Improved shipboard aircraft departure efficient scheduling optimization method

Abstract

The invention discloses an improved shipboard aircraft departure and departure efficient scheduling optimization method, and belongs to the field of shipboard aircraft scheduling optimization, and the method comprises the steps: constructing a shipboard aircraft departure and departure scheduling model which is a workshop scheduling model with multiple constraints; and solving the scheduling model by adopting an improved meta-heuristic recurrent neural algorithm, representing a solution by adopting a job allocation matrix, calculating a new solution through a disturbance update strategy, and carrying out iterative optimization to obtain an optimal scheduling scheme. According to the method, the problems of low solving efficiency, premature convergence, difficulty in ensuring that all constraints are satisfied and the like when the existing method faces specific high-dimension and strong-coupling constraints of shipboard aircraft scheduling are solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of carrier-based aircraft dispatch optimization, and particularly relates to an improved efficient dispatch optimization method for carrier-based aircraft takeoff. BACKGROUND

[0002] Carrier-based aircraft is a key component of a carrier battle group, and undertakes and completes most combat missions. The rapid dispatch and landing scheduling of a carrier-based aircraft group is a key problem in the scheduling of carrier-based aircraft operations. Maintaining the efficiency of carrier-based aircraft dispatch is of great significance to improving the overall combat effectiveness of a carrier. The carrier-based aircraft dispatch problem usually involves the dispatch of multiple carrier-based aircraft, and needs to consider hard constraints such as takeoff position allocation, runway allocation, takeoff sequence arrangement, machine resource scheduling, takeoff time interval constraints, and spatial anti-collision safety constraints. While considering all the constraints, the given fleet aircraft group dispatch time is minimized, and the number of carrier-based aircraft dispatched in a given time is maximized.

[0003] Traditional scheduling arrangements use simple and direct methods such as manual placement and manual experience scheduling, and excessively rely on existing experience. The scheduling efficiency of manual decision-making is not high, and it is difficult to apply to the dispatch of a large number of carrier-based aircraft. With the development of computer technology, intelligent decision-making and optimization scheduling to some extent make up for the defects of the past methods that are difficult to arrange the entire dispatch of a carrier-based aircraft group in a limited time. At present, there are many achievements for carrier-based aircraft dispatch, such as inverse reinforcement learning, mixed integer programming, deck operations planning system DCAP, path planning and tracking control, integrated optimization of carrier-based aircraft deck maintenance service scheduling and resource allocation, human-computer interaction intelligent decision-making, and carrier-based aircraft support scheduling based on genetic algorithm.

[0004] In addition, metaheuristic algorithms are also commonly used for carrier-based aircraft dispatch optimization. Metaheuristic algorithms are a class of algorithms used to solve complex optimization problems, aiming to gradually approach the optimal solution in the search space. Common metaheuristic algorithms include genetic algorithms, particle swarm optimization, differential evolution algorithms, and ant colony algorithms. However, existing intelligent methods often have low solution efficiency, premature convergence, and difficulty in ensuring that all constraints are met when facing the high dimensionality and strong coupling constraints specific to carrier-based aircraft scheduling. SUMMARY

[0005] In view of the above deficiencies in the prior art, the improved efficient dispatch optimization method for carrier-based aircraft takeoff provided by the present application solves the problems of low solution efficiency, premature convergence, and difficulty in ensuring that all constraints are met when existing methods face the high dimensionality and strong coupling constraints specific to carrier-based aircraft scheduling.

[0006] In order to achieve the above application purposes, the technical scheme adopted by the present application is as follows: an improved efficient dispatch optimization method for carrier-based aircraft takeoff, comprising the following steps:

[0007] S1: constructing a carrier-based aircraft dispatching off-deck scheduling model, the scheduling model being a job shop scheduling model with multiple constraints;

[0008] S2: solving the scheduling model by using an improved meta-heuristic recursive neural algorithm, obtaining an optimal scheduling scheme by using a job assignment matrix to represent a solution, calculating a new solution by using a perturbation update strategy and iteratively optimizing.

