An aircraft fleet dynamic recovery method, system, device, and medium

Abstract

The application discloses an aircraft fleet dynamic recovery method, system, device and medium, and relates to the field of air traffic control.The method comprises the following steps: acquiring state data of each aircraft in the current decision stage aircraft fleet; constructing a state space of the current decision stage based on the state data; inputting the state space into a strategy network to obtain action probabilities of each aircraft in the current decision stage; determining action mask vectors of each aircraft in the current decision stage based on the values of the remaining fuel quantity and the recovery state; optimizing the action probabilities based on the action mask vectors, and performing normalization processing on the optimized action probabilities to obtain normalized action probabilities of each aircraft in the current decision stage; and recovering the aircraft with the highest normalized action probability and the smallest remaining fuel quantity. The application can improve the real-time performance, accuracy and efficiency of aircraft recovery.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of air traffic control, in particular to a method, system, device and medium for dynamic recovery of aircraft fleet. BACKGROUND

[0002] In modern air management operations, aircraft recovery sequencing is the core link to determine the recovery efficiency. In some cases, the aircraft recovery sequencing method mainly relies on predefined priority rules, such as determining the order according to the fuel emergency level or task level, and there are also cases of using heuristic algorithms for static scheduling.

[0003] However, in the complex recovery environment of the open sea with high real-time performance, the above methods have significant defects: (1) insufficient dynamic response: unable to handle dynamic events such as changes in fuel consumption and sudden mechanical failures in real time; (2) rigid constraint processing: when abnormal conditions such as aircraft reflight and mechanical failure occur, manual intervention is required to adjust the recovery queue; (3) single optimization goal: only the fuel emergency level or task level is determined as the optimization goal, and it is difficult to balance multiple goals such as recovery success rate, task priority and resource utilization.

[0004] Although the existing reinforcement learning method has the potential for dynamic decision-making, there are two major bottlenecks in the design of state-action space: on the one hand, the state representation fails to integrate the coupling effect between mechanical integrity and real-time fuel consumption, resulting in an inaccurate and comprehensive description of the aircraft state; on the other hand, the action space lacks an embedded processing mechanism for physical constraints such as cooling forbidden selection rules and emergency channel restrictions, making it difficult for the decision-making process to conform to the actual physical conditions. SUMMARY

[0005] The purpose of the present application is to provide a method, system and device for dynamic recovery of aircraft fleet, which can improve the real-time performance, accuracy and efficiency of aircraft recovery.

[0006] To achieve the above purpose, the present application provides the following solutions:

[0007] In a first aspect, the application provides an aircraft fleet dynamic recovery method, comprising: obtaining state data of each aircraft in the aircraft fleet in a current decision stage; constructing a state space of the current decision stage based on the state data; the state space comprises numerical values of residual fuel quantity, mechanical integrity, task priority, recovery success rate and recovery state; inputting the state space into a policy network to obtain action probabilities of each aircraft in the current decision stage; the policy network is obtained by training using a PPO algorithm; determining an action mask vector of each aircraft in the current decision stage based on the numerical values of the residual fuel quantity and the recovery state; optimizing the action probabilities based on the action mask vector, and performing normalization processing on the optimized action probabilities to obtain normalized action probabilities of each aircraft in the current decision stage; and recovering an aircraft with the highest normalized action probability and the smallest residual fuel quantity.

[0008] In a second aspect, the application provides an aircraft fleet dynamic recovery system, comprising: an obtaining module configured to obtain state data of each aircraft in the aircraft fleet in a current decision stage; a constructing module configured to construct a state space of the current decision stage based on the state data; the state space comprises numerical values of residual fuel quantity, mechanical integrity, task priority, recovery success rate and recovery state; a policy module configured to input the state space into a policy network to obtain action probabilities of each aircraft in the current decision stage; the policy network is obtained by training using a PPO algorithm; a determining module configured to determine an action mask vector of each aircraft in the current decision stage based on the numerical values of the residual fuel quantity and the recovery state; an optimization and normalization module configured to optimize the action probabilities based on the action mask vector, and perform normalization processing on the optimized action probabilities to obtain normalized action probabilities of each aircraft in the current decision stage; and a recovery module configured to recover an aircraft with the highest normalized action probability and the smallest residual fuel quantity.

[0009] In a third aspect, the application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aircraft fleet dynamic recovery method described above.

