A large-scale constellation earth observation task planning method and device

Abstract

The application provides a large-scale constellation earth observation task planning method and device, and belongs to the technical field of satellite task planning. The large-scale constellation earth observation task problem is divided into a multi-star cluster cluster allocation problem and a single-star cluster cluster autonomous task planning problem by the application, an improved adaptive large neighborhood search algorithm combined with tabu search is used to allocate the large-scale constellation earth observation task to different star clusters, and then a star cluster autonomous task planning algorithm based on an extended contract net mechanism is used to realize multi-star task planning in the star cluster. In this way, the problem scale can be reduced, the search space can be greatly reduced, the solving efficiency can be improved, the cooperation between satellites in the star cluster can be increased, especially in the face of satellite damage and interference, the anti-interference ability and cooperative completion efficiency of the task can be improved, and the robustness and reliability of the task planning can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite mission planning, in particular to a large-scale constellation earth observation mission planning method and device. BACKGROUND

[0002] With the development of satellite technology, the number and observation capability of earth observation satellites are continuously improved. How to efficiently arrange these satellite resources to meet the growing demand of users has become an important research topic.

[0003] Currently, when facing the problem of large-scale constellation earth observation mission planning, traditional methods such as exact algorithms and meta-heuristic algorithms are usually used for solving, and centralized planning is carried out on the ground. However, with the continuous increase of satellites and tasks, the solving time of the existing technology increases exponentially, which leads to the difficulty of solving the problem of large-scale constellation earth observation mission planning; at the same time, the centralized planning mode cannot be applied to the scene of satellite damage and interference, which leads to insufficient satellite autonomous cooperation and anti-interference ability. SUMMARY

[0004] The present application provides a large-scale constellation earth observation mission planning method and device to solve the problems of difficulty in solving and insufficient satellite anti-interference ability of related technologies when facing the problem of large-scale constellation earth observation mission planning.

[0005] In order to solve the above problems, the present application adopts the following technical solutions:

[0006] In a first aspect, the present application provides a large-scale constellation earth observation mission planning method, which comprises:

[0007] An improved adaptive large neighborhood search algorithm based on fusion tabu search is used to optimize and solve a preset large-scale constellation earth observation mission planning model, so as to distribute a large-scale constellation earth observation mission to multiple star clusters and obtain a target star cluster distribution scheme; wherein the target star cluster distribution scheme represents that multiple star clusters are distributed with a corresponding set of tasks to be planned respectively;

[0008] For any star cluster, a star cluster autonomous mission planning algorithm based on an extended contract net mechanism is used to distribute tasks in the set of tasks to be planned to satellites in the star cluster, so as to obtain a star cluster mission planning scheme of the star cluster;

[0009] Based on the star cluster mission planning scheme of each star cluster, a large-scale constellation earth observation mission planning scheme corresponding to the large-scale constellation earth observation mission is generated.

[0010] In an embodiment of the present application, the large-scale constellation earth observation mission planning model comprises a target function and a set of constraint conditions; wherein,

[0011] the objective function aims to maximize the total revenue of the observation tasks;

[0012] the constraint condition set comprises:

[0013] a single-orbit storage constraint condition for constraining that the storage space required by all tasks of each orbit observation of each satellite must be less than or equal to the storage capacity limit of each orbit of the satellite;

[0014] a single-orbit energy constraint condition for constraining that the energy consumed by all tasks of each orbit observation of each satellite must be less than or equal to the energy limit of each orbit of the satellite;

[0015] a task execution times constraint condition for constraining that each task needs to be observed only once on one orbit to be considered as completed observation;

[0016] a task uniqueness constraint condition for constraining that each satellite can only execute one task at the same time;

[0017] a task completion time constraint condition for constraining that each task needs to be completed before a preset task completion time;

[0018] an observation start and end time constraint condition for constraining that the actual observation start time and the actual observation end time of all tasks are within the observable time window of the satellite;

[0019] a satellite attitude adjustment constraint condition for constraining that the time length for attitude adjustment of the satellite and the time length for switching on and off of the satellite payload must be less than the time interval between the start time of the next task after the current task and the observation activity completion time of the current task.