[0009] Further, the objective function of the job shop scheduling model in S1 is:

[0010]

[0011] wherein, denotes the total time for all aircraft to complete all processes, denotes the end time of the aircraft, denotes the end time of the machine, denotes the end time of the first process stage of the aircraft, denotes the end time of the second process stage of the aircraft, and denote the number of aircraft and processes, respectively;

[0012] The constraints of the job shop scheduling model include:

[0013] Process exclusive constraint:

[0014]

[0015] wherein, denotes the set of processing machines of the process, denotes that the first process of the aircraft is processed by the machine, denotes that the first process of the aircraft is not processed by the machine, denotes that the second process of the aircraft is not processed by the machine, denotes an arbitrary symbol; Machine exclusive constraint: Time constraint:

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] ​​​​​​

[0023] wherein, and respectively represent carrier-based aircrafts the start time of the machine of the machine of the start time and processing time of the machine, represents, represents the machine number in the previous process, represents the end time of the previous process, respectively represent carrier-based aircrafts the start time of the machine of the machine of the processing end time of the machine, respectively represent carrier-based aircrafts the processing end time of the machine of the machine of the processing end time of the machine, and represent different carrier-based aircraft numbers, represent the processing sequence of carrier-based aircrafts of the same machine in the same process;

[0024] Spatial anti-collision safety constraints:

[0025]

[0026] wherein, represents the set of processing machines of the 4th process, respectively represent carrier-based aircrafts of the 4th process in the machine .

[0027] Further, the S2 includes the following sub-steps:

[0028] S21: initialize the job assignment matrix, and the element value of the job assignment matrix encodes the priority sequence of carrier-based aircrafts in each process;

[0029] S22: calculate the perturbation matrix and perform normalization processing;

[0030] S23: generate multiple perturbation candidate solutions based on the perturbation matrix;

[0031] S24: convert the perturbation candidate solutions into feasible scheduling schemes through decoding scheduling process, and calculate the overall scheduling time of the carrier-based aircraft group corresponding to each feasible scheduling scheme;

[0032] S25: compare the scheduling time of each feasible scheduling scheme, select a better scheduling scheme to update the current solution, and maintain the global optimal solution;

[0033] S26: determine whether the maximum number of iterations is reached, if not, return to S21, if yes, output the optimal scheduling scheme.

[0034] Further, the job assignment matrix in S21 is:

[0035]

[0036] wherein, denotes the job assignment matrix, denotes the process of carrier-based aircraft launch priority matrix, denotes the carrier-based aircraft launch priority in the process , denotes the transpose of the matrix, is the current iteration number, denotes the real field.

[0037] Further, the disturbance matrix in S22 is:

[0038]

[0039]

[0040] wherein, denotes the disturbance matrix, is a standard normal random matrix, denotes the Frobenius norm.

[0041] Further, the disturbance candidate solution in S23 is:

[0042]

[0043]

[0044]

[0045] wherein, , and denote three disturbance candidate solutions, denotes the job assignment matrix of the th iteration, d denotes the neighborhood search step size, is a set of setting parameters.

[0046] Further, the decoding scheduling process in S24 includes the following steps:

[0047] S241: For each process , generate a carrier-based aircraft scheduling sequence based on the job assignment matrix:

[0048]

[0049] wherein, denotes the dispatch sequence of the aircrafts;

[0050] S242: initialize the earliest time vector of the machines and the earliest startable time vector of the aircraft group :

[0051]

[0052]

[0053] wherein K denotes the total number of machines;

[0054] S243: traverse the aircrafts in sequence and perform the operations of machine selection, determination of the take-off start time, determination of the take-off end time in turn, and update the earliest time vector of the machines and the earliest startable time vector of the aircraft group ;

[0055] machine selection is denoted as:

[0056]

[0057] take-off start time is:

[0058]

[0059] take-off end time is:

[0060]

[0061] wherein, denotes the aircraft the time consumption of the process ;

[0062] update the earliest time vector of the machines and the earliest startable time vector of the aircraft group is:

[0063]

[0064]

[0065] S244: record all dispatch tasks and generate a dispatch scheme, wherein the dispatch task contains the aircraft number, the machine number, the process number, the start time and the end time;

[0066] dispatch task is:

[0067]

[0068] Generate scheduling scheme For:

[0069] .

