Aircraft passenger integrated dynamic recovery method and system considering speed control and medium

Abstract

The invention discloses an aircraft passenger integrated dynamic recovery method and system considering speed control and a medium. The method comprises the following steps: firstly generating an interference scene possibly encountered by an airline company in an operation process, solving the interference scene based on an ALNS heuristic algorithm to obtain a static recovery scheme, constructing an interference scheme pool, and when the airline company encounters interference in actual operation, matching the most similar interference scene in the scheme pool to quickly obtain a recovery scheme. And the scheme is adaptively modified according to the actually callable resources. The method is different from a traditional interference recovery method that the interference is considered to be static and predictable, the interference is considered to be dynamic and unpredictable, resources before the interference is generated cannot be utilized during recovery, and the method is more in line with an actual operation scene of an airline company. Moreover, compared with a traditional heuristic algorithm, the method can obtain a relatively optimal recovery scheme more quickly, and meets the urgency requirement of an airline company for the recovery scheme.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of civil aviation transportation management, and particularly relates to an aircraft passenger integrated dynamic recovery method considering speed control, a system and a medium. BACKGROUND

[0002] In the civil aviation transportation industry, airlines always expect each flight to arrive on time and all passengers to complete their journey smoothly, but in reality, the operation of the aircraft is always disturbed by various factors, including but not limited to air traffic control caused by weather factors, and cannot be used due to mechanical failure of the aircraft. The occurrence of these unexpected events will cause flight delays, cancellations or need to be rescheduled, thereby affecting the normal execution of the flight plan, so the airlines need to take effective measures to deal with these interference factors to ensure the normal execution of the flight.

[0003] Currently, there have been many research results on the problem of abnormal flight recovery, including "Optimization Scheduling of Disturbed Flights Based on Improved Column Generation Algorithm" published by Tian Qianan in System Engineering Theory and Practice in November 2019, "Airport Large-area Flight Recovery Planning Based on Uncertain Factor Disturbance" published by Zhou Yu in Science, Technology and Engineering in 2018, the invention patent "Abnormal Flight Recovery Optimization Method Based on Weight and Loss Cost" with publication number CN117689155A, and the invention patent "Flight Recovery System Based on Flight Value Evaluation" with publication number CN111353721A. Although the current journal papers and patents consider the flight disturbance recovery problem from many different angles, in these studies, the occurrence of disturbance is considered as predictable, rather than real-time and dynamic, which makes the research inconsistent with the actual situation. Therefore, under the premise of considering speed control as a recovery means, the application constructs an ALNS-IRAP (ALNS for Integrated Recovery of Aircraft and Passengers) algorithm, and based on this, further proposes a dynamic recovery method based on the interference scheme pool. SUMMARY

[0004] The purpose of the application is to provide an aircraft passenger integrated dynamic recovery method considering speed control, a system and a medium.

[0005] The purpose of the application is achieved by the following technical solutions: An aircraft passenger integrated dynamic recovery method considering speed control, comprising: Step 1: When the flight operation of a new day begins, generate a number of interference scenarios according to the flight operation data and historical interference data of the airline; Step 2: Use the ALNS-IRAP algorithm to generate a recovery scheme for each generated interference scenario and form a scheme pool; Step 3: When actual interference occurs, if there are enough interference scenarios in the scheme pool, then find the interference scenario most similar to the actual interference from the scheme pool, and adaptively improve its recovery scheme to obtain the actual recovery scheme; if there are not enough interference scenarios, then use the ALNS-IRAP algorithm to directly solve the actual interference to obtain the actual recovery scheme. Step 4: Perform the recovery operation using the actual recovery plan described above; Step 5: Determine whether the airline's operations for the day have ended. If not, continue to generate new interference scenarios based on the actual recovery plan, and repeat steps 2 to 4 until operations end.

[0006] Furthermore, the generated interference scenarios include: aircraft type, number of aircraft of each type, passenger capacity of each type of aircraft, available time of each aircraft, initial airport of each aircraft, departure and arrival airports, estimated departure and arrival times of flights, number of seats sold for each flight, rebooking cost per passenger when rebooking between flights, delay cost per passenger when flights are delayed, refund cost per passenger when flights are canceled, number of flights each aircraft is scheduled to operate, curfew times for each airport, and fuel prices; the generated interference scenarios include: aircraft being unavailable for a period of time, airports being unable to land for a period of time, and flight delays.