[0020] In an embodiment of the present application, an improved adaptive large neighborhood search algorithm based on fusion tabu search is used to optimize and solve a preset large-scale constellation earth observation task planning model, so as to distribute the large-scale constellation earth observation tasks to multiple star clusters and obtain a target star cluster distribution scheme, comprising:

[0021] generating an initial solution of the algorithm and taking the initial solution as a current solution and a current optimal solution;

[0022] repeating the iterative solving step until the algorithm meets a preset iteration stopping condition; wherein the iterative solving step is: in at least one destruction operator and at least one repair operator, a target destruction operator and a target repair operator are respectively determined, and based on the tabu attribute of each task, a preset number of scheduled tasks are deleted from the current solution by the target destruction operator, and a preset number of unscheduled tasks are inserted by the target repair operator to generate a new solution; based on the new solution, the current solution and the current optimal solution are updated;

[0023] In a case where it is determined that the algorithm satisfies the iteration stopping condition, the current optimal solution is assigned as the target star cluster allocation scheme.

[0024] In an embodiment of the present application, in the preset at least one destruction operator and at least one repair operator, a target destruction operator and a target repair operator are respectively determined, comprising:

[0025] The target destruction operator is determined from the at least one destruction operator based on a current operator weight corresponding to each of the at least one destruction operator; wherein the at least one destruction operator comprises one or more of a random deletion operator, a minimum revenue deletion operator, a maximum attitude adjustment conflict deletion operator, a maximum imaging opportunity deletion operator and a maximum time window conflict deletion operator.

[0026] The target repair operator is determined from the at least one repair operator based on a current operator weight corresponding to each of the at least one repair operator; wherein the at least one repair operator comprises one or more of a maximum revenue insertion operator, a minimum time window conflict insertion operator, a minimum attitude adjustment conflict insertion operator and a minimum imaging opportunity insertion operator.

[0027] In an embodiment of the present application, based on the new solution, the current solution and the current optimal solution are updated, comprising:

[0028] In a case where the revenue of the new solution is greater than the revenue of the current solution and greater than the revenue of the current optimal solution, the new solution is updated as the current optimal solution and the current solution.

[0029] In a case where the revenue of the new solution is greater than the revenue of the current solution and less than the revenue of the current optimal solution, the new solution is updated as the current solution.

[0030] In a case where the revenue of the new solution is less than the revenue of the current solution, a target probability is determined based on the revenues of the new solution and the current solution and a current annealing temperature; and whether to accept the new solution as the current solution is determined based on the target probability.

[0031] In an embodiment of the present application, the taboo attribute comprises a deletion taboo attribute and an insertion taboo attribute; the method further comprises:

[0032] In a case where a new solution containing any task is accepted as a current solution, or in a case where a new solution after deleting the task is not accepted as a current solution, a deletion taboo attribute of the task is set to a first attribute value; the first attribute value is used to indicate that the task is prohibited from being deleted in subsequent preset number of iterations of solution.

[0033] In a case that a new solution after deleting the task is accepted as the current solution or a new solution including the task is not accepted as the current solution, the insertion taboo attribute of the task is set as a second attribute value; the second attribute value is used to indicate that the task is prohibited from being inserted in subsequent preset times of iterative solving.

[0034] In an embodiment of the present application, the method further comprises:

[0035] For any one of the at least one destruction operator and the at least one repair operator, a new solution generated based on the operator is obtained;

[0036] In a case that the new solution is updated as the current optimal solution, the operator is given a first score;

[0037] In a case that the benefit of the new solution is greater than the benefit of the current solution and less than the benefit of the current optimal solution, the operator is given a second score;

[0038] In a case that the benefit of the new solution is less than the benefit of the current solution and the new solution is accepted as the current solution, the operator is given a third score;

[0039] In a case that the benefit of the new solution is less than the benefit of the current solution and the new solution is not accepted as the current solution, the operator is given a fourth score; wherein the first score, the second score, the third score and the fourth score decrease in turn;

[0040] Every interval preset times of iterative solving, a total score of the operator in the preset times of iterative solving is determined, and a current operator weight of the operator is updated based on the total score and a historical operator weight of the operator.