[0070] Further, the S25 in the selection of a more optimal scheduling scheme to update the current solution, and maintain the global optimal solution, the formula is:

[0071]

[0072]

[0073]

[0074] Wherein, The updated job allocation matrix after the first iteration, The scheduling decoding function, The one out of the scheduling scheme set, The scheduling scheme set, the function returns the maximum completion time as the end of the aircraft in the current process and machine processing time ;

[0075] Maintain the global optimal solution:

[0076]

[0077] Wherein, The global optimal solution in the iteration process.

[0078] The beneficial effects of the present application are: the present application proposes an improved meta-heuristic recursive neural algorithm for the dispatching of aircraft carrier aircrafts. The aircraft carrier aircraft is taken as the research object, and the dispatching of the aircraft carrier aircraft is analyzed. A multi-constraint job shop scheduling model is established, and complex constraints such as process timing, resource exclusivity, time window and space collision avoidance are systematically integrated to optimize the overall aircraft carrier aircraft dispatching time. A low-dimensional job allocation matrix solution representation mechanism is designed, and a random disturbance strategy is combined to effectively overcome the premature convergence problem of traditional intelligent algorithms in high-dimensional strong constraint space. The framework realizes accurate mapping from the solution space to the feasible scheduling scheme through the constraint satisfaction decoder, and provides a new solution paradigm for complex scheduling problems. The present application is a powerful supplement to the research on the rapid dispatching of aircraft carrier aircrafts, and to some extent, it makes up for the shortcomings of existing research on aircraft carrier aircraft scheduling, provides a new scheduling solution method, solves the problems of low efficiency of traditional methods and insufficient solution of existing intelligent algorithms under complex constraints, and provides a new theoretical method and practical tool for efficient dispatching of aircraft carrier aircrafts. BRIEF DESCRIPTION OF DRAWINGS

[0079] Figure 1 The flowchart of the improved aircraft carrier aircraft dispatching optimization method of the present application.

[0080] Figure 2 The overall solution flowchart of the dispatching optimization.

[0081] Figure 3 The Gantt chart of 8 aircraft carrier aircraft dispatching-aircraft operation scheduling.

[0082] Figure 4 The Gantt chart of 8 aircraft carrier aircraft dispatching-machine operation scheduling.

[0083] Figure 5 The Gantt chart of 12 aircraft carrier aircraft dispatching-aircraft operation scheduling.

[0084] Figure 6 The Gantt chart of 12 aircraft carrier aircraft dispatching-machine operation scheduling.

[0085] Figure 7 The Gantt chart of 18 aircraft carrier aircraft dispatching-aircraft operation scheduling.

[0086] Figure 8 The Gantt chart of 18 aircraft carrier aircraft dispatching-machine operation scheduling.

[0087] Figure 9 The algorithm iteration convergence curve graph under different dispatching scales.

[0088] Figure 10 The algorithm iteration convergence curve graph under different aircraft carrier aircraft dispatching scales.

[0089] Figure 11 This is the iterative convergence curve of the PSO algorithm under different carrier-based aircraft deployment scales.

[0090] Figure 12 Iterative convergence curve of DE algorithm under different aircraft deployment scales. DETAILED DESCRIPTION

[0091] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0092] Aircraft sortie dispatching and departure operations are a multi-step process. Each aircraft sortie must complete multiple steps in sequence, each of which must be performed on specific deck machinery and subject to a series of constraints. This process primarily includes the following: the preparation phase, during which aircraft are moved from their parking positions to the preparation area and inspected, relying on machinery resources such as towing equipment and taxiways; the loading and refueling phase, during which aircraft complete pre-combat preparations such as weapon loading and fuel replenishment, relying on specialized support equipment related to loading and refueling; the takeoff readiness phase, during which aircraft enter their designated takeoff positions on the deck and complete takeoff commissioning, relying on machinery resources related to takeoff inspections; and the departure and takeoff phase, during which aircraft are launched or taxied, relying on resources such as the deck takeoff runway. Aircraft sortie dispatching and departure operations are a typical multi-constraint shop floor scheduling optimization problem. Aircraft sortie efficiency directly impacts the overall combat effectiveness of an aircraft carrier. Solving this scheduling optimization problem and rapidly determining a carrier aircraft departure scheduling solution while satisfying all complex constraints is crucial for ensuring a robust aircraft carrier's combat capability. Based on the above description of the carrier-based aircraft dispatch and departure scheduling problem, the present invention mainly performs the following steps.