[0007] Furthermore, the ALNS-IRAP algorithm includes: (12) Constructing the initial solution set: The initial flight recovery plan, i.e. the initial solution, is generated by the natural delay of the original plan; (13) Record the current solution and the optimal solution as the initial solution. Iterate and optimize the current solution through the destruction operator and the repair operator to form a new solution. Compare the recovery cost of the new solution and the optimal solution. If the recovery cost of the new solution is less than that of the optimal solution, update both the optimal solution and the current solution to the new solution and proceed to the next iteration. If the recovery cost of the new solution is greater than that of the optimal solution, there is a probability that the new solution will be updated to the current solution, the optimal solution will remain unchanged, and proceed to the next iteration. (14) The algorithm ends when the number of iterations reaches the preset maximum value.

[0008] Furthermore, the method of generating an initial flight recovery plan based on the natural delays of the original plan includes: scheduling flight departure times according to the actual available aircraft time, with subsequent flights naturally adjusting for delays, and if the aircraft... Available time is The first flight in the first flight loop corresponding to this aircraft. The scheduled departure time is ,like earlier than then all the flights corresponding to the aircraft run as scheduled; if later than then the scheduled departure time of the first flight will be changed to , the departure time of the second flight in the first flight loop will be changed to , and the rest of the flights corresponding to the aircraft perform the same operation as above, obtaining the departure and arrival time information of all flights in the flight loop that the aircraft flies; if the subsequent flights cannot meet the curfew time limit, they will be directly canceled.

[0009] Further, the destruction operator is selected from the destruction operator set in a roulette manner according to a set probability, including a shaw removal operator, a random removal operator, and a worst removal operator; and the repair operator is selected from the repair operator set in a roulette manner according to a set probability, including a greedy insertion operator, a regret insertion operator, a check-in priority insertion operator, and a check-out priority insertion operator.

[0010] Further, the recovery cost is minimized by the following formula: = +

[0011] wherein, is the recovery cost of the entire recovery scheme; F is all flight sets ; is the delay cost of the flight along the arc ; is the additional fuel cost of the flight due to acceleration; is the cost of exchanging aircrafts of flights; is the cost of refunding tickets in the flight.

[0012] Further, the disturbance most similar to the actual disturbance is found from the scheme pool, and the similarity calculation formula is:

[0013] wherein, , , respectively represent the distance of flight disturbance, the distance of aircraft disturbance, and the distance of airport disturbance in two disturbance scenarios, and are represented using Hausdorff distance.

[0014] Furthermore, the corresponding recovery scheme is adaptively improved based on the partial search heuristic algorithm, specifically including: Reposition all disrupted flight sequences in descending order of disruption severity: For each disrupted flight sequence, attempt to remove it from the original aircraft path and insert it into other aircraft paths, assess the change in recovery costs, and accept the repositioning that maximizes the reduction in recovery costs; repeat this process for all disrupted flight sequences until the objective cannot be improved. Cross-insertion between interfering flight sequences and another flight sequence from a different path: Select the flight sequence with severe interference, randomly select another flight sequence, evaluate the cost of the exchange insertion, and accept it if the cost is reduced; repeat the iteration until no improvement can be obtained.

[0015] Furthermore, in the process of finding the most similar interference scenarios, when there are too many scenarios in the solution pool, the k-means clustering algorithm is used to cluster the interference scenarios, and the similarity of the scenarios at the cluster centers is calculated.

[0016] A computer device / equipment / system includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of an integrated dynamic recovery method for aircraft passengers that takes speed control into account.

[0017] The beneficial effects of this invention are as follows: This invention provides a solution to the flight recovery problem, enabling recovery plans to better align with airlines' actual operations and offering significantly faster response times. The provided recovery plans not only meet airlines' needs but also can be derived within a short timeframe, effectively achieving a good solution to the flight recovery problem. Attached Figure Description

[0018] Figure 1 This is a flowchart of the integrated dynamic recovery method for aircraft passengers that takes speed control into account, as described in this invention. Figure 2 This is a flowchart of the ALNS-IRAP algorithm.

[0019] Figure 3 This is a comparison chart of three recovery methods: dynamic algorithm, static algorithm, and delay-only. Detailed Implementation

[0020] The present invention will now be further described with reference to the accompanying drawings.

[0021] This invention discloses an integrated dynamic recovery method, system, and medium for aircraft passengers that considers speed control. Figure 1 The specific steps of the method are as follows: Step 1: When the new day's flight operation begins, generate disturbance scenarios according to the disturbance data suffered by the airline in the past.

