Flight arrival and landing scheduling method based on hyper-heuristic algorithm

Abstract

The invention discloses a flight arrival and landing scheduling method based on a hyper-heuristic algorithm, and mainly relates to the field of air traffic management and intelligent optimization. An existing flight arrival and landing scheduling method is poor in speed and precision, and an optimal landing scheme cannot be found in a short time. According to the method, a randomly selected hyper-heuristic strategy is adopted, and a better scheduling scheme can be obtained within limited time through four low-level heuristic operations based on disturbance; and meanwhile, a distributed parallel evolution thought is combined, a plurality of optimizers operate at the same time, and an optimal sorting scheme is ensured to be obtained in a short time by the method in a manner of periodically synchronizing an optimal solution.

Description

Technical Field

[0001] The present invention belongs to the technical field of blockchain privacy protection and supervision, and particularly relates to a flight arrival and landing scheduling method based on a hyper-heuristic algorithm. Background Art

[0002] The flight arrival and landing scheduling problem is a core issue in air traffic management. Especially in a busy airport environment, how to efficiently and safely arrange the landing sequence and time of aircraft is directly related to the operation efficiency of the airport, the flight punctuality rate, and aviation safety. With the continuous growth of global air transportation volume, the runway resources and air traffic control capabilities of airports are facing increasing pressure.

[0003] Traditional flight scheduling methods mainly adopt the first-come, first-served strategy, that is, arranging landings according to the order in which aircraft arrive at the airport. However, this strategy has obvious limitations in practical applications. Since different types of aircraft need to maintain different minimum safe separation times during landing, the first-come, first-served strategy often fails to make full use of runway resources, resulting in an increase in aircraft waiting time and a decrease in airport operation efficiency. For example, the difference in the minimum safe separation time between heavy aircraft and light aircraft is relatively large. If the landing sequence is arranged according to the first-come, first-served strategy, it may lead to unnecessary delays.

[0004] To overcome the limitations of the first-come, first-served strategy, researchers have proposed various optimization algorithms to solve the flight scheduling problem. Early research mainly focused on mathematical models and exact algorithms, such as dynamic programming, branch and bound method, etc. Although these methods can find the optimal solution, as the problem scale increases, the computational complexity rises sharply, making it difficult to apply in practice. Therefore, researchers have gradually turned to heuristic algorithms and meta-heuristic algorithms to find approximate optimal solutions within a reasonable time.

[0005] Heuristic algorithms can quickly find better solutions by using specific knowledge in the problem domain, but it is often difficult to guarantee the global optimum. Meta-heuristic algorithms, on the other hand, can search for better solutions in a larger solution space by combining multiple heuristic strategies or introducing a random search mechanism. For example, meta-heuristic algorithms such as genetic algorithms, ant colony algorithms, and simulated annealing algorithms have been widely used in this problem. These algorithms can find better solutions in a relatively short time by simulating natural evolution, swarm intelligence, etc., but there are still problems such as being easily trapped in local optima and slow convergence speed.

[0006] In recent years, hyper-heuristic algorithms have gradually become an effective means to solve complex optimization problems. Instead of directly operating on the solution space of the problem, hyper-heuristic algorithms manage a set of low-level heuristics and adaptively select appropriate strategies to guide the search process. This method can make full use of the advantages of different heuristic algorithms, avoid the limitations of a single heuristic algorithm, and thus explore and develop more effectively in the solution space. Hyper-heuristic algorithms have achieved remarkable results in many fields such as the bin-packing problem, scheduling problem, vehicle routing problem, etc., but their application in the flight landing scheduling problem is still in the exploratory stage.