[0010] In a fourth aspect, the application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the steps of the aircraft fleet dynamic recovery method described in any one of the above aspects.

[0011] According to the specific embodiments provided in the application, the application has the following technical effects.

[0012] (1) The action probability of each aircraft in the current decision stage is optimized and normalized by the action mask vector of each aircraft in the current decision stage, so that the normalized action probability of each aircraft in the current decision stage can be obtained in real time, dynamic optimization is realized, and the real-time performance of aircraft recovery is improved.

[0013] (2) The state space is represented by the numerical values of the remaining fuel quantity, mechanical integrity, task priority, recovery success rate and recovery state, various abnormal conditions of the aircraft can be considered in all aspects, dynamic adjustment can be realized through the strategy network, and the problem of manual intervention adjustment is solved; meanwhile, the action mask vector of each aircraft in the current decision stage is determined by the numerical values of the remaining fuel quantity and the recovery state, and the action probability generated by the state space is optimized through the action mask vector, so that the optimization of the action probability generated by the state space by the numerical values of the remaining fuel quantity and the recovery state is realized, and the singleness of optimization by the fuel emergency level or the task level is solved.

[0014] (3) The action probability of each aircraft in the current decision stage is optimized and normalized by the action mask vector of each aircraft in the current decision stage, so that the normalized action probability obtained is more accurate, and the subsequent selected action is more accurate, thereby improving the accuracy of aircraft recovery.

[0015] (4) The action probability of each aircraft in the current decision stage is optimized and normalized by the action mask vector of each aircraft in the current decision stage, so that the action probability of 0 can be masked by the mask vector, and after normalization and screening, the action in the current decision stage can be quickly screened out, thereby improving the efficiency of aircraft recovery. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technical solutions, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0017] Figure 1 A flowchart of a kind of aircraft fleet dynamic recovery method provided by an embodiment of the present application.

[0018] Figure 2 A structure diagram of a kind of aircraft fleet dynamic recovery system provided by an embodiment of the present application.

[0019] Figure 3 A structure diagram of a kind of computer equipment provided by an embodiment of the present application. DETAILED DESCRIPTION

[0020] With reference to the drawings, the technical solutions in the embodiments of the present application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0021] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0022] In one exemplary embodiment, as shown in Figure 1 An aircraft fleet dynamic recovery method is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or can be executed by a terminal and a server together. In the embodiments of the present application, the method is applied to a server as an example, and includes the following steps S1 to S6.

[0023] Step S1: Obtain state data of each aircraft in the current decision stage of the aircraft fleet.

[0024] Further, the state data includes initial remaining fuel quantity, initial mechanical integrity, initial task priority and flight state.

[0025] Step S2: Construct a state space of the current decision stage based on the state data; the state space includes values of remaining fuel quantity, mechanical integrity, task priority, recovery success rate and recovery state.

[0026] Specifically, the state space of the current decision stage includes , wherein, is a remaining fuel quantity set; is a mechanical integrity set; is a task priority set; is a recovery success rate set; is a recovery state value set. The remaining fuel quantity set includes the remaining fuel quantity of all aircraft, the mechanical integrity set includes the mechanical integrity of all aircraft, the task priority set includes the task priority of all aircraft, the recovery success rate set includes the recovery success rate of all aircraft, and the recovery state value set includes the recovery state of all aircraft.

[0027] Further, step S2 specifically includes steps S21 to S26.

[0028] Step S21: Determine the value of the recovery status based on the flight status.

[0029] Furthermore, flight status includes: recovered status and unrecovered status; unrecovered status includes: airborne standby status, go-around status, and mechanical failure status.

[0030] Step S21 specifically includes: when the flight status is "recovered", the recovery status value is determined to be 1; when the flight status is "in-flight standby", the recovery status value is determined to be 0; when the flight status is "go-around", the recovery status value is determined to be -1; when the flight status is "mechanical failure", the recovery status value is determined to be -2. The numerical value of the recovery status of each aircraft .

[0031] Specifically, when the flight status is initially in a go-around state, the first Aircraft cooling timer Updated after each decision-making stage ,when The system automatically switches the flight status to an aerial standby status, thus determining the recovery status value to be 0.

[0032] Step S22: Normalize the initial remaining fuel quantity to obtain the remaining fuel quantity.

[0033] Furthermore, the formula for calculating the remaining fuel quantity is as follows.