[0022] The flight operation data includes: the type of aircraft, the number of each type of aircraft, the passenger capacity of each type of aircraft, the available time of each aircraft, the initial airport of each aircraft, the departure and arrival airport of the flight, the estimated departure time and the estimated arrival time of the flight, the number of seats sold for the flight, the rebooking cost of each passenger when rebooking between flights, the delay cost of each passenger when the flight is delayed, the refund cost of each passenger when the flight is canceled, the number of flight trips each aircraft is expected to fly, the curfew time of each airport, and the fuel price.

[0023] The disturbance scenarios include three scenarios: aircraft unavailability for a period of time, airport unavailability for a period of time, and flight delay.

[0024] Step 2: For each generated scenario, an ALNS-IRAP (ALNS for Integrated Recovery of aircraft and passengers) algorithm is used to generate a recovery scheme according to the disturbance scenario, and the recovery scheme is stored in the scheme pool.

[0025] According to Figure 2 , the flow of the ALNS-IRAP algorithm is as follows: (1) Construct an initial solution. Generate an initial flight recovery scheme by naturally delaying the original scheme, and pass the solution to step 3; where the flight recovery scheme is represented in the form of a flight sequence of aircraft, and the specific rules are as follows: assuming that the number of aircraft operated by the airline each day is n, the entire flight recovery scheme is represented as , , where is the flight sequence of the th aircraft, , , .

[0026] (2) Destroy and repair the current solution by the ALNS-IRAP algorithm. Compare the solution obtained in each iteration with the initial solution of the iteration, update the current solution with a certain probability, and update the weights of the corresponding operators.

[0027] (3) Compare the solution obtained in each iteration with the optimal solution to determine whether to update the optimal solution, with the optimization objective being to minimize the recovery cost. Loop and update the simulated annealing temperature and operator weights every certain number of iterations, and end when the number of iterations reaches the initial maximum number of iterations.

[0028] The generating initial recovery scheme adopts a natural delay scheme, specifically: arranging the flight departure time according to the actual available time of the aircraft, and naturally adjusting the subsequent flight delay, if the available time of the aircraft is , the scheduled departure time of the first flight in the flight sequence corresponding to the aircraft is , if is earlier than , then all flights corresponding to the aircraft can be operated according to the original plan, if is later than , then the scheduled departure time of will be changed to , and the departure time of the second flight in the first flight ring will be changed to , wherein 30 is the minimum interval required for the same aircraft to continuously fly two flights. The same operation is performed on the remaining all flight rings corresponding to the aircraft to obtain the departure and landing time information of the flights in all flight rings flown by the aircraft , and if the subsequent flight cannot meet the curfew time limit, it will be directly cancelled. The minimum recovery cost formula is:

[0029] = +

[0030] The parameter meanings are as follows: wherein, is the recovery cost of the entire recovery scheme; F is all flight set ; is the delay cost of flight along the arc ; is the additional fuel cost of flight due to acceleration; is the cost of flight exchange aircraft; is the cost of ticket refund in the flight.

[0031] The destruction of the solution is realized by removing the operator: The removal operator is selected from the removal operator set by roulette in a probability, and the similar method is also suitable for the selection of the insertion operator. In the ALNS algorithm first proposed, three removal operators, shaw removal operator, random removal operator and worst removal operator, are used, and the present application transplants them into the aircraft passenger dynamic recovery problem considering speed control.

[0032] The shaw removal operator is: ​​The Shaw removal operator selects and removes a certain number of elements based on a similarity metric. By comparing features such as distance, time window, and service time between elements, it identifies and removes similar elements. After removal, these elements are reinserted into the solution to obtain a new solution. The Shaw removal operator is based on the high substitutability between similar elements, thus having a high probability of obtaining a better solution after swapping them. Therefore, it is necessary to define the similarity between two flight loops / flight sequences.

[0033]

[0034]

[0035] It includes five items, among which flight sequence and Distance between departure airports and flight sequence and Distance between destination airports ,sequence and The time difference between the departure of the first flight in the middle and sequence and Time difference of arrival of the last flight Flight sequence and Total delay time difference between Flight sequence and The difference in the total number of passengers and sequences and Aircraft type can be used to assess the similarity of service flights, which is calculated and measured here by the number of aircraft available to cover that flight. In the formula, these are respectively determined by… .

[0036] Based on the above definition of similarity, the design of the Shaw removal operator is as follows: 1) Randomly select a flight ring or flight sequence and remove it from the solution.

[0037] 2) Randomly select one of the removed flight loops as a baseline, and calculate the similarity between the remaining flight loops and flight sequences and the removed flight loops or flight sequences.