[0093] like Figure 1 As shown, an improved carrier-based aircraft dispatch and departure efficient scheduling optimization method includes the following steps:

[0094] S1: Constructing a carrier-based aircraft dispatch and departure scheduling model, which is a shop-floor scheduling model with multiple constraints;

[0095] The goal of carrier-based aircraft dispatch scheduling is to minimize the total time for all carrier-based aircraft to complete all processes, which is a shop-floor scheduling problem with multiple constraints.

[0096] The objective function of the workshop scheduling model in S1 is:

[0097]

[0098] in, Indicates the total time it takes for all carrier-based aircraft to complete all processes. Indicates carrier-based aircraft In the machine Complete the The end time of the process stage, and Represent the number of carrier-based aircraft and processes respectively;

[0099] The constraints of the workshop scheduling model include:

[0100] Any carrier-based aircraft and process can only be executed on one machine at a time. The process exclusive constraint is expressed as:

[0101]

[0102] in, Indicates the process The processing machine set, Indicates carrier-based aircraft No. The process in the machine processing, Indicates carrier-based aircraft No. The process is not on the machine processing, Represents any symbol;

[0103] Each machine in any process can only process at most one carrier-based aircraft at a time. The cumulative number of all carrier-based aircraft processed by the same machine in the same process is 1. The machine exclusive constraint is expressed as:

[0104]

[0105] The processing time constraint of the carrier-based aircraft on the machine satisfies that the end time is not earlier than the start time processing time. The time constraint is expressed as:

[0106]

[0107] The previous and next processes of the same carrier-based aircraft shall be completed in sequence, and each carrier-based aircraft must meet the starting time no earlier than the ending time of the previous process:

[0108]

[0109] Priority constraints are set for the processing operations of the preceding and following aircraft on the same machine to ensure that the end time of the preceding processing operation is not earlier than the start time of the following processing operation:

[0110]

[0111]

[0112] in, and Represents carrier-based aircraft In process machine Start time and processing time, express, denotes the machine number in the previous process, denotes the end time of the previous process, denotes the carrier aircraft in the process machine , the start time, denotes the carrier aircraft in the process machine , the processing end time, and denotes the different carrier aircraft numbers, denotes the carrier aircraft processing sequence of the same machine in the same process;

[0113] The spatial anti-collision safety constraint requirement of the take-off and launch process, while allowing at most one carrier aircraft to take off and launch, is expressed as:

[0114]

[0115] wherein, denotes the set of processing machines of the 4th process, denotes the carrier aircraft of the 4th process in the machine processing.

[0116] S2: An improved meta-heuristic recursive neural algorithm is used to solve the scheduling model, a job assignment matrix is used to represent the solution, a new solution is calculated by a perturbation update strategy and iteratively optimized to obtain an optimal scheduling scheme.

[0117] The improved meta-heuristic recursive neural algorithm is used to solve the above-mentioned integer optimization problem, the core is to realize constraint conversion through the iteration process of random perturbation direction search and decoding scheduling, and the specific process is as shown in Figure 2 .

[0118] The iteration process is the core part of the take-off scheduling optimization, and mainly functions in exploring better solutions, calculating fitness values, and updating global optimal solutions. The present application takes the four processes of preparation, mounting, readiness and departure of carrier aircraft take-off scheduling as an optimization example, uses a job assignment matrix as a mathematical representation of the solution, and takes the overall carrier aircraft take-off scheduling time as the fitness value. The element values of the job assignment matrix encode the priority sequence of the carrier aircraft in each process, wherein the size of the row vector element value directly determines the take-off order of the carrier aircraft in the process, and the smaller the value, the higher the priority. Through the decoding scheduling process, the job assignment matrix can be converted into a carrier aircraft take-off scheduling scheme.

[0119] The S2 includes the following steps:

[0120] S21: initialize a job assignment matrix, an element value of the job assignment matrix encoding a priority sequence of the carrier-based aircraft at each process;

[0121] The job assignment matrix is:

[0122]

[0123] wherein, denotes the job assignment matrix, denotes a process of the carrier-based aircraft sortie priority matrix, denotes the carrier-based aircraft sortie priority at the process , denotes the transpose of the matrix, is the current iteration number, denotes the real field.

