constraints are as follows:

[0130]

[0131] Where E represents the set of all flight connection pairs in the original passenger itinerary. As an indicator parameter, if resetting passenger itinerary j includes connecting flights and If the result is 1, then it equals 1; otherwise, it equals 0. For passenger connection variables, if passengers cannot connect to flights under scenario s. and If it is true, then it is 1; otherwise, it is 0. Indicates connecting flights and The shortest connection time required for passengers between the two routes. The passenger flow conservation constraint is:

[0132] in, Let i represent the set of candidate reset passenger itineraries provided by the original passenger itinerary. Let i be the number of passengers in the original passenger itinerary i. The feasibility constraint for resetting passenger itineraries is:

[0133] The flight capacity constraint is as follows:

[0134] in, As an indicator parameter, it equals 1 if resetting passenger itinerary j includes flight copy t, and 0 otherwise. Indicates airplane Seating capacity, This represents the delayed repetition set of flight l obtained from flight string r.

[0135] In this optional embodiment, the core algorithm module 120 of system 100 performs data mining and processing on the acquired weather forecast information, relevant information of each flight, and parameter configuration information. It can also determine the original passenger itinerary set consisting of all passenger itineraries under normal flight operation conditions when unaffected by weather; the candidate reset passenger itinerary set consisting of all arrangable candidate reset passenger itineraries when affected by weather; the set of flights included in the original passenger itinerary and their corresponding candidate reset passenger itinerary sets; the set of flight connection pairs in all original passenger itineraries; the shortest connection time required between connecting flights; and the unit cost of refunding or rebooking to other airlines for any passenger whose original passenger itinerary is interrupted, and any original passenger itinerary under any scenario. The number of passengers who need refunds or rebooking to other airlines, the cost of relocating passengers from their original itineraries to a candidate reset itinerary, the number of passengers who can be relocated from their original itineraries to a candidate reset itinerary in any given scenario, the set of available candidate reset itineraries for any original itinerary, the number of passengers in any original itinerary, the seat capacity of any aircraft in the aircraft set, the set of delayed duplicates of the corresponding flights obtained from any flight string (in flight recovery, a flight may take off on time or be delayed, and different flight recovery plans need to be distinguished; the set of delayed duplicates is the set of all possible flight recovery plans (also known as "flight duplicates" or "delay duplicates") for the corresponding flight), etc.

[0136] Through the core algorithm module 120, the early cancellation cost, passenger recovery cost, and expected recovery cost of each flight can be comprehensively considered. A constrained mathematical optimization model is used to construct a flight cancellation and recovery strategy generation model, where the objective function of this model is:

[0137] Where I represents the original set of passenger itineraries, and J represents the candidate set of passenger itineraries to be reset. This represents the set of flights included in the original passenger itinerary i. This represents the set of candidate reset passenger itineraries provided by the original passenger itinerary i. This represents the unit cost of refunding or rebooking a passenger whose original itinerary was interrupted. This indicates the number of passengers in scenario s whose original passenger itinerary i requires a refund or rebooking to another airline. Let $\frac{ ... This represents the number of passengers who are reassigned from the original passenger itinerary i to a candidate reset passenger itinerary j under scenario s.

[0138] The constraints of the flight cancellation and resumption strategy generation model also include: passenger connection constraints, passenger flow conservation constraints, feasibility constraints for resetting passenger itineraries, and flight capacity constraints. The passenger connection constraint is as follows:

[0139]

[0140] Where E represents the set of flight connection pairs in all original passenger itineraries. As an indicator parameter, if resetting passenger itinerary j includes connecting flights and If the result is positive, then it equals 1; otherwise, it equals 0.

[0141] in, Let be the passenger connection variable, a binary variable, and its values ​​are as follows: If, under scenario s, the passenger cannot connect to a flight... and If the result is positive, the value is 1; otherwise, it is 0.

[0142] in, Indicates connecting flights and The shortest connection time required for passengers between the two routes.

[0143] The passenger flow conservation constraint is:

[0144] in, This represents the set of candidate reset passenger itineraries provided by the original passenger itinerary i. The number of passengers in the original passenger itinerary i.

[0145] The feasibility constraint for resetting passenger itineraries is:

[0146] The flight capacity constraint is:

[0147] in, The parameter is 1 if resetting passenger itinerary j includes flight copy t, and 0 otherwise. Indicates airplane Seating capacity. This represents the delayed repetition set of flight l obtained from flight string r.

[0148] Similarly, the method described in the above-mentioned optional embodiments can be used to solve the optimal solution of the objective function in this optional embodiment, thereby determining the early cancellation and flight recovery strategies for each scenario.

[0149] According to another aspect of this application, a computer-readable medium is also provided, on which computer-readable instructions are stored, which can be executed by a processor to implement some or all of the foregoing method embodiments and / or optional embodiments.

[0150] It should be noted that the method embodiments and / or optional embodiments in this application do not strictly limit the order of execution of each step, as long as the method embodiments and / or optional embodiments can solve the defects existing in the prior art, achieve the inventive purpose of this application, and obtain beneficial effects. The method embodiments and / or optional embodiments in this application can be implemented in software and / or combinations of software and hardware. The software program involved in this application can be executed by a processor to implement the steps or functions of the above embodiments. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium.