[0038] 3) For the remaining flight loops and flight sequences, based on their relationship with the removed flight loops or flight sequences... Sort by similarity and generate a random number. Remove the first a number of flight sequences. Where the algorithm controls the randomness of the selection process through the parameter p.

[0039] 4) Repeat 2) and 3) until a pre-set removal number threshold is reached.

[0040] The random removal operator: This operator randomly selects a number of elements and removes them from the current solution. These removed elements will be re-inserted into the solution later to find new solutions. The design idea of is to increase the exploration of the solution space. By randomly removing elements, the algorithm can explore different areas in the solution space. This helps to find the global optimal solution and avoid the algorithm falling into local optimum, which is designed as follows:

[0041] 1) Randomly select flight rings or flight sequences from the solution; 2) Remove these flight rings or flight sequences from the solution and put them into the "flight ring pool" or "flight sequence pool" for subsequent insertion for use; 3) Return the updated solution.

[0042] The worst removal operator: The worst removal operator first evaluates the impact of each element on the quality of the solution, and then removes the worst elements from the current solution according to the evaluation results. The design idea of the worst removal operator is to identify and eliminate low-quality parts of the solution to improve the overall quality of the solution.

[0043] 1) For each flight ring or flight sequence in the original solution, calculate its cost; the cost is defined as the difference between the cost of the flight ring or flight sequence in the original solution and the cost of the flight ring or flight sequence not in the original solution; the cost can be calculated using the number of times the flight ring or flight sequence violates the speed control limit in the original solution; 2) For flight rings or flight sequences with a cost higher than the average cost, remove them from the original solution and add them to the "flight ring pool" or "flight sequence pool" according to a certain probability; at this time, the randomness and convergence speed of the algorithm can be controlled by changing the probability; 3) Return the updated solution.

[0044] The repair of the solution in the ALNS-IRAP algorithm is implemented through the insertion operator: After removing the flight sequences in the current aircraft path solution, the algorithm will apply the insertion operator to reassign these flight sequences to other aircraft paths. Given a flight sequence to be inserted and an aircraft path, the algorithm will try to insert the flight sequence into the aircraft path at an appropriate position. If each flight in the flight sequence and each flight after the insertion position in the aircraft path meets the maximum flight departure delay limit and the airport curfew time limit, the target change value can be calculated by the following formula .

[0045]

[0046]

[0047]

[0048]

[0049] wherein, f represents a set of flights in a flight sequence l n represents the position where the flight sequence f can be inserted into the aircraft path p, represents a set of flights after position n in the aircraft path f, represents the cruise speed of the aircraft p during the recovery period of flight f, represents the original cruise speed, the = for short-haul flights, and the for long-haul flights, represents the unit delay cost of each passenger, represents the actual departure time of flight f, which can be calculated by adding the maximum value of the actual arrival time of the previous flight and the minimum transfer time to the planned departure time of flight f.

[0050] There are four kinds of insertion operators: greedy insertion operator, regret insertion operator, sign-in priority insertion operator, and sign-out priority insertion operator.

[0051] The greedy insertion operator: The greedy insertion operator inserts elements into the position that minimizes the objective function value based on the greedy strategy. The greedy strategy can find a seemingly suitable insertion position in a short time, thereby speeding up the search process. However, the greedy strategy may lead to a local optimal solution, which is designed as follows:

[0052] 1) For each flight sequence in the "flight ring pool" or "flight sequence pool", calculate its cost of insertion into the aircraft path; 2) Find the flight sequence that minimizes the delay cost , and insert it into the best insertion aircraft path; 3) Repeat step 2) until no more flight sequences can be inserted.

[0053] The regret insertion operator: The regret insertion operator takes into account the potential regret value during the insertion process. In this algorithm, the regret value is the loss that can be caused by not repairing a certain flight loop or flight sequence in this insertion process. The operator selects the element with the largest regret value for insertion and inserts it into the position that minimizes the regret value.

[0054] 1) For each flight loop or flight sequence in the "flight loop pool" or "flight sequence pool", calculate its insertion cost into the aircraft path ; 2) Sort all aircraft routes to ensure that is satisfied; ≤ ; 3) Select a flight sequence and insert it into the best insertion aircraft path; 4) Repeat steps 2) and 3) until no more flight loops or flight sequences can be inserted.

[0055] The check-in priority insertion operator: The previous insertion operator mainly focuses on the delay cost, additional fuel cost and flight exchange cost. However, it ignores the contribution of passenger itinerary rescheduling to the objective function. Based on this, the general idea of this operator is to insert flight sequences that can provide more seats for passenger itinerary rescheduling. Therefore, the flight sequence can be calculated by the following formula, where represents the number of itineraries that passengers can be rescheduled to cover flights . Its steps are similar to the greedy insertion operator.