[0124] S22: calculate a perturbation matrix and perform normalization processing;

[0125] The perturbation matrix is:

[0126]

[0127]

[0128] wherein, denotes the perturbation matrix, is a standard normal random matrix, denotes the Frobenius norm.

[0129] S23: generate a plurality of perturbation candidate solutions based on the perturbation matrix;

[0130] The perturbation candidate solution is:

[0131]

[0132]

[0133]

[0134] wherein, , and denote three perturbation candidate solutions, denotes the job assignment matrix of the iteration, d denotes a neighborhood search step size, and controls the perturbation amplitude, is a set of setting parameters.

[0135] S24: Transform the perturbation candidate into a feasible scheduling scheme by decoding the scheduling process, and calculate the overall scheduling time of the aircraft group corresponding to each feasible scheduling scheme;

[0136] The aircraft dispatch scheduling decoding process outputs a scheduling scheme containing multiple task information and meeting all the above constraints according to the input job allocation matrix, aircraft time consumption matrix of each process stage, and optional machine sequence. Each task information contains the aircraft number, machine number, process number, current task start time, and current task end time of the current task.

[0137] The decoding scheduling process includes the following steps:

[0138] The job allocation matrix contains the aircraft dispatch sequence information of each process. The selection of machines is based on the process number sequence from small to large to select the aircraft dispatch number sequence of each process. The aircraft dispatch number is determined by the sequence corresponding to the job allocation matrix of the obtained process.

[0139] S241: For each process , generate an aircraft scheduling sequence based on the job allocation matrix:

[0140]

[0141] Wherein, represents the aircraft scheduling sequence, and satisfies:

[0142]

[0143] Wherein, represents the aircraft process corresponding to the job allocation matrix element value;

[0144] S242: Initialize the earliest machine time vector and the earliest start time vector of the aircraft group :

[0145]

[0146]

[0147] Wherein, K represents the total number of machines;

[0148] S243: Traverse the aircraft in sequence , and perform the operations of machine selection, dispatch start time determination, and dispatch end time determination in turn, and update the earliest machine time vector and the earliest start time vector of the aircraft group ;

[0149] In machine selection, the machine with the earliest available machine time is selected as the machine assigned to the current aircraft for the current process. According to the process exclusive constraint, each aircraft is assigned to only one machine in each process, and the machine selection satisfying the constraint is denoted as:

[0150]

[0151] The launch start time of the aircraft in the process is determined by ensuring that the aircraft has finished the previous process of the current process, the launch start time is not earlier than the end time of the previous task of the machine, and the assigned machine is idle. The launch start time is:

[0152]

[0153] The launch end time of the aircraft in the process is determined by ensuring that the aircraft has finished the previous process of the current process, the launch end time is not earlier than the end time of the previous task of the machine, and the assigned machine is idle. The launch end time is:

[0154]

[0155] wherein denotes the time consumption of the aircraft in the process ; After determining the launch end time, the earliest time vector of the machine

[0156] and the earliest start time vector of the aircraft group are updated as:

[0157]

[0158]

[0159] S244: Record all scheduling tasks, generate a scheduling scheme, wherein the scheduling tasks include aircraft number, machine number, process number, start time and end time;

[0160] The scheduling task is:

[0161]

[0162] The scheduling scheme is:

[0163] .

[0164] ​​​The maximum value in all scheduling task end times is taken as the scheduling time, and the scheduling time is:

[0165]

[0166] The algorithm realizes the machine exclusive constraint by dynamically maintaining the machine time state vector. When the aircraft carrier is allocated to the machine , the start time of the aircraft carrier satisfies , and the and are updated immediately after the processing task is allocated, when multiple aircraft carriers need to use the same machine and the same process, the algorithm is sequentially allocated according to the global optimal solution order, ensures that the end time of the first processing process is not earlier than the start time of the later processing process constraint, strictly establishes the process chain relationship, and ensures the continuity of the current machine time window. In physical meaning, this processing method avoids the time overlap of different aircraft carrier operations on the same machine.

[0167] A separate machine for the last departure and takeoff is set, combined with the above aircraft carrier related constraints, to limit the maximum number of aircraft carriers that can take off at the same time in the departure and takeoff stage, strictly guarantee the departure safety constraint, meet the spatial anti-collision safety constraint requirement of the departure and takeoff process, and allow at most one aircraft carrier to take off.