[0041] In an embodiment of the present application, for any one of the star clusters, a star cluster autonomous task planning algorithm based on an extended contract net mechanism is used to allocate tasks in the set of tasks to be planned to satellites in the star cluster, to obtain a star cluster task planning scheme of the star cluster, comprising:

[0042] For any one of the star clusters, a master satellite and a slave satellite of the star cluster are determined;

[0043] The master satellite is used to publish tasks in the set of tasks to be planned to the slave satellite for bidding;

[0044] Each of the slave satellites generates a bidding scheme and delivers the bidding scheme to the master satellite;

[0045] The master satellite determines a task planning scheme corresponding to the star cluster based on the bidding scheme of each of the slave satellites.

[0046] In an embodiment of the present application, the bid scheme is generated by each slave star, including:

[0047] For any slave star, the tasks are sorted in descending order based on the priority of the tasks published by the master star, to obtain a set of observation tasks;

[0048] In an embodiment of the present application, the task planning is performed on the tasks in the set of observation tasks in order from high to low priority; the task planning includes: in the case that there is at least one observable time window that meets the task requirements of the task in the set of current observable time windows of the slave star, the earliest observable time window is determined as the target window of the task; the task is planned into a set of task plans, and the current observable time window is updated based on the target window.

[0049] In the case that the task planning of the last task in the set of observation tasks is completed, the bid scheme is obtained based on the set of task plans.

[0050] In a second aspect, based on the same inventive concept, embodiments of the present application provide a large-scale constellation earth observation task planning device, the device comprising:

[0051] An upper task allocation module is configured to optimize and solve a preset large-scale constellation earth observation task planning model based on an improved adaptive large neighborhood search algorithm of fusion tabu search, so as to allocate large-scale constellation earth observation tasks to a plurality of star clusters to obtain a target star cluster allocation scheme; wherein the target star cluster allocation scheme represents that a plurality of star clusters are allocated with respective sets of tasks to be planned.

[0052] A lower task allocation module is configured to, for any star cluster, allocate tasks in the set of tasks to be planned to satellites in the star cluster based on a star cluster autonomous task planning algorithm of an extended contract net mechanism, to obtain a star cluster task planning scheme of the star cluster.

[0053] A planning scheme generation module is configured to generate a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation task based on the respective star cluster task planning schemes of a plurality of star clusters.

[0054] In an embodiment of the present application, the large-scale constellation earth observation task planning model includes a target function and a set of constraint conditions; wherein,

[0055] The target function aims to maximize the total revenue of observation tasks;

[0056] The set of constraint conditions includes:

[0057] A single-track storage constraint condition is used to restrict that the storage space required by all tasks of each orbit observation of each satellite must be less than or equal to the storage capacity limit of each orbit of the satellite;

[0058] A single-track energy constraint condition is used to restrict that the energy consumed by all tasks of each orbit observation of each satellite must be less than or equal to the energy limit of each orbit of the satellite;

[0059] A task execution times constraint condition is used to restrict that each task needs to be successfully observed only once on an orbit to be considered as completed observation;

[0060] A task uniqueness constraint condition is used to restrict that each satellite can only execute one task at the same time;

[0061] A task completion time constraint condition is used to restrict that each task needs to be completed before a preset task completion time;

[0062] An observation start and end time constraint condition is used to restrict that the actual observation start time and the actual observation end time of all tasks are within the observable time window of the satellite;

[0063] A satellite attitude adjustment constraint condition is used to restrict that the time length for satellite attitude adjustment and the time length for satellite payload switching must be less than the time interval between the start time of the next task of the current task and the observation activity completion time of the current task.

[0064] In an embodiment of the present application, the upper task allocation module comprises:

[0065] An initial solution generation submodule is configured to generate an initial solution of the algorithm, and take the initial solution as a current solution and a current optimal solution;

[0066] An iterative solution submodule is configured to repeatedly perform an iterative solution step until the algorithm meets a preset iteration stopping condition; wherein the iterative solution step comprises: determining a target destruction algorithm and a target repair algorithm from at least one destruction algorithm and at least one repair algorithm respectively, and based on the taboo attribute of each task, deleting a preset number of scheduled tasks from the current solution by using the target destruction algorithm, inserting a preset number of unscheduled tasks by using the target repair algorithm, to generate a new solution; and updating the current solution and the current optimal solution based on the new solution;

[0067] A star cluster allocation scheme determination submodule is configured to, in a case where it is determined that the algorithm meets the iteration stopping condition, take the current optimal solution as the target star cluster allocation scheme.