[0151] Furthermore, part or all of this application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions invoking the methods of this application may be stored in a fixed or removable recording medium, and / or transmitted via data streams in broadcast or other signal carrying media, and / or stored in the working memory of a computer device operating according to the program instructions.

[0152] According to another aspect of this application, an apparatus for generating flight cancellation and recovery strategies is also provided. The apparatus includes: a memory for storing computer-readable instructions and one or more processors for executing the computer-readable instructions, wherein when the computer-readable instructions are executed by the processor, the apparatus is triggered to run part or all of the methods and / or technical solutions of the foregoing embodiments.

[0153] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0154] In this application, when terms such as "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" are used, the indicated orientation and / or positional relationship is based on the orientation and / or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation. Furthermore, some of the above terms, in addition to indicating orientation or positional relationship, can also be used to indicate other meanings; for example, the term "upper" can also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application according to the specific circumstances. Furthermore, the terms "installation," "setup," "equipped with," "connection," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral structure; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection via an intermediate medium; and they can refer to an internal connection between two devices, components, or constituent parts. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0155] Furthermore, the terms "first," "second," etc., are primarily used to distinguish different devices, units, modules, elements, circuits, or components (which may be the same or different in specific type and construction), and are not intended to indicate or imply the relative importance, order, and / or quantity of the indicated devices, units, modules, elements, circuits, or components. Unless otherwise stated, "a plurality of" means two or more.

[0156] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in the apparatus claims may also be implemented by a single unit or device through software and / or hardware.

Claims

1. A method for flight cancellation and recovery strategy generation, characterized in that, The method comprises: acquiring weather forecast information, relevant information of each flight and parameter configuration information; constructing a target function of a flight cancellation and recovery strategy generation model according to an early cancellation cost of a flight and an expected recovery cost, wherein the expected recovery cost comprises a delay cost, an aircraft exchange cost, a real-time cancellation cost and an airport imbalance cost, and wherein the target function is: wherein L denotes a set of flights, S denotes a set of scenarios, A denotes a set of aircrafts, denotes a set of flight strings executable by aircraft a, B denotes a set of airports where aircrafts completing flight string r arrive at the end of the pre-specified flight recovery period, denotes a cost generated by the early cancellation of flight l, denotes a probability of scenario s occurring, denotes a cost of real-time cancellation of flight l under scenario s, denotes a cost of assigning aircraft a to flight string r, denotes a penalty cost of airport b appearing unbalanced, is a first flight variable, which is 1 if flight l is cancelled early, and 0 otherwise, is a second flight variable, which is 1 if flight l is cancelled in real time under scenario s, and 0 otherwise, is a flight string variable, which is 1 if aircraft a flies flight string r under scenario s, and 0 otherwise, is a slack variable, which denotes the number of aircrafts needed to be supplemented at airport b to meet the balance requirement under scenario s; wherein the flight cancellation and recovery strategy generation model comprises flight coverage constraints, aircraft allocation constraints, time period capacity constraints and airport balance constraints, wherein the flight coverage constraints are: wherein is an indicator parameter, 1 if the flight string r contains the flight I, otherwise 0, the aircraft allocation constraints are: the time period capacity constraints are: wherein, denotes the set of flight strings executable by aircraft a under scenario s, is the number of takeoffs and landings used by flight string r in time period o, denotes the maximum number of takeoffs and landings for time period o under scenario s, denotes the set of takeoff and landing time periods under scenario s, the airport balance constraints are: wherein is a parameter indicating if the last flight of the flight string r landed at the airport b normally within the preset flight recovery period, then 1, otherwise 0, is the number of aircrafts required at the airport b at the end of the preset flight recovery period, wherein corresponding parameters of the target function, the flight coverage constraints, the aircraft allocation constraints, the time period capacity constraints and the airport balance constraints are determined according to the weather forecast information, the relevant information of each flight and the parameter configuration information; solving an optimal solution of the target function, and determining an early cancellation flight and a flight recovery strategy under each scenario according to the optimal solution.

2. The method of claim 1, wherein, The solving of the optimal solution of the target function comprises: An unexpected constraint variable for each flight under each scenario is set, the unexpected constraint variable for flight l under scenario s The following constraints are satisfied: wherein the flight coverage constraints are changed to: replacing the first flight variable in the objective function replacing the flight unexpected constraint variable in each scenario with a first objective function, wherein the first objective function is: solving an optimal integer solution of the first target function, and taking the optimal integer solution of the first target function as the optimal solution of the target function.