[0034] .

[0035] in, For the first The aircraft is in the decision-making stage. The remaining fuel quantity; For the first The aircraft is in the decision-making stage. The initial remaining fuel quantity; The number of aircraft in the aircraft fleet; For the first The aircraft is in the decision-making stage. The minimum initial remaining fuel quantity; For the first The aircraft is in the decision-making stage. The maximum value of the initial remaining fuel quantity.

[0036] Specifically, For the first The aircraft is in the decision-making stage. The initial remaining fuel quantity (i.e., the actual remaining fuel quantity (unit: tons)) is updated using the following fuel consumption formula: ,in, For the first The aircraft is in the decision-making stage. The initial remaining fuel quantity, No. The aircraft is in the decision-making stage. The initial remaining fuel quantity, For consumption rate, For adjacent decision time steps, Second, Tons per second is the typical oil consumption rate.

[0037] Step S23: Normalize the initial mechanical integrity to obtain the mechanical integrity value.

[0038] Furthermore, the formula for calculating mechanical integrity is as follows.

[0039] .

[0040] in, For the first The mechanical integrity of an aircraft; For the first The initial mechanical integrity of an aircraft; For maximum mechanical integrity, .

[0041] Step S24: Normalize the initial task priorities to obtain the task priorities.

[0042] Furthermore, the formula for calculating the normalized initial task priority is as follows.

[0043] .

[0044] in, For the first Normalized initial mission priority for each aircraft; For the first Initial mission priority for each aircraft; As the highest task priority, .

[0045] Furthermore, the normalized initial task priority is optimized by using the values ​​of the recycled state to obtain the task priority.

[0046] Specifically, the optimized expression for task priority is as follows.

[0047]

[0048] in, For the first The mission priority of each aircraft For the dynamic adjustment factor, when , , when , , the rest of the case .

[0049] Step S25: Calculate the recovery success rate based on the remaining fuel quantity and the mechanical integrity.

[0050] Further, the calculation formula of the recovery success rate is as follows.

[0051] .

[0052] Among them, is the recovery success rate of the i th aircraft; is the minimum recovery success rate, .

[0053] Step S26: Based on the values of the remaining fuel quantity, the mechanical integrity, the task priority, the recovery success rate and the recovery state, the state space of the current decision stage is constructed.

[0054] Step S3: Input the state space into the policy network to obtain the action probability of each aircraft in the current decision stage; the policy network is obtained by training using the PPO algorithm.

[0055] Specifically, the network structure of the policy network is a 3-layer fully connected network, each layer has 128 neurons, the activation function is ReLU, the output layer is Softmax, and the hyperparameters include the discount factor , the target function clip range , the value function coefficient , the entropy regularization coefficient , and the learning rate .

[0056] The policy network is trained using sample state space and sample action, and the generation process of the sample state space and the sample action is as follows.

[0057] (1) The hyperparameters are set as follows: the discount factor , the target function clip range , the value function coefficient , the entropy regularization coefficient , and the learning rate .

[0058] (2) Initialize the environment, set .

[0059] (3) Obtain the sample state space of the current decision stage. ​

[0060] (4) Input the sample state space of the current decision stage to the policy network, and output the sample action probability of each aircraft in the current decision stage.

[0061] (5) Optimize the sample action probability through the action mask vector, and normalize the optimized sample action probability to obtain the normalized sample action probability of each aircraft in the current decision stage. Select the aircraft with the highest normalized sample action probability in the normalized sample action probability of each aircraft in the current decision stage. If the normalized sample action probability is equal, select the aircraft with the highest normalized sample action probability and the lowest remaining fuel.

[0062] (6) The aircraft selected in (5) is taken as the sample action, and the sample action is executed to realize the recovery of the aircraft.

[0063] (7) After execution, the environment is interacted to obtain the sample state space of the next decision stage.

[0064] Among them, the acquisition process of the flight state in the sample state space of the next decision stage: recovery success determination, the determination condition is: the uniform random number is less than or equal to the recovery success rate of the th aircraft , wherein , is a uniform random number in the interval , and is the recovery success rate of the th aircraft. If yes (i.e. ), it is determined that the flight state of the aircraft is , if not (i.e. ). At the same time, the aircraft in the reflight state is updated, and the value of the recovery state is determined to be 0 when re-entering the queue.