[0068] In an embodiment of the present application, the iterative solution submodule comprises:

[0069] The destruction operator determination unit is configured to determine the target destruction operator from the at least one destruction operator based on a current operator weight corresponding to each of the at least one destruction operator; the at least one destruction operator comprises one or more of a random deletion operator, a minimum revenue deletion operator, a maximum attitude adjustment conflict deletion operator, a maximum imaging opportunity deletion operator, and a maximum time window conflict deletion operator.

[0070] The repair operator determination unit is configured to determine the target repair operator from the at least one repair operator based on a current operator weight corresponding to each of the at least one repair operator; the at least one repair operator comprises one or more of a maximum revenue insertion operator, a minimum time window conflict insertion operator, a minimum attitude adjustment conflict insertion operator, and a minimum imaging opportunity insertion operator.

[0071] In an embodiment of the present application, the iteration solving submodule comprises:

[0072] The first updating unit is configured to update the new solution as the current optimal solution and the current solution when the revenue of the new solution is greater than the revenue of the current solution and greater than the revenue of the current optimal solution.

[0073] The second updating unit is configured to update the new solution as the current solution when the revenue of the new solution is greater than the revenue of the current solution and less than the revenue of the current optimal solution.

[0074] The third updating unit is configured to determine a target probability based on the revenues of the new solution and the current solution and a current annealing temperature when the revenue of the new solution is less than the revenue of the current solution, and determine whether to accept the new solution as the current solution based on the target probability.

[0075] In an embodiment of the present application, the tabu attribute comprises a deletion tabu attribute and an insertion tabu attribute; and the large-scale constellation earth observation task planning apparatus further comprises:

[0076] The deletion tabu attribute determination module is configured to set a deletion tabu attribute of any task to a first attribute value when a new solution containing the task is accepted as a current solution, or when a new solution after the task is deleted is not accepted as a current solution; the first attribute value is used to indicate that the task is prohibited from being deleted in subsequent iteration solving for a preset number of times.

[0077] The insertion tabu attribute determination module is configured to, for any task, set an insertion tabu attribute of the task to a second attribute value in a case where a new solution after deletion of the task is accepted as a current solution, or in a case where a new solution including the task is not accepted as the current solution; the second attribute value is used to indicate that the task is prohibited from being inserted in subsequent preset numbers of iterations.

[0078] In an embodiment of the present application, the large-scale constellation earth observation task planning device further comprises:

[0079] The new solution obtaining module is configured to, for any one of the at least one destruction operator and the at least one repair operator, obtain a new solution generated based on the operator;

[0080] The first score module is configured to, in a case where the new solution is updated to the current optimal solution, assign a first score to the operator;

[0081] The second score module is configured to, in a case where a benefit of the new solution is greater than a benefit of the current solution and less than a benefit of the current optimal solution, assign a second score to the operator;

[0082] The third score module is configured to, in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is accepted as the current solution, assign a third score to the operator;

[0083] The fourth score module is configured to, in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is not accepted as the current solution, assign a fourth score to the operator; wherein the first score, the second score, the third score and the fourth score decrease in turn.

[0084] The operator weight updating module is configured to, every interval of preset numbers of iterations, determine a total score of the operator in the preset numbers of iterations, and update a current operator weight of the operator based on the total score and a historical operator weight of the operator.

[0085] In an embodiment of the present application, the lower-layer task allocation module comprises:

[0086] The satellite type determination submodule is configured to, for any one of the star clusters, determine a primary star and a secondary star of the star cluster;

[0087] The bidding submodule is configured to publish a task in a set of tasks to be planned to the secondary stars for bidding through the primary star;

[0088] The bidding submodule is configured to, for any one of the star clusters, determine a primary star and a secondary star of the star cluster;

[0089] A scheme determination sub-module is configured to determine a task planning scheme corresponding to the star cluster based on the bidding scheme of each slave star by the master star.

[0090] In an embodiment of the present application, the bidding sub-module comprises:

[0091] A sorting unit is configured to sort, in descending order, the observation tasks of any slave star based on the priority of the tasks issued by the master star to obtain an observation task set;

[0092] A task planning unit is configured to sequentially plan the tasks in the observation task set in the order from high to low of the priority; the task planning comprises: in the case that there is at least one observable time window satisfying the task requirement of the task in the current observable time window set of the slave star, determining the earliest observable time window as the target window of the task; planning the task into a task planning set, and updating the current observable time window based on the target window.