[0007] Although existing methods have made some progress in the problem, there are still some challenges. First, when dealing with large-scale problems, existing methods often struggle to balance computational speed and solution quality. Second, due to the dynamic and complex nature of the flight scheduling problem, the robustness and adaptability of algorithms in practical applications still need to be further improved. Summary of the Invention

[0008] The main objective of the present invention is to overcome the drawbacks and deficiencies of the prior art and provide a flight arrival and landing scheduling method based on hyper-heuristic algorithms. This method uses the idea of hyper-heuristic methods and distributed parallel evolution, significantly improving the solution speed and quality of the flight arrival and landing scheduling method. For new flight arrival and landing scheduling scenarios, this method can quickly generate efficient scheduling plans, achieving a "real-time" effect.

[0009] To achieve the above objective, the present invention adopts the following technical solutions:

[0010] A flight arrival and landing scheduling method based on hyper-heuristic algorithms, wherein the flight arrival and landing scheduling method comprises the following steps:

[0011] S1. Construct an initial landing scheduling plan. First, according to the estimated arrival time of the flight, use the first-come-first-served strategy to generate an initial flight landing scheduling plan. The process is as follows:

[0012] S101. Read flight data, including information such as flight number, aircraft type, estimated arrival time, etc. For example, if multiple flights arrive in sequence, each flight has three pieces of information (flight number, aircraft type, estimated arrival time). Encode one of the flights as (1, 1, 0), indicating that the flight number is 1, the aircraft type is 1, and the estimated arrival time is the 0th second;

[0013] S102. Sort according to the expected arrival time PLT of the flights to generate an initial landing sequence. Suppose there are 5 flights with codes (1, 1, 0), (2, 1, 80), (3, 2, 120), (4, 2, 200), and (5, 1, 300) respectively. Then the initial landing sequence is (1, 2, 3, 4, 5), indicating that each flight lands in the order of the expected arrival time.

[0014] S103. According to the shortest interval time between the landings of the two aircraft types, obtain the actual landing time ALT of all flights. The sum of the differences between the actual landing time ALT and the expected arrival time PLT of all flights can be used to calculate the total delay time of the sequence, which serves as the benchmark for subsequent optimization.

[0015] For example, after a flight of aircraft type 1 lands, it is necessary to wait 100 seconds for the next flight of aircraft type 1 to land and 120 seconds for a flight of aircraft type 2 to land; after a flight of aircraft type 2 lands, it is necessary to wait 80 seconds for the next flight of aircraft type 1 to land and 100 seconds for a flight of aircraft type 2 to land. Then the actual landing time ALT of the initial landing sequence assumed in S102 is (0, 100, 220, 320, 400), which is obtained through Equation 1:

[0016]

[0017] In the above formula, ALT i is the actual landing time of the i-th flight, PLT i is the expected arrival time of the i-th flight, LTI(x, y) is the minimum interval time between the previously landed flight x and the subsequently landed flight y, and Ti represents the type of the flight. The total delay TAD is the sum of the differences between the actual landing time ALT and the expected arrival time PLT of each flight, that is:

[0018]

[0019] where N represents the total number of flights to be calculated, and i represents the flight number;

[0020] S2. In order to perform diverse searches in the solution space and find the landing sequence with the shortest total delay time, the present invention designs four perturbation-based low-level heuristic operations. The functions of these four low-level heuristics are only to adjust the order of flight landings. The expected arrival time PLT of the flights will not change accordingly, but the actual landing time ALT needs to be recalculated due to the change in the flight landing order. The specific operations are as follows:

[0021] S201. Random insertion: Randomly select a flight F from the current flight landing schedule sequence and randomly insert it into a position whose distance from the position of F is no more than k1. This operation can perform a search within a local range through a small perturbation, avoiding a large fluctuation in the quality of the solution.

[0022] S202. Random swap: Randomly select a flight F1 from the current flight landing schedule sequence, then select another flight F2 from positions whose distance from the position of F1 is no more than k2, and finally swap the positions of F1 and F2. This operation can perform a diverse search within a local range by swapping the positions of flights.