[0065] (8) Determine whether the sample action is successful, and according to the judgment result, the environment returns the reward .

[0066] After all the aircraft are recovered, collect the trajectory samples (sample state space, sample action and sample environment return reward), T is the total number of decision stages, and the training cycle is independent. After that, update the parameters according to the following formula, and update 500 times.

[0067] Based on the above trajectory samples, the parameters of the policy network and the value learning network are updated. The parameter update formula of the policy network is as follows.

[0068] .

[0069] .

[0070] in, For the parameters of the policy network, As expected, For policy networks, For sample actions, Let the sample state space be... These are the initial parameters for the policy network. Therefore The strategy network and The policy network in Select action in state The probability ratio of actions, For the dominant function, As a shearing function, it can be used to... The shearing function is restricted to , This is the parameter of the shearing function, usually taken as 0.2.

[0071] The parameter update formula for the value learning network is as follows.

[0072] .

[0073] in, For the parameters of the value learning network, For value networks, The discount factor is a power of itself. For the stage The sample environment returns a reward. This is a stage index.

[0074] During initial training, the parameters of the policy network are the same as the initial parameters of the policy network. In subsequent training, every 10 training iterations, the initial parameters of the policy network are updated to the parameters of the policy network in the 10th iteration, thus realizing parameter updates.

[0075] Step S4: Based on the values ​​of remaining fuel quantity and recovery status, determine the action mask vector of each aircraft in the current decision-making phase.

[0076] Further, step S4 specifically includes: when the value of the recovery status is 1 or -1, determining the action mask vector of the aircraft in the current decision-making stage to be 0; when the value of the recovery status is -2 or the remaining fuel is less than the remaining fuel threshold, determining the action mask vector of the aircraft in the current decision-making stage to be 2; when the value of the recovery status is 0 or the remaining fuel is greater than or equal to the remaining fuel threshold, determining the action mask vector of each aircraft in the current decision-making stage to be 1.

[0077] Specifically, the current decision-making stage is... The mask vector of each aircraft .

[0078] Step S5: Optimize the action probability based on the action mask vector, and normalize the optimized action probability to obtain the normalized action probability of each aircraft in the current decision-making stage.

[0079] Further, step S5 specifically includes: if the action mask vector is 0, then the action probability is optimized to 0; if the action mask vector is 1, then the action probability remains unchanged; if the action mask vector is 2, then the action probability is optimized to 1.

[0080] Furthermore, the formula for calculating the normalized action probability of each aircraft in the current decision-making stage is as follows.

[0081] .

[0082] in, For the first Normalized probability of an aircraft's actions; For the first The motion mask vector of an aircraft; For the first The optimized probability of each aircraft's actions; For the first The optimized probability of each aircraft's actions; For the first The motion mask vector of each aircraft.

[0083] Step S6: Recover the aircraft with the highest normalized motion probability and the smallest remaining fuel.

[0084] Specifically, , ,in, For discrete action space, This represents the total number of aircraft in the fleet that have not been recovered. , The aircraft with the highest normalized probability of action and the smallest remaining fuel is selected.

[0085] The beneficial effects of the dynamic recovery method for aircraft fleet proposed in this application are mainly reflected in the following aspects.

[0086] (1) This application comprehensively considers the various situations of the aircraft in the five-dimensional state space by normalizing the remaining fuel quantity, normalizing the mechanical integrity, normalizing the task priority, recovery success rate and state marking, so as to achieve dynamic adjustment through the combination of strategy network and action mask vector, thereby realizing dynamic optimization and improving the real-time performance of aircraft recovery.

[0087] (2) The application innovatively introduces the action mask mechanism to dynamically shield invalid actions, adjusts the emergency probability of the fault state or the remaining fuel amount of the aircraft to 1, can quickly filter out the actions in the current decision stage, improves the efficiency of aircraft recovery, and at the same time, makes the normalized action probability more accurate, so that the subsequently filtered actions are more accurate, and the accuracy of aircraft recovery is improved.

[0088] Based on the same inventive concept, the embodiments of the application also provide an aircraft fleet dynamic recovery system. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method, so the specific limitations in one or more aircraft fleet dynamic recovery system embodiments provided below can refer to the limitations of the aircraft fleet dynamic recovery method in the foregoing, which will not be repeated here.