[0093] A bidding scheme generation unit is configured to obtain the bidding scheme based on the task planning set in the case that the task planning of the last task in the observation task set is completed.

[0094] Compared with the prior art, the present application has the following advantages:

[0095] The large-scale constellation earth observation task planning method provided by the embodiments of the present application firstly optimizes and solves a preset large-scale constellation earth observation task planning model based on an improved adaptive large neighborhood search algorithm with tabu search to allocate the large-scale constellation earth observation tasks to multiple star clusters to obtain a target star cluster allocation scheme; then, for any star cluster, a star cluster autonomous task planning algorithm based on an extended contract net mechanism is used to allocate the tasks in a to-be-planned task set to the satellites in the star cluster to obtain a star cluster task planning scheme of the star cluster; and finally, a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation task is generated based on the star cluster task planning schemes corresponding to the multiple star clusters. In this way, on the one hand, by dividing the large-scale constellation earth observation task problem into a multi-star cluster allocation problem and a single-star cluster autonomous task planning problem, the dimensionality of the problem can be reduced, and the search space can be greatly reduced to improve the solving efficiency; on the other hand, by allocating a large number of tasks to different star cluster groups and performing collaborative autonomous task planning in the star cluster, the collaborative ability between the satellites in the star cluster can be improved, especially in the case of satellite damage and interference, the anti-interference ability and collaborative completion efficiency of the tasks can be improved, and the robustness and reliability of the task planning can be ensured. BRIEF DESCRIPTION OF DRAWINGS

[0096] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only need to be some embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort.

[0097] Figure 1 is a step flow chart of a large-scale constellation earth observation task planning method in an embodiment of the present application.

[0098] Figure 2 is a planning solution framework diagram of large-scale constellation resource earth observation tasks in an embodiment of the present application.

[0099] Figure 3 is a working principle schematic diagram of a minimum yield deletion operator in an embodiment of the present application.

[0100] Figure 4 is a working principle schematic diagram of a maximum imaging opportunity deletion operator in an embodiment of the present application.

[0101] Figure 5 is a working principle schematic diagram of a maximum time window conflict deletion operator in an embodiment of the present application.

[0102] Figure 6 is a task planning schematic diagram based on a contract net mechanism in an embodiment of the present application.

[0103] Figure 7 is a module schematic diagram of a large-scale constellation earth observation task planning device in an embodiment of the present application. DETAILED DESCRIPTION

[0104] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without any creative effort belong to the scope of protection of the present application.

[0105] It should be noted that, with the development of satellite technology, the number and observation capability of earth observation satellites are continuously improved. How to efficiently arrange these satellite resources to meet the growing demand of users has become an important research topic. Large-scale constellation earth observation task planning problems have been studied by a large number of scholars. The existing solution algorithms are mainly as follows:

[0106] (1) Exact algorithm: Exact algorithm is a kind of algorithm for finding the exact solution of a problem, such as branch and bound algorithm, dynamic programming algorithm, etc. It is mainly used to solve the problem of small number of satellites and small scale of tasks.

[0107] (2) Meta-heuristic algorithm: Meta-heuristic algorithm is based on the optimization mechanism in nature and social phenomena, as well as the combination of random algorithm and local search algorithm, mainly including genetic algorithm, annealing simulation algorithm, large neighborhood search algorithm, particle swarm optimization algorithm.

[0108] (3) Heuristic algorithm: Heuristic algorithm mainly realizes the solution of the problem based on certain rules, such as tabu search, greedy search.

[0109] For large-scale constellation earth observation task planning problem, the existing technologies usually have the following shortcomings:

[0110] (1) Difficulty in solving large-scale constellation earth observation task planning problem. With the increasing number of satellites and tasks, the complexity of large-scale constellation task planning under massive demand is high, which makes the solving time of existing technologies increase exponentially, and it is impossible to solve in a short time, which does not meet the requirements of lightweight deployment. At the same time, the nonlinearity of constraint conditions and objective function, and the continuous expansion of problem scale make the constraint conditions and objective function more complex, which makes the existing technologies unable to apply to large-scale constellation earth observation task planning problem.