[0023] S203. Neighbor swap: Randomly select a flight F1 from the current flight landing schedule sequence and swap it with the flight in the next position. This operation can perform a small adjustment within a local range by swapping adjacent flights.

[0024] S204. Destroy and reconstruct: Randomly select a flight F1 from the current flight landing schedule sequence and re - randomly arrange k3 flights including F1 around it. This operation can perform a search within a global range through a large perturbation, avoiding the algorithm falling into a local optimum.

[0025] S3. Randomly call a heuristic to operate on the solution. In each iteration step, randomly select a low - level heuristic operation with equal probability to perturb the current flight landing schedule solution, generate a new solution, and calculate the total delay time (TAD) of the new solution.

[0026] S4. Update the current optimal solution. Compare the total delay time TAD of the new solution with that of the current solution. If the total delay time of the new solution is less than that of the current solution, then take the new solution as the current optimal solution; otherwise, retain the current solution. Then continue to iteratively execute S2 - S4 until the established number of iterations fes is reached to end the iteration, and output the landing schedule solution with the minimum total delay time TAD; or when the number of iterations of collaborative search is reached, execute S5.

[0027] S5. To improve the solution speed and quality of the algorithm, the present invention adopts a distributed parallel evolution strategy. Each program that independently completes steps S2 - S4 is called a hyper - heuristic optimizer. Multiple hyper - heuristic optimizers run in parallel. Every certain number of iterations, synchronize the current optimal solution to all optimizers, which can optimize the search efficiency. Finally, when all optimizers reach the set number of iterations fes, select the one with the shortest total delay time among all optimizers as the optimal flight schedule solution.

[0028] Further, in step S2, k1, k2, and k3 can all be adjusted according to the problem scale. It is recommended that k1 and k2 be set to [1, 5], and k3 be set to [3, 5]. This range is obtained from repeated experiments. Under this range, the optimizer can make full use of computing resources and search for more possible landing sequences in a short time.

[0029] Further, by adjusting the number n of hyper-heuristic optimizers and the number of iterations fes of each optimizer, a trade-off can be made between the solution quality and the computing time. According to the problem scale and computing resources, the number of hyper-heuristic optimizers can be varied. The number of hyper-heuristic optimizers and the number of iterations of each hyper-heuristic optimizer need to increase with the increase of the problem scale. It is recommended that n be set to [4, 8], and fes be set to [500, 2000]. This range is obtained from repeated experiments. Under this range, it can be ensured to a great extent that the optimal solution can be obtained in a short time under normal circumstances.

[0030] Further, by adjusting the number of generations of optimal solution synchronization, the search efficiency of the algorithm can be affected. Since the results generated by each optimizer in each iteration are generally different, this strategy can make full use of all optimizers within a certain period of time to explore a larger solution space and find the optimal solution among them. It should be noted that if the interval of generations is too small, the diversity of solutions will be greatly reduced, and it is easier for all optimizers to fall into the local optimal solution at the same time; while if the interval of generations is too large, some optimizers may fall into the local optimal solution for a long time, resulting in reduced efficiency. It is recommended that the number of generations of optimal solution synchronization be set to [50, 200]. This range is obtained from experiments and can greatly improve the search efficiency of the algorithm.

[0031] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0032] (1) Through the hyper-heuristic algorithm, the present invention can adaptively select a variety of low-level heuristic operations (such as random insertion, random exchange, neighbor exchange, destruction and reconstruction, etc.), so as to perform diverse searches in the solution space. This mechanism can not only effectively prevent the algorithm from falling into the local optimal solution, but also reduce the number of evaluations of solutions, significantly accelerating the convergence speed. Compared with the traditional single heuristic algorithm, the present invention can find a better solution in a shorter time, greatly improving the solution efficiency.