[0168] The decoding algorithm realizes the strict satisfaction of all constraints in a smaller time complexity through the cooperative update of the state vectors , . This design based on space-time resource management not only guarantees the rigorousness of the analysis, but also meets the real-time requirement of the aircraft carrier deck operation.

[0169] The three perturbation candidate solutions obtained by the decoding scheduling process are decoded and scheduled by the decoding scheduling process to generate a takeoff scheduling scheme, and the overall scheduling time of the aircraft carrier group is calculated by a scheduling decoding function , is the input takeoff scheduling scheme, and the better scheduling scheme is selected by comparing the overall scheduling time of the aircraft carrier group corresponding to all perturbation candidate solutions.

[0170] S25: compare the scheduling time of each feasible scheduling scheme, select a better scheduling scheme to update the current solution, and maintain the global optimal solution:

[0171]

[0172]

[0173]

[0174] Among them, represents the first the updated job assignment matrix after the sub-iteration, denotes a scheduling decoding function, denotes a dispatching scheduling scheme in the scheduling scheme set, denotes a scheduling scheme set, denotes that the function returns the maximum completion time;

[0175] maintain the global optimal solution:

[0176]

[0177] wherein, denotes the global optimal solution in the iteration process.

[0178] This targeted perturbation strategy effectively balances the local search and global exploration ability, ensuring efficient optimization of the algorithm in the solution space.

[0179] In an embodiment of the present application, the departure scheduling of a group of carrier-based aircraft is taken as the research object, and a carrier operation scene containing 3 available takeoff positions and 18 deck parking positions is constructed. Dynamic scheduling simulation analysis is carried out for three typical task scales of 8, 12 and 18 carrier-based aircrafts respectively. The relevant data of the example is based on public data and reasonable assumptions. Through preliminary parameter comparison, it is determined that , the neighborhood search step d=7. The remaining experimental parameters are: number of processes num_phases=4, total number of machines num_machines=10, and number of iterations max_iter=250.

[0180] As Figures 3-8 Under certain carrier operation conditions, the Gantt charts for the departure scheduling of 8, 12 and 18 carrier-based aircrafts are respectively shown, which include carrier-based aircraft operation Gantt charts and machine operation Gantt charts. As shown in Figure 3 , different color blocks represent different process stages, the vertical axis is the carrier-based aircraft number, and the numbers in the color blocks represent the machine number and the process number respectively, such as "M7-1" corresponding to the 1st stage of the 1st carrier-based aircraft on the 7th machine, the length of the color block represents the required time for the operation, and so on. As shown in Figure 4 , different color blocks represent different process stages of different carrier-based aircrafts processed by each machine, the vertical axis is the machine number, and the numbers in the color blocks represent the carrier-based aircraft number and the process number respectively, such as "J6-1" corresponding to the 1st stage of the 1st machine processing the 6th carrier-based aircraft, and the length of the color block represents the start time of the processing and the required time for the operation. As can be seen from Figures 3-8 , the operation scheduling in each subgraph satisfies the model constraint condition, each stage and machine are mutually exclusive and compatible, and there is no conflict situation, i.e. the given scheme is completely feasible, and the algorithm can effectively solve the departure scheduling design.

[0181] Figure 9 The iteration convergence characteristics of the aircraft scale of 8, 12 and 18 are shown, the horizontal axis represents the iteration number, and the vertical axis represents the maximum departure time target value. The three convergence curves clearly show the iteration optimization trajectory of the dispatching time under different dispatching scales, and the target function value presents a significant downward trend and approaches the stable optimal solution within a limited number of iterations. The star points of the figure represent the feasible solutions found in the iteration process that meet the constraint conditions, and record the update time of the feasible solution. The optimization trajectory is obviously related to the dispatching scale, and the increase of the dispatching scale increases the target value.

[0182] Table 1 is the algorithm related index of three aircraft dispatching scales, the convergence process of the algorithm is shown in Figure 9 The target values of the aircraft dispatching scales of 8, 12 and 18 are 252s, 327s and 480s respectively, and the average aircraft dispatching time decreases with the increase of the dispatching scale, and the average completion time of each aircraft is between 26 and 32 seconds, which is close to the actual dispatching operation condition, and the total time required for the takeoff of 18 aircrafts under the condition of dispatching condition guarantee is within 10 minutes.