[0056]

[0057] The check-out priority insertion operator: Similar to the check-in priority insertion operator, this operator focuses on the availability of passenger reassignment to other itineraries. It aims to insert flight loops or flight sequences that serve the smallest passenger reassignment opportunities, as these flights correspond to itineraries that are difficult to find suitable rescheduling strategies, so once the flight loop or flight sequence is not repaired and is forced to cancel, the refund cost will increase sharply. Therefore, the algorithm calculates the check-out value of each flight loop or flight sequence by the following formula, and then selects the element for insertion from small to large according to the check-out value, which symbolizes the size of the reassignment opportunity of the flight loop or flight sequence. Where is the itinerary that can receive the reassignment of passengers from the itinerary covering flights , is the travel time of the original itinerary and the travel time of the new itinerary the ratio of the difference between the actual arrival time and the original departure time , , the larger indicates that the travel time of the trip re-planning is more similar to the original trip. Its steps and greedy insertion operator are similar, the only difference is that the index of the flight sequence or flight ring is signed out value rather than cost .

[0058]

[0059]

[0060] Step 3: When the actual disturbance occurs, check whether enough disturbance scenarios are stored in the scheme pool, if yes, find the most similar disturbance from the scheme pool to the actual disturbance, and adaptively improve the solution; if not, directly solve the problem using the ALNS-IRAP algorithm; The most similar disturbance scenario is found, when the scene in the scheme pool is too much, the k-means clustering algorithm is used to cluster the disturbance scenarios, and the similarity of the scene of the cluster center is calculated; The similarity calculation formula is , wherein , , respectively represent the distance of flight disturbance, the distance of aircraft disturbance and the distance of airport disturbance in two disturbance scenarios.

[0061] Step 4: Judge whether the operation of the airline on the day is over, if not, continue to generate new disturbance scenarios based on the recovery scheme obtained in S3, generate a recovery scheme according to the disturbance scenario using the ALNS-IRAP algorithm, and store the recovery scheme in the scheme pool. If a new disturbance occurs in reality, continue to repeat step S3 until the operation of the airline on the day is over.

[0062] Embodiment: The application is a dynamic recovery problem of aircraft passenger integration considering speed control, and the problem aims to minimize the total cost related to passenger delay, transfer, fuel and ticket refund, etc.

[0063] The difference between it and the static recovery problem is that the static recovery problem assumes that all disturbances are known before the airline starts a new day of operation, and all aircraft, crew and other resources within the day can be mobilized to recover the disturbance. The dynamic recovery problem assumes that the disturbance occurs in real time during the operation, and cannot mobilize the resources used before the disturbance is found, such as knowing that a disturbance will occur after 14 o'clock at 13 o'clock. It is impossible to recover by scheduling aircraft resources before 13 o'clock.

[0064] Since the problem considered in this invention is dynamic, disturbances occurring randomly during the whole planning execution period are handled dynamically, the invention assumes that the airline is able to monitor the occurrence of disturbances effectively during the planning execution process and inform the crew and operation control center in time. In this case, the dynamic recovery problem can be divided into different sub-problems according to the occurrence of disturbances. When all disturbances are recovered in one stage, the problem will become a completely static problem.

[0065] For each sub-problem in the dynamic recovery problem, it can be considered as a static recovery problem, ensuring that no conflict with the previous sub-problem occurs when adding constraints. This means that only resources from the current time and after can be used for recovery when solving each sub-problem. In the sub-problem, the invention retains most of the assumptions of the static problem, including that a passenger itinerary contains at most two flights, i.e. one transfer; crew and aircraft are considered to be one-to-one correspondence, without additional consideration of their matching and re-matching; only two classes of business and economy are concerned, without involving other classes, and the transfer of passengers between different classes is not discussed; four recovery measures, i.e. flight delay, cancel flight, flight exchange and control speed, are considered, without considering other measures such as spare aircraft; in the process of recovering passengers, passengers are considered to follow the airline's arrangement, and the recovery of passenger itinerary is carried out through delay, transfer and refund, without considering the loss caused by passengers against their will; delay cost, transfer cost, cancellation cost and fuel cost are concerned, without involving other costs such as the average cost of using aircraft; only the airline's plan within one day is considered, and the recovery time limit is within the day.

[0066] In addition, the dynamic recovery problem relaxes the constraint that the number of different types of aircraft at each airport at the end of the recovery time window should be the same as at the beginning of the plan, as it is difficult to ensure that this constraint is met during the recovery process due to the dynamic nature of disturbance occurrence in dynamic recovery.