[0033] (2) The present invention adopts the idea of distribution and the method of parallel evolution. By running multiple hyper-heuristic optimizers in parallel and synchronizing the solutions at regular intervals of iteration, it ensures that the algorithm searches for the optimal solution globally. This strategy of distributed parallel evolution can not only make full use of computing resources but also significantly improve the robustness and stability of the algorithm. Even if a certain optimizer falls into a local optimum, other optimizers can still continue to search globally, thereby greatly improving the solution quality. In addition, the method of distributed parallel evolution can explore more solution spaces within the same time, further accelerating the convergence speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0035] Figure 1 is a flowchart of the flight arrival and landing scheduling method based on the hyper-heuristic algorithm in the embodiment of the present invention;

[0036] Figure 2 is an effect diagram of the flight arrival and landing scheduling method based on the hyper-heuristic algorithm in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] In order to enable those skilled in the art of the present technology to better understand the solutions of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of protection of the present application.

[0038] Referring to "embodiments" in the present application means that the specific features, structures, or characteristics described in connection with the embodiments may be included in at least one embodiment of the present application. The phrase appears at various positions in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments.

[0039] Embodiment 1

[0040] Refer to Figure 1。This embodiment discloses a flight arrival and landing scheduling method based on a hyper-heuristic algorithm. In this scenario, the airport flights are relatively busy. According to the minimum time interval condition for flight landings shown in Table 1, if the traditional strategy of landing in the order of arrival time is adopted, it will lead to significant flight delays. The specific flight data is shown in Table 2.

[0041] Table 1. Minimum Time Interval Table between Two Flights (Unit / Second)

[0042]

[0043] In Table 1, Aircraft Type 1, Aircraft Type 2, Aircraft Type 3, and Aircraft Type 4 represent Boeing-747, Boeing-727, Boeing-707, and Douglas DC-9 respectively.

[0044] Table 2. Flight Data Table in Embodiment 1

[0045] Flight Number Flight Model Estimated Arrival Time (seconds) 1 1 0 2 1 79 3 1 144 4 2 204 5 1 264 6 1 320 7 2 528 8 1 635 9 2 730 10 2 766 11 1 790 12 1 920

[0046] The operation steps of the flight arrival and landing scheduling method are as follows: S1. Construct an initial landing scheduling plan and initialize the program. For Figure 2 the flight arrival sequence in Embodiment 2, code each flight according to the arrival time order, and the landing sequence is (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20). According to Figure 2 the minimum interval time of different aircraft types, calculate the actual landing time of the flights in the current order, the delay time of each flight, and the total delay time of all flights. The total delay time of the sequential landing sequence in Example 2 is 4578 seconds, which will be used as the subsequent optimization benchmark. After the calculation is completed, create 4 hyper-heuristic optimizers, set the number of iterations of each optimizer to 1000 generations, set the synchronization algebra of the optimal solution to 100 generations, input the initial landing scheduling plan into each optimizer, and start running them distributively, and execute the subsequent steps S2 - S5.

[0048] S2. Adjust the parameters of the low-level heuristic operations according to the scale of the problem. Since the sequence of 12 flights can be regarded as a small-scale problem, the parameters k1 and k2 of random exchange and random insertion are set to 3, and the parameter k3 of destruction and reconstruction is set to 5.

[0049] S3. Use the program to generate random numbers, and with equal probability, select one of the 4 low-level heuristics to perturb the current flight landing scheduling sequence, obtain a new landing sequence, and recalculate the actual landing time and total delay time after the sequence perturbation.

[0050] S4. Compare the total delay time of the new landing sequence with that of the previous generation. If the total delay time of the new landing sequence is shorter, select this sequence as the current sequence of this optimizer and proceed to the next step.

[0051] S5. Since the synchronization algebra of the optimal solution is set to 100, the optimizer needs to perform synchronization every 100 generations. During this process, compare the current sequences of each optimizer, select the sequence with the shortest total delay time as the global optimal solution, and replace the current sequences of all optimizers. Subsequently, each optimizer continues to execute steps S3 - S5 based on this global optimal solution. If all optimizers reach the set number of iterations, the program will directly output the sequence with the shortest total delay time among all optimizers as the final landing scheduling plan.