[0183] Table 1 is the dispatching simulation results of different scales of aircraft

[0184]

[0185] In this embodiment, the PSO and DE intelligent search algorithms are combined for simulation comparison experiment, the population size is initialized to 50, the iteration number is 500, and the parameters are set as follows: the inertia weight in PSO is 0.9, the acceleration factor is (1.5, 1.8), and the population size is 50; the mutation factor in DE is 0.9, the crossover probability is 0.9, and the population size is 50. Figures 10-12 The iteration processes of the three algorithms under the conditions of the aircraft dispatching scales of 8, 12 and 18 are compared. Figure 10 It is shown that the algorithm of the application can efficiently search for the optimal solution, and the target value rapidly decreases in the iteration process and is updated within about 200 iterations. Figure 11 It is shown that the PSO algorithm is easy to fall into local optimal value in the process of searching for the optimal value, and the quality of the optimal solution found is not high, and the solving effect for more decision variables does not reach the optimal value, and the iteration number required for optimization is more. Figure 12 It can be seen that the solving performance of the DE algorithm in the process of searching for the optimal value is low, the iteration update efficiency is low, the optimal solution found is less and the quality is not high.

[0186] The performance index of the algorithm of the present application and the PSO algorithm and DE algorithm is shown in Table 2. Table 2 compares the scheduling performance index of the classical swarm intelligence optimization algorithm and the algorithm of the present application, including the main indexes such as solution running time, decision variable quantity, best target value, average single aircraft carrier aircraft sortie time, etc. For the solution running, since the decision variable realized by the algorithm of the present application is less, the overall solution running time is better than that of the PSO and DE algorithms. Under the sortie scale of 12 and 18 aircrafts, the algorithm of the present application has a better target value and faster solution time. The comprehensive results show that the algorithm of the present application has significant advantages for the scheduling optimization of a large number of aircrafts and has a good effect on sortie scheduling.

[0187] Table 2 Performance comparison of different algorithms for sortie scheduling

[0188]

[0189] The improved meta-heuristic search algorithm proposed in the present application can quickly search for an ideal solution by introducing random disturbance search of a neighboring solution, and the random disturbance matrix is determined by the current solution, and few additional variables are used. The algorithm realizes scheduling decoding in combination with specific problem constraints, accurately converts the decision solution (job allocation matrix) into a feasible scheduling scheme, and effectively avoids the problem of difficulty in searching for an ideal solution caused by multiple constraints. By comparing with the particle swarm optimization algorithm, the differential evolution algorithm and other classical optimization algorithms, the feasibility and superiority of the algorithm in solving the aircraft sortie scheduling problem are verified. The present application has the following advantages:

[0190] Efficient solution representation and update: a low-dimensional job allocation matrix is used to represent the solution, and a random disturbance update strategy inspired by signal denoising is designed to perform efficient neighborhood search in the solution space;

[0191] Decoding mechanism: a departure scheduling decoder that satisfies constraints is designed to accurately convert the job allocation matrix into a feasible scheduling scheme that satisfies all complex hard constraints, ensuring the feasibility of the solution;

[0192] Significant efficiency advantage: by reducing the dimension of the decision variable, the algorithm can obtain a high-quality solution in a very short time.

[0193] Those skilled in the art will realize that the embodiments described herein are intended to help the reader understand the principles of the present application and should be understood as not limiting the scope of protection of the present application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations according to the technical inspirations disclosed in the present application without departing from the essence of the present application, and these modifications and combinations are still within the scope of protection of the present application.

Claims

1. An improved method for optimizing the dispatch of carrier-based aircraft for efficient departure, characterized in that, The method comprises the following steps: S1: constructing a carrier-based aircraft dispatching departure model, the dispatching model being a job shop scheduling model with multiple constraints; S2: solving the dispatching model by using an improved meta-heuristic recursive neural algorithm, representing a solution by using a job assignment matrix, calculating a new solution by using a perturbation update strategy and iteratively optimizing to obtain an optimal scheduling scheme.

2. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 1, wherein, The objective function of the job shop scheduling model in S1 is: ; wherein, Ttotal represents the total time for all aircraft to complete all processes, Taircraft represents the time for an aircraft to complete a process, Tprocess represents the time for a process to be completed on a machine, Tprocess represents the time for a process to be completed on a machine, Tprocess represents the time for a process to be completed on a machine, Tprocess represents the time for a process to be completed on a machine, Tprocess represents the time for a process to be completed on a machine, The constraints of the job shop scheduling model include: Process exclusive constraint: ; wherein indicates a process machineset, indicates a carrier-based aircraft first process is machined machines, indicates a carrier-based aircraft first process is not machined machines, indicates an arbitrary symbol; Machine exclusive constraint: ; Time constraint: ; ; ; ; wherein, and respectively represent carrier-based aircraft the start time and processing time of the machine in the process , represent, represent the machine number in the previous process, represent the end time of the previous process, represent carrier-based aircraft the start time of the machine in the process , represent carrier-based aircraft the processing end time of the machine in the process , and represent different carrier-based aircraft numbers, represent the processing sequence of carrier-based aircraft in the same process of the same machine; Space anti-collision safety constraint: ; wherein represents the machining machine set of the 4th process, represents the shipboard aircraft of the 4th process in the machine processing.

3. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 2, wherein, The step S2 comprises the following steps: S21: initializing a job assignment matrix, an element value of the job assignment matrix encoding a priority sequence of a carrier-based aircraft in each process; S22: calculating a perturbation matrix and performing normalization processing; S23: generating multiple perturbation candidate solutions based on the perturbation matrix; S24: converting the perturbation candidate solutions into feasible scheduling schemes by using a decoding scheduling process, and calculating a carrier-based aircraft group overall scheduling time corresponding to each feasible scheduling scheme; S25: comparing scheduling times of each feasible scheduling scheme, selecting a better scheduling scheme to update a current solution, and maintaining a global optimal solution; S26: judging whether a maximum iteration number is reached, if not, returning to S21, and if yes, outputting the optimal scheduling scheme.

4. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 3, wherein, The job assignment matrix in S21 is: ; wherein, denotes the job assignment matrix of the tth iteration, denotes the process of carrier-based aircraft launch priority matrix, denotes the carrier-based aircraft launch priority of the process , denotes the transpose of a matrix, is the current iteration number, denotes the field of real numbers.

5. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 4, wherein, The perturbation matrix in S22 is: ; ; wherein denotes a perturbation matrix, is a standard normal random matrix, denotes the Frobenius norm.

6. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 5, wherein, The perturbation candidate solution in S23 is: ; ; ; wherein, , and denote three perturbed candidate solutions, denotes the job assignment matrix of the th iteration, d denotes the neighborhood search step size, is a set of setting parameters.

7. The improved carrier aircraft launch and departure high efficiency scheduling optimization method of claim 6, wherein, The decoding scheduling process in S24 comprises the following steps: S241: For each process , generate a carrier aircraft scheduling sequence based on the job assignment matrix: ; wherein denotes the carrier-based aircraft dispatch sequence; S242: initialize machine earliest time vector and carrier air group earliest start time vector : ; ; Wherein, K represents a total number of machines; S243: by sequence Traverse the aircrafts, and sequentially perform the operations of machine selection, determination of the start time, determination of the end time, and update of the earliest time vector of the machine and the earliest start time vector of the aircraft group ; Machine selection is represented as: ; Dispatch start time for: ; Dispatch end time Is: ; wherein embodiments in process time-consuming Updated machine earliest time vector And carrier air group earliest start time vector Is: ; ; S244: recording all scheduling tasks to generate a scheduling scheme, wherein the scheduling tasks include a carrier-based aircraft number, a machine number, a process number, a start time and an end time; Scheduling tasks To: ; Generating a set of scheduling schemes For: 。 8. The improved carrier-based aircraft dispatch and departure efficient scheduling optimization method according to claim 7 is characterized in that: The formula for selecting a better scheduling scheme to update a current solution and maintaining a global optimal solution in S25 is: ; ; ; wherein, represents the updated job assignment matrix after the th iteration, represents a scheduling decoding function, represents one of the scheduling schemes in the set of scheduling schemes, represents the set of scheduling schemes, the function returns the maximum completion time as the end of processing time of the aircraft at the current process and machine ; The global optimal solution is maintained as follows: ; wherein, denotes the global optimum solution in the iteration process.

Images