[0067] The goal of dynamic recovery is still to minimize the total recovery cost, but unlike the goal of static recovery, the target value of the dynamic recovery problem is not obtained by solving it once, but by synthesizing the results of solving all sub-problems. Since each sub-problem is incomplete, its target value is also local, and the recovery scheme in the current segment may directly enter the next segment because the next disturbance cannot be fully executed, so its target value cannot be simply added. After the last sub-problem is solved, the flight plan and passenger itinerary plan of the whole dynamic recovery problem are finally determined. Therefore, for each sub-problem, only the resources determined in the current segment are considered when calculating the target, and finally the target values of all sub-problems are added to obtain the target value of the total problem.

[0068] Therefore, the goal of the dynamic recovery problem can be defined as: : recovery cost; total recovery period; : total recovery period of the first segment, satisfying ; no earlier than segment flight related flight arc set, wherein ; : no earlier than spatiotemporal point set of the segment; The total goal of dynamic recovery is:

[0069] wherein, is the sub-problem target value of the segment.

[0070] Although the spatiotemporal network model reduces the complexity of the model by discretizing the nonlinear function, the number of constraints increases exponentially due to the large number of delay and acceleration arcs in the spatiotemporal network, which is tens or even hundreds of times the number of flights, which brings great challenges to the exact algorithm for solving. Moreover, as the size of the flight increases, even the heuristic algorithm, the solving time also increases significantly, especially under the consideration of dynamic recovery, the requirement for the response speed of the algorithm is further improved, generally the heuristic algorithm is used, therefore the application proposes an ALNS-IRAP algorithm with good effect to construct the interference recovery scheme pool, and further proposes a dynamic recovery method based on the pre-generated interference scheme pool.

[0071] Figure 1 The framework of the proposed dynamic recovery model is shown. First, the flight plan and passenger itinerary plan are normally executed, and preparation for possible disturbances is started, and the possible disturbances and recovery schemes are pre-generated, at this time the pre-generated solutions will be used for the recovery of the disturbance occurred in the next stage. At the same time, it is constantly detected whether a disturbance has occurred, if no disturbance has occurred, the execution of the plan is continued.

[0072] When the disturbance occurs, the size of the pre-generated solution pool may not be enough to effectively recover the sub-problem of this stage, at this time, the ALNS-IRAP algorithm is selected to recover the sub-problem of the next stage. If the size of the pre-generated solution pool is sufficient, the disturbance situation generated in the last stage is matched, the most similar disturbance situation is found, the corresponding solution is adaptively improved, and is used as the recovery plan of this stage.

[0073] In the aspect of matching the disturbance situation, the present application assumes that if the position, type and severity of two disturbances are more similar, the matching degree of them is higher.

[0074] Although the candidate recovery solution of the matching scene is close to the prepared solution before the interruption, it may be far from optimal or even feasible for the actual disturbance. Therefore, the present application will adjust and improve the candidate recovery solution of the matching scene, and then select the best solution as the final solution of the actual disturbance.

[0075] The key idea of this adjustment is that the candidate recovery solution is used as the input of the recovery process to preheat the new solution, and all the requests served before the disturbance will be removed from the new solution. According to the candidate solution, the vehicle route may be the same, but the arrival time and service time may be completely different due to the difference between the actual disturbance and the historical interruption scene. In addition, the newly arrived request should also be added to the set of unfinished requests.

[0076] Therefore, the relevant travel cost and the cost of deviating from the original schedule will be adjusted according to the actual disturbance situation. The improvement strategy is mainly based on a local search heuristic algorithm, which has two steps executed iteratively in the following order.

[0077] (1) Re-position all disturbed flight sequences in descending order of disturbance severity. The algorithm first tries to remove the most severe flight sequence from the original aircraft path, and evaluates the result if it is inserted into all other aircraft paths. For the disturbed flight sequence, the re-positioning that reduces the recovery cost the most among all possible insertions will be accepted. This process will be repeated for all disturbed flight sequences until no improvement can be made.

[0078] (2) Cross insertion between a disturbed flight sequence and another flight sequence from a different path. Similar to the first strategy, the algorithm first sorts the disturbed flight sequences in descending order of disturbance severity, and then selects the most severe flight sequence. Then, the algorithm randomly selects another flight sequence from other aircraft paths, and evaluates the overhead if the algorithm inserts each flight sequence into another aircraft path. If the cost from the first step is lower than the current solution, the algorithm will adopt it. Repeat the iteration until no improvement can be obtained.