[0111] (2) Insufficient satellite autonomous cooperation and anti-disturbance ability under centralized planning mode. The existing on-board autonomous task planning method usually assigns tasks on the ground, and then uploads the tasks to a single satellite for autonomous task planning. This solving method cannot be applied to the scene of satellite damage and interference. When the satellite is damaged and interfered, the task cannot be effectively cooperated and completed, which leads to poor robustness and reliability of task planning.

[0112] In view of the problems of solving difficulty and insufficient satellite anti-disturbance ability of related technologies in the face of large-scale constellation earth observation task planning problem, the present application aims to provide a large-scale constellation earth observation task planning method. On the one hand, by dividing the large-scale constellation earth observation task problem into multi-star cluster allocation problem and single-star cluster autonomous task planning problem, the dimensionality of the problem can be reduced, and the search space can be greatly reduced, thereby improving the solving efficiency. On the other hand, by allocating massive tasks to different star clusters and performing cooperative autonomous task planning in the star cluster, the cooperation ability between satellites in the star cluster can be improved, especially in the scene of satellite damage and interference, the anti-disturbance ability and cooperative completion efficiency of the task can be improved, and the robustness and reliability of the task planning can be ensured.

[0113] Reference Figure 1The application discloses a large-scale constellation earth observation task planning method, which can comprise the following steps:

[0114] S101: An improved adaptive large neighborhood search algorithm based on fusion tabu search is used to optimize and solve a preset large-scale constellation earth observation task planning model, so that large-scale constellation earth observation tasks are allocated to multiple star clusters, and a target star cluster allocation scheme is obtained.

[0115] In the embodiment, a large-scale constellation earth observation task planning model is first established based on agile satellite task scheduling analysis and constraint conditions.

[0116] It should be noted that in the multi-agile satellite earth observation task planning, the scene of the agile satellite in the actual observation activity is very complex, and some constraint conditions and scenes cannot be modeled. Therefore, in order to simplify the scene and solve the difficulty, the large-scale constellation earth observation task planning model can be constructed based on the following assumptions:

[0117] (1) It is assumed that all tasks are point targets, that is, a task can be completely covered by a satellite, and the observation can be completed within a visible time window.

[0118] (2) It is assumed that all tasks are observed once, and there is no need for repeated observation.

[0119] (3) It is assumed that all tasks meet the optical imaging requirements, and factors such as cloud cover, equipment failure and electromagnetic interference are not considered.

[0120] (4) The differences in equipment and operation between different satellites are ignored, and it is assumed that the storage space and total power of all satellites are the same.

[0121] (5) It is assumed that the data transmission process is not considered in the observation task scheduling.

[0122] In the embodiment, for the convenience of subsequent model description and establishment, the variables appearing in the model are defined and symbolically explained, and the symbolic explanation is shown in Table 1.

[0123] Table 1 Symbolic explanation

[0124]

[0125]

[0126] In the embodiment, the large-scale constellation earth observation task planning model comprises a target function for maximizing the total revenue of observation tasks. Specifically, the target function is shown in formula (1):

[0127]

[0128] Wherein, pi Let T represent the reward value for task i; T represents the set of imaging tasks; S represents the set of satellites; O g Represents the set of orbital resources of a satellite; Represents a binary decision variable, when At that time, mission i is observed on the k-th orbit of the satellite, if Then task i was not observed in the kth orbit of the satellite.

[0129] In this embodiment, the large-scale constellation Earth observation mission planning model also includes a set of constraints; wherein, the set of constraints includes:

[0130] (1) Single-track fixed-storage constraint conditions.

[0131] The single-orbit storage constraint is used to ensure that the storage space required for all tasks of observation in each orbit of each satellite must be less than or equal to the storage capacity limit of each orbit of that satellite. Specifically, the single-orbit storage constraint is shown in formula (2):

[0132]

[0133] It should be noted that the storage space required for all tasks of each satellite's observation in each orbit must be less than or equal to the fixed storage capacity limit of each orbit of that satellite. If, during the execution of observation activities, the sum of the storage space occupied by the tasks already observed in that orbit is close to the fixed storage threshold of that orbit, then all other tasks that have been allocated to that orbit cannot be observed.

[0134] (2) Single-track energy constraint conditions.