[0052] The recorded data, after going through steps S1 - S5, the output landing plan is (1, 2, 3, 5, 6, 7, 4, 9, 10, 8, 11, 12), the time used is 45 milliseconds, and the total delay time is 1372 seconds. Compared with the traditional first - come - first - served method (total delay time is 2524 seconds), the optimization effect of the present invention is significant.

[0053] Example 2

[0054] See Figure 1 This example describes a flight arrival and landing scheduling method based on a hyper - heuristic algorithm. In this scenario, the airport flights are relatively busy. According to the minimum time interval conditions for flight landings shown in Table 1, if the traditional strategy of landing in the order of arrival time is adopted, it will result in significant flight delays. The specific flight data is shown in Table 3,

[0055] Table 3. Flight data table in Example 2

[0056] Flight Number Flight Model Estimated Arrival Time (seconds) 1 4 0 2 3 107 3 1 272 4 1 293 5 2 327 6 3 365 7 1 459 8 1 530 9 3 630 10 4 844 11 2 880 12 4 918 13 1 945 14 4 1244 15 2 1373 16 3 1379 17 3 1691 18 2 1779 19 1 1900 20 2 1945

[0057] The operating steps of the flight arrival and landing scheduling method are as follows: S1. Construct an initial landing scheduling plan and initialize the program. For Figure 2 the flight arrival sequence in Example 2, encode each flight in the order of arrival time, and the landing sequence is (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20). According to Figure 2The minimum interval time for different aircraft types calculates the actual landing time of flights in the current order, the delay time of each flight, and the total delay time of all flights. The total delay time of the landing sequence in Example 2 is 4578 seconds, which will be used as the subsequent optimization benchmark. After the calculation is completed, 4 hyper-heuristic optimizers are created. The number of iterations for each optimizer is set to 1000 generations, and the synchronization algebra of the optimal solution is set to 100 generations. The initial landing scheduling plan is input into each optimizer, and they start running distributively, and then perform the subsequent steps S2 - S5.

[0059] S2. Adjust the parameters of the low-level heuristic operations according to the scale of the problem. Set the parameters k1 and k2 of random exchange and random insertion to 5, and the parameter k3 of destruction and reconstruction to 5.

[0060] S3. Use the program to generate random numbers, and select one of the 4 low-level heuristics with equal probability to perturb the current flight landing scheduling sequence to obtain a new landing sequence, and recalculate the actual landing time and total delay time after the sequence perturbation.

[0061] S4. Compare the total delay time of the new landing sequence with that of the previous generation. If the total delay time of the new landing sequence is shorter, then select this sequence as the current sequence of this optimizer and proceed to the next step.

[0062] S5. Since the synchronization algebra of the optimal solution is set to 100, the optimizers need to be synchronized every 100 generations. During this process, compare the current sequences of each optimizer, select the sequence with the shortest total delay time as the global optimal solution, and replace the current sequences of all optimizers. Subsequently, each optimizer continues to execute steps S3 - S5 based on this global optimal solution. If all optimizers reach the set number of iterations, the program will directly output the sequence with the shortest total delay time among all optimizers as the final landing scheduling plan.

[0063] The recorded data, after going through steps S1 - S5, the output landing plan is (1, 2, 3, 4, 7, 8, 9, 6, 5, 11, 10, 12, 14, 15, 16, 13, 17, 18, 20, 19), the time used is 69 milliseconds, and the total delay time is 2702 seconds. Compared with the traditional first-come, first-served method (the total delay time is 4578 seconds), the optimization effect of the present invention is significant.

[0064] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as these combinations of technical features do not conflict, they should all be considered to be within the scope described in this specification.

[0065] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications made without departing from the spirit and principle of the present invention shall be equivalent substitution methods and are all included in the protection scope of the present invention.