[0079] To further illustrate the specific effect of the present application, a test example is given here based on the actual flight recovery of a small airline on a certain day. In this example, there are 47 aircrafts and 162 flights involved in the recovery. The original ALNS-IRAP algorithm and the dynamic recovery algorithm are used respectively to give a solution to the flight recovery, and the iteration number is set to 5000 times. The final results are as follows Figure 3 The D1-5 scenarios respectively include multiple disturbances occurring at different time periods. After running, it can be seen that in the D2, D4 and D5 examples, the dynamic solution method differs greatly from the static solution method, which is difficult to explain by the difference between the dynamic solution framework and the ALNS-IRAP algorithm. This part of the difference reflects the cost of the reactive solution of the dynamic recovery. Compared with the static recovery, the dynamic recovery can only recover the disturbance with partial information and partial resources, so it cannot comprehensively and integrally recover. In the static solution, the present application assumes that all disturbances can be predicted in advance, but in the real world, the airline cannot predict the occurrence of disturbances, so if the static method is used to solve the disturbance recovery problem, the recovery cost will be greatly underestimated.

[0080] From the optimization results, it can be seen that the technical solution in the present application can solve the flight recovery problem, so that the recovery scheme is more in line with the actual operation of the airline, and the response speed is significantly faster, that is, the recovery scheme can meet the needs of the airline, and the recovery scheme can be obtained in a relatively short time, and the flight recovery problem can be solved well.

[0081] In particular, in some preferred embodiments of the present application, a computer device is also provided, which includes a memory and a processor and a computer program stored in the memory, and the processor implements the steps of the flight passenger integrated dynamic recovery method considering speed control in any of the above embodiments when executing the computer program.

[0082] In some other preferred embodiments of the present application, a computer readable storage medium is also provided, which stores a computer program / instructions, and the computer program is executed by a processor to implement the steps of the flight passenger integrated dynamic recovery method considering speed control in any of the above embodiments.

[0083] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the flight passenger integrated dynamic recovery method considering speed control in the above-mentioned embodiments, which will not be repeated here.

[0084] Computer-readable storage media encompasses a variety of media, both physical and non-physical that can be accessed by a computer. Such media include, but are not limited to, optical, electrical and electromagnetic media such as compact disc (CD), digital versatile disc (DVD), memory devices such as read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), non-volatile memory (e.g., flash), volatile memory, etc., magnetic media, such as diskette, floppy disk or hard drive, solid state media, such as memory stick, etc., and any other medium that can be used to either access, store, transport, or carry information. Note that the computer-readable media discussed herein are computer-readable storage media. Computer-readable storage media does not include transitory media, such as modulated data signals and carrier waves.

[0085] Those skilled in the art will further appreciate that the functionality of the modules can be implemented by logic encoded on one or more tangible media (e.g., embedded logic such as an ASIC, discrete elements such as resistors, capacitors, inductors, transistors, etc., or combinations of the above) that work in concert to create the required functionality, and that expressions, such as "programmed / instructions" and "executing program steps" can include the logic encoded on one or more tangible media and one or more processors, such as an FPGA or ASIC, to facilitate the enabling, disabling, and modifying of the functionality described herein.

[0086] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in RAM, flash memory, ROM, electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC.

[0087] The implementations described herein are implemented as logical steps in one or more algorithms that are implemented and executed in a computing environment. The computing environment can include one or more computer processors. The one or more computer processors can be specialized processors or general purpose computing devices. The implementations can be described in the context of a personal computer, in which the one or more computer processors are general purpose computers. Embodiments described herein can be implemented in the context of general purpose computers and / or special purpose computers. The implementations described herein can be implemented in the context of a general purpose computer and / or a special purpose computer.

Claims

1. An aircraft passenger integration dynamic recovery method considering speed control, characterized in that, The application relates to a flight recovery method based on ALNS-IRAP algorithm. Step 1: when a new day of flight operation starts, a plurality of interference scenarios are generated according to flight operation data and historical interference data of an airline; Step 2: an ALNS-IRAP algorithm is used to generate a recovery scheme for each interference scenario, and a scheme pool is formed; Step 3: when an actual interference occurs, if there are enough interference scenarios in the scheme pool, a recovery scheme most similar to the actual interference is searched from the scheme pool, and the recovery scheme is adaptively improved to obtain an actual recovery scheme; if there are not enough interference scenarios, the actual interference is directly solved by using the ALNS-IRAP algorithm to obtain an actual recovery scheme; Step 4: the actual recovery scheme is used for recovery operation; Step 5: whether the operation of the airline is ended is judged, if not, a new interference scenario is generated based on the actual recovery scheme, and steps 2 to 4 are repeated until the operation is ended.