[0089] In an exemplary embodiment, as shown in Figure 2 An aircraft fleet dynamic recovery system is provided, including: an acquisition module configured to acquire state data of each aircraft in a current decision stage of an aircraft fleet; a construction module configured to construct a state space of the current decision stage based on the state data; the state space includes values of a remaining fuel amount, a mechanical integrity, a task priority, a recovery success rate, and a recovery state; a strategy module configured to input the state space into a strategy network to obtain action probabilities of each aircraft in the current decision stage; the strategy network is obtained by training using a PPO algorithm; a determination module configured to determine an action mask vector of each aircraft in the current decision stage based on the values of the remaining fuel amount and the recovery state; an optimization and normalization module configured to optimize the action probabilities based on the action mask vector, and perform normalization processing on the optimized action probabilities to obtain normalized action probabilities of each aircraft in the current decision stage; and a recovery module configured to recover an aircraft with the highest normalized action probability and the smallest remaining fuel amount.

[0090] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram of the computer device can be as shown in Figure 3As shown in the figure. The computer device includes a processor, a memory, an Input / Output (I / O) interface, and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the state data of each aircraft in the current decision stage aircraft fleet. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement an aircraft fleet dynamic recovery method.

[0091] Those skilled in the art can understand that, Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement. In one exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in each of the method embodiments described above.

[0092] In one exemplary embodiment, a computer readable storage medium is provided, storing a computer program, which is executed by a processor to implement the steps in each of the method embodiments described above.

[0093] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0094] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to a memory, a database or other medium used in the embodiments provided in the present application can include at least one of a non-volatile and a volatile memory. The non-volatile memory can include a read-only memory (ROM), a magnetic tape, a floppy disk, a flash memory, an optical storage, a high-density embedded non-volatile memory, a resistive random access memory (ReRAM), a magnetoresistive random access memory (MRAM), a ferroelectric random access memory (FRAM), a phase change memory (PCM), a graphene memory, etc. The volatile memory can include a random access memory (RAM) or an external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as a static random access memory (SRAM) or a dynamic random access memory (DRAM), etc.

[0095] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0096] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0097] The principles and implementation modes of the present application are described by applying specific examples in the present application. The above embodiments are only used to help understand the method and its core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range can be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for dynamic recovery of a fleet of aircraft, characterized in that, The aircraft fleet dynamic recovery method comprises: Obtaining state data of each aircraft in the current decision stage aircraft fleet; the flight state comprises: a recovered state, an unrecovered state; the unrecovered state comprises: an airborne standby state, a reflight state and a mechanical failure state; when the flight state is the recovered state, the value of the recovery state is determined as 1; when the flight state is the airborne standby state, the value of the recovery state is determined as 0; when the flight state is the reflight state, the value of the recovery state is determined as -1; when the flight state is the mechanical failure state, the value of the recovery state is determined as -2; Based on the state data, a state space of the current decision stage is constructed; the state space comprises: a remaining fuel quantity, a mechanical integrity, a task priority, a recovery success rate and the value of the recovery state; The state space is input into a policy network to obtain an action probability of each aircraft in the current decision stage; the policy network is obtained by training using a PPO algorithm; Based on the remaining fuel quantity and the value of the recovery state, an action mask vector of each aircraft in the current decision stage is determined, specifically comprising: when the value of the recovery state is 1 or -1, the action mask vector of the aircraft in the current decision stage is determined as 0; when the value of the recovery state is -2 or the remaining fuel quantity is less than a remaining fuel quantity threshold, the action mask vector of the aircraft in the current decision stage is determined as 2; when the value of the recovery state is 0 or the remaining fuel quantity is greater than or equal to the remaining fuel quantity threshold, the action mask vector of each aircraft in the current decision stage is determined as 1; Based on the action mask vector, the action probability is optimized, and the optimized action probability is normalized to obtain a normalized action probability of each aircraft in the current decision stage, specifically comprising: if the action mask vector is 0, the action probability is optimized as 0; if the action mask vector is 1, the action probability remains unchanged; if the action mask vector is 2, the action probability is optimized as 1; The calculation formula of the normalized action probability of each aircraft in the current decision stage is: ； wherein, is the normalized action probability for the th aircraft; is the action mask vector for the th aircraft; is the optimized action probability for the th aircraft; is the optimized action probability for the th aircraft; is the action mask vector for the th aircraft; The aircraft with the highest normalized action probability and the smallest remaining fuel quantity is recovered.