Claims

1. A flight arrival and landing scheduling method based on a hyper-heuristic algorithm, characterized in that, The flight arrival and landing scheduling method includes the following steps: S1. Construct an initial landing scheduling plan. First, according to the expected arrival time of the flight, use the first-come, first-served strategy to generate an initial flight landing scheduling plan. The process is as follows: S101. Read flight data, including information such as flight number, aircraft type, expected arrival time, etc.; S102. Sort according to the expected arrival time PLT of the flight to generate an initial landing sequence; S103. According to the shortest interval time for the landing of heavy aircraft and light aircraft, obtain the actual landing time ALT of all flights. The sum of the differences between the actual landing time ALT and the expected arrival time PLT of all flights can be used to calculate the total delay time of the sequence, which serves as the benchmark for subsequent optimization. ALT is calculated by the following formula: Among them, ALT i is the actual landing time of the i-th flight, PLT i is the estimated arrival time of the i-th flight, LTI(x, y) is the minimum interval time between the previously landed flight x and the subsequently landed flight y, T i represents the model of the i-th flight. The total delay TAD is the sum of the difference between the actual landing time ALT and the estimated arrival time PLT for each flight, that is: where N represents the total number of flights to be calculated, and i represents the flight number; S2. Find the landing sequence with the shortest total delay time through the following 4 disturbance-based low-level heuristic operations. The process is as follows: S201. Random insertion: Randomly select a flight F from the current flight landing scheduling sequence and randomly insert it into a position whose distance from the position of F does not exceed k1; S202. Random swap: Randomly select a flight F1 from the current flight landing scheduling sequence, then select another flight F2 from the positions whose distance from the position of F1 does not exceed k2, and finally swap the positions of F1 and F2; S203. Neighbor swap: Randomly select a flight F1 from the current flight landing scheduling sequence and swap this flight with the flight in the next position; S204. Destroy and reconstruct: Randomly select a flight F1 from the current flight landing scheduling sequence and randomly rearrange k3 flights including F1 around it; S3. Randomly call a heuristic to operate on the plan. In each iteration step, randomly select a low-level heuristic operation with equal probability to perturb the current flight landing scheduling plan to generate a new solution, and calculate the total delay time TAD of the new solution; S4. Update the current optimal solution: Compare the total delay time TAD of the new solution and the current solution. If the total delay time of the new solution is less than the total delay time of the current solution, then use the new solution as the current optimal solution. Otherwise, retain the current solution, and then continue to iterate and execute S2 - S4 until the established number of iterations fes is reached to end the iteration, and output the landing scheduling plan with the smallest total delay time TAD; or when the number of iterations of collaborative search is reached, execute S5; S5. Adopt a distributed parallel evolution strategy. Each independent completion of steps S2 - S4 is called a hyper-heuristic optimizer. Multiple hyper-heuristic optimizers run in parallel. Every certain number of iterations, synchronize the current optimal solution to all optimizers. Finally, when all optimizers reach the set number of iterations fes, extract the one with the shortest total delay time among all optimizers as the optimal flight scheduling plan.

2. The flight arrival and landing scheduling method based on the hyper-heuristic algorithm according to claim 1, characterized in that In the above step S2, the value ranges of the parameters k1 and k2 for low-level heuristic random swap and random insertion are [1, 5], and the value range of the parameter k3 for destroy and reconstruct is [3, 10].

3. The flight arrival and landing scheduling method based on the hyper-heuristic algorithm according to claim 1, characterized in that, In the step S4, the value range of the number of iterations of each hyper-heuristic optimizer is [500, 2000].

4. The flight arrival and landing scheduling method based on the hyper-heuristic algorithm according to claim 1, characterized in that In the step S5, the value range of the number n of hyper-heuristic optimizers is [4, 8], and the interval algebra for synchronizing the optimal solutions is set to 100.

Images