3. The method of claim 2, wherein, The solving of the optimal integer solution of the first target function comprises: introducing a Lagrange multiplier for non-anticipatory constraints, and decomposing the first target function into second target functions under each scenario, wherein a second target function under scenario s is: adopting column generation to solve an optimal solution of the second target function under each scenario, and according to the optimal solution of the second target function under each scenario, determining a linear relaxation solution of the first target function, and according to the linear relaxation solution of the first target function, determining a dual price corresponding to the non-anticipatory constraints and assigning the dual price to the Lagrange multiplier to update the second target function, and iteratively solving until an iteration end condition is met, to obtain an optimal linear solution of the non-anticipatory constraint variables of the first target function; if there is a non-integer in the optimal linear solution of the non-anticipatory constraint variables of the first target function, adopting a diving heuristic algorithm to iteratively adjust the non-integer in the optimal linear solution of the non-anticipatory constraint variables of the first target function and solving an optimal solution of the non-anticipatory constraint variables of the second target function under each scenario by column generation until there is no non-integer in the optimal linear solution of the non-anticipatory constraint variables of the first target function, to obtain an optimal integer solution of the non-anticipatory constraint variables of the first target function; according to the optimal integer solution of the non-anticipatory constraint variables of the first target function, solving an optimal integer solution of the second target function under each scenario, and according to the optimal integer solution of the second target function under each scenario, obtaining an optimal integer solution of the first target function.

4. The method of claim 2, wherein, Before solving the optimal solution of the target function, the method further comprises: constructing an information perfect model corresponding to each scenario of the flight cancellation and recovery strategy generation model, solving a dual feasible solution of an objective function of the information perfect model of each scenario, and determining the flight to be cancelled in advance according to the dual feasible solution of the objective function of the information perfect model of each scenario.

5. The method of claim 4, wherein, The solving of the dual feasible solution of the objective function of the information perfect model of each scenario, and the determination of the flight to be cancelled in advance according to the dual feasible solution of the objective function of the information perfect model of each scenario, comprises: According to the information perfect model of each scenario, an information perfect model in dual form is obtained, and an objective function of the information perfect model in dual form of each scenario is solved to obtain a corresponding dual feasible solution; According to each dual feasible solution, the flight to be cancelled in advance or in real time in at least one dual feasible solution is determined as a to-be-identified flight; The first dual cost and the second dual cost of the to-be-identified flight when the to-be-identified flight is cancelled in advance in each scenario are calculated respectively, and compared, and if the first dual cost is less than the second dual cost, the to-be-identified flight is determined as a flight to be cancelled in advance.

6. The method of claim 1, wherein, The expected recovery cost further comprises a crew connection cost, and the objective function is: where D denotes the set of flight connection pairs, the cost of crew connection disruption for flying a connection flight and , is the crew connection variable, which is 1 if the crew flying a connection flight and is disrupted in scenario s, and 0 otherwise, The flight cancellation and recovery strategy generation model further comprises a crew connection constraint, wherein the crew connection constraint is: wherein M is a preset positive integer, is an indicator parameter, if the flight string r contains a connecting flight and is 1, otherwise 0, denotes the actual take-off time of a flight in the flight string r; denotes the actual landing time of a flight in the flight string r; denotes the minimum ground rest time required by the crew for flying a connecting flight and .

7. The method of claim 1, wherein, The expected recovery cost further comprises a passenger recovery cost, and the objective function is: where I denotes the original passenger itinerary set, J denotes the candidate reset passenger itinerary set, denotes the set of flights contained in the original passenger itinerary i, denotes the set of candidate reset passenger itineraries offered for the original passenger itinerary i, denotes the unit cost of passenger refund or rebooking to other airlines for the disruption of the original passenger itinerary i, denotes the number of passengers in the original passenger itinerary i that need to be refunded or rebooked to other airlines under scenario s, denotes the cost of relocating passengers from the original passenger itinerary i to the reset passenger itinerary j, denotes the number of passengers relocated from the original passenger itinerary i to a candidate reset passenger itinerary j under scenario s, The flight cancellation and recovery strategy generation model further comprises a passenger connection constraint, a passenger flow conservation constraint, a reset passenger itinerary feasibility constraint, and a flight capacity constraint, wherein, The passenger connection constraint is: where E denotes the set of all flight connection pairs in the original passenger itinerary, is 1 if resetting passenger itinerary j contains a connection flight and is 1, otherwise 0, is the passenger connection variable, which is 1 if the passenger cannot connect flights and in scenario s, otherwise 0, denotes the minimum connection time required by the passenger between connecting flights and , The passenger flow conservation constraint is: wherein, a set of candidate repositioning passenger trips provided for the original passenger trip i, the number of passengers in the original passenger trip i, The reset passenger itinerary feasibility constraint is: The flight capacity constraint is: in, As an indicator parameter, it equals 1 if resetting passenger itinerary j includes flight copy t, and 0 otherwise. Indicates airplane Seating capacity, This represents the delayed repetition set of flight l obtained from flight string r.

8. A computer readable medium, characterized in that, computer readable instructions are stored thereon, the computer readable instructions are executed by a processor to implement part or all of the method of any one of claims 1 to 7.

9. An apparatus for generating flight cancellation and recovery strategies, characterized in that, The device comprises: one or more processors; and a memory storing computer readable instructions which, when executed, cause the processor to perform part or all of the operations of the method of any one of claims 1 to 7.

Images