2. The method of claim 1, wherein, The state data comprises: an initial remaining fuel quantity, an initial mechanical integrity, an initial task priority and a flight state; Based on the state data, a state space of the current decision stage is constructed, specifically comprising: Based on the flight state, the value of the recovery state is determined; The initial remaining fuel quantity is normalized to obtain the remaining fuel quantity; The initial mechanical integrity is normalized to obtain the mechanical integrity; The initial task priority is normalized to obtain a normalized initial task priority; The normalized initial task priority is optimized by the value of the recovery state to obtain the task priority; The recovery success rate is calculated based on the remaining fuel quantity and the mechanical integrity; The state space of the current decision stage is constructed based on the remaining fuel quantity, the mechanical integrity, the task priority, the recovery success rate and the value of the recovery state.

3. The method of claim 2, wherein, The calculation formula of the remaining fuel quantity is: ； in, For the first The aircraft is in the decision-making stage. The remaining fuel quantity; For the first The aircraft is in the decision-making stage. The initial remaining fuel quantity; The number of aircraft in the aircraft fleet; For the first The aircraft is in the decision-making stage. The minimum initial remaining fuel quantity; For the first The aircraft is in the decision-making stage. The maximum value of the initial remaining fuel quantity; The calculation formula of the mechanical integrity is: ； wherein, is the mechanical integrity of the th aircraft; is the initial mechanical integrity of the th aircraft; is the maximum mechanical integrity; The calculation formula of the normalized initial task priority is: ； in, For the first Normalized initial mission priority for each aircraft; For the first Initial mission priority for each aircraft; Set as the highest task priority; The calculation formula of the recovery success rate is: ； wherein, is the recovery success rate for the th aircraft; is the minimum recovery success rate.

4. An aircraft fleet dynamic recovery system, characterized by, The aircraft fleet dynamic recovery system comprises: An acquisition module is configured to acquire state data of each aircraft in the aircraft fleet at a current decision stage; the flight state comprises a recovered state, an unrecovered state, and the unrecovered state comprises an airborne standby state, a reflight state, and a mechanical failure state; when the flight state is the recovered state, the value of the recovery state is determined to be 1; when the flight state is the airborne standby state, the value of the recovery state is determined to be 0; when the flight state is the reflight state, the value of the recovery state is determined to be -1; and when the flight state is the mechanical failure state, the value of the recovery state is determined to be -2; A construction module is configured to construct a state space of the current decision stage based on the state data; the state space comprises the remaining fuel quantity, the mechanical integrity, the task priority, the recovery success rate, and the value of the recovery state; A strategy module is configured to input the state space into a policy network to obtain an action probability of each aircraft at the current decision stage; the policy network is obtained by training using a PPO algorithm; A determination module is configured to determine an action mask vector of each aircraft at the current decision stage based on the remaining fuel quantity and the value of the recovery state; specifically, when the value of the recovery state is 1 or -1, the action mask vector of the aircraft at the current decision stage is determined to be 0; when the value of the recovery state is -2 or the remaining fuel quantity is less than a remaining fuel quantity threshold, the action mask vector of the aircraft at the current decision stage is determined to be 2; and when the value of the recovery state is 0 or the remaining fuel quantity is greater than or equal to the remaining fuel quantity threshold, the action mask vector of each aircraft at the current decision stage is determined to be 1; An optimization and normalization module is configured to optimize the action probability based on the action mask vector, and perform normalization processing on the optimized action probability to obtain a normalized action probability of each aircraft at the current decision stage; specifically, if the action mask vector is 0, the action probability is optimized to 0; if the action mask vector is 1, the action probability is kept unchanged; and if the action mask vector is 2, the action probability is optimized to 1; The calculation formula of the normalized action probability of each aircraft at the current decision stage is: ； in, For the first Normalized probability of an aircraft's actions; For the first The motion mask vector of an aircraft; For the first The optimized probability of each aircraft's actions; For the first The optimized probability of each aircraft's actions; For the first The motion mask vector of an aircraft; A recovery module is configured to recover an aircraft with the highest normalized action probability and the smallest remaining fuel quantity.

5. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the aircraft fleet dynamic recovery method of any one of claims 1-3.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the aircraft fleet dynamic recovery method of any one of claims 1-3.

Images