2. The method of claim 1, wherein, The flight operation data includes the types of aircraft, the number of each type of aircraft, the passenger capacity of each type of aircraft, the available time of each aircraft, the initial airport of each aircraft, the takeoff and landing airport of a flight, the expected takeoff time and the expected landing time of the flight, the number of sold seats of the flight, the rebooking cost of each passenger when rebooking between flights, the delay cost of each passenger when a flight is delayed, the refund cost of each passenger when a flight is cancelled, the flight class of each aircraft, the curfew time of each airport and the fuel price; the generated interference scenario includes that an aircraft is unavailable for a period of time, an airport cannot land for a period of time and a flight is delayed.

3. The method of claim 1, wherein, The ALNS-IRAP algorithm comprises the following steps: (3) constructing an initial solution set: generating an initial flight recovery scheme by naturally delaying the original scheme, that is, an initial solution; (4) recording the current solution and the optimal solution as the initial solution, iteratively optimizing the current solution by a destruction operator and a repair operator to form a new solution, comparing the recovery cost of the new solution and the optimal solution, if the recovery cost of the new solution is smaller than that of the optimal solution, updating the optimal solution and the current solution to the new solution, and performing the next iteration; if the recovery cost of the new solution is larger than that of the optimal solution, updating the new solution to the current solution with a probability, and keeping the optimal solution unchanged, and performing the next iteration; (3) ending the algorithm when the number of iterations reaches a preset maximum value.

4. The method of claim 3, wherein, The method of generating an initial flight recovery plan based on natural delays from the original plan includes: scheduling flight departure times according to actual available aircraft slots, with subsequent flights naturally adjusting for delays; if the aircraft... Available time is The first flight in the first flight loop corresponding to this aircraft. The scheduled departure time is ,like Earlier Then all flights corresponding to that aircraft will operate as originally planned; if Later Then the first flight The scheduled departure time will become The second flight in the first loop The takeoff time will become ,airplane Perform the same operation as above on all other flights to obtain the aircraft. Information on the departure and arrival times of all flights operated by the company within the designated area; if subsequent flights cannot comply with the curfew time restrictions, they will be cancelled directly.

5. The method of claim 3, wherein, The destruction operator is selected from a destruction operator set in a roulette manner with a set probability, and the destruction operator set comprises a shaw removal operator, a random removal operator and a worst removal operator; the repair operator is selected from a repair operator set in a roulette manner with a set probability, and the repair operator set comprises a greedy insertion operator, a regret insertion operator, a check-in priority insertion operator and a check-out priority insertion operator.

6. The method of claim 3, wherein, The recovery cost is minimized by the following formula: = + The parameters have the following meanings: where is the recovery cost of the overall recovery scheme; F is the set of all flights ; is the flight delay cost along the arc ; is the flight additional fuel cost due to acceleration; is the flight change aircraft cost; is the flight cancellation cost.

7. The method of claim 1, wherein, The interference most similar to the actual interference is searched from the scheme pool, and the similarity calculation formula is as follows: wherein, , , respectively denote the flight interference distance, the aircraft interference distance and the airport interference distance in the two interference scenarios and are expressed using the Hausdorff distance.

8. The method of claim 1, wherein the method further comprises: The corresponding recovery scheme is adaptively improved based on a local search heuristic algorithm, and the adaptive improvement comprises the following steps: repositioning all disturbed flight sequences in descending order of disturbance severity: for each disturbed flight sequence, try to remove it from the original aircraft path and insert it into other aircraft paths, evaluate the change in recovery cost, accept the repositioning that reduces the recovery cost the most; repeat this operation for all disturbed flight sequences until no improvement can be made to the objective; cross-insertion between a disturbed flight sequence and another flight sequence from a different path: select a severely disturbed flight sequence, randomly select another flight sequence, evaluate the cost after swapping insertion, if the cost is reduced, accept it; repeat the iteration until no improvement can be obtained.

9. The method of claim 1, wherein the method further comprises: The most similar interference scenarios are found, and when there are too many scenarios in the scheme pool, a k-means clustering algorithm is used to cluster the interference scenarios, and the similarity of the scenarios of the cluster centers is calculated.

10. A computer apparatus / device / system comprising a memory, a processor, and a computer program stored on the memory, characterized in that: The processor executes the computer program to implement the steps of the method of any one of claims 1 to 9.

Images