[0135] The single-orbit energy constraint is used to ensure that the energy consumed by all missions for each orbit observation of each satellite is less than or equal to the energy limit of each orbit of that satellite. Specifically, the single-orbit energy constraint is shown in formula (3):

[0136]

[0137] It should be noted that the energy consumed by all tasks observing each orbit of each satellite must be less than or equal to the energy limit of each orbit of that satellite. If, during the execution of observation activities, the sum of the energy consumed by the tasks already observed in that orbit is close to the energy threshold of that orbit, then all other tasks that have been assigned to that orbit cannot be observed.

[0138] (3) Task execution count constraint. The task execution count constraint is used to ensure that each task is considered complete only if it successfully observes once on a single orbit. Specifically, the task execution count constraint is shown in formula (4):

[0139]

[0140] It should be noted that each task needs to be observed only once on a track to be considered complete observation, and does not need to be observed repeatedly.

[0141] (4) Task uniqueness constraint condition.

[0142] The task uniqueness constraint condition is used to constrain each satellite to perform only one task at the same time. Specifically, the task uniqueness constraint condition is as shown in formula (5):

[0143]

[0144] It should be noted that formula (5) ensures that each task except the first task and the last task has only one preceding task and only one subsequent task at the same time.

[0145] (5) Task completion time constraint condition.

[0146] The task completion time constraint condition is used to constrain each task to be completed before the preset task completion time. Specifically, the task completion time constraint condition is as shown in formula (6):

[0147]

[0148] It should be noted that when the user has specific time limit requirements for some tasks, the task needs to be completed before the time limit required by the task, and therefore, a time window later than the required completion time is invalid.

[0149] (6) Observation start and end time constraint condition.

[0150] The observation start and end time constraint condition is used to constrain the actual observation start time and the actual observation end time of all tasks to be within the observable time window of the satellite. Specifically, the observation start and end time constraint condition is as shown in formula (7):

[0151]

[0152] It should be noted that the actual observation start time and the actual observation end time of all tasks are within the observable time window of the satellite for the target, and observation activities cannot be performed outside the observable time window.

[0153] (7) Satellite attitude adjustment constraint condition.

[0154] The satellite attitude adjustment constraint condition is used to constrain the time length of the satellite for attitude adjustment and the time length of the satellite payload switch-off must be less than the time interval between the start time of the next task after the current task and the completion time of the observation activity of the current task. Specifically, the satellite attitude adjustment constraint condition is shown in formula (8):

[0155]

[0156] It should be noted that, unlike the traditional earth observation satellite, the agile satellite can complete the observation activity in the maneuvering state and has the ability of "non-steady imaging" and does not need attitude stabilization time. When calculating the attitude adjustment amplitude of the agile satellite, the free degree with the longest maneuvering time of pitch and roll is taken as the attitude adjustment time. The next task after the current task means the next task closest to the current task.

[0157] It should be noted that the parameter meanings of formulas (2)-(8) refer to Table 1 and will not be repeated here.

[0158] In the embodiment, after the large-scale constellation earth observation task planning model is constructed, the improved adaptive large neighborhood search algorithm fused with the tabu search is used to optimize and solve the large-scale constellation earth observation task planning model based on the objective function and the constraint condition set, so as to allocate the large-scale constellation earth observation task to multiple star clusters and obtain a target star cluster allocation scheme. The target star cluster allocation scheme represents that multiple star clusters are allocated with respective corresponding task set to be planned.

[0159] In the embodiment, the star cluster represents a set of n satellites with similar spatial distances in the same orbit plane, and all the satellites in the same star cluster share the same task information, where n is a positive integer greater than 1. It should be noted that the number of star clusters can be set based on actual needs, for example, a group of satellites in the same orbit plane can be combined into a star cluster.

[0160] It should be noted that the adaptive large neighborhood search algorithm (ALNS) can effectively jump out of the local optimal solution, explore a larger range of solution space, and the adaptive operator update can also adjust the search strategy according to the current search process. However, the traditional adaptive large neighborhood search algorithm may also repeatedly search for recent solutions, resulting in reduced search efficiency. By fusing the tabu search algorithm (TS), the algorithm can well avoid repeatedly searching for recent solutions and greatly improve the search efficiency. Therefore, the embodiment adds a tabu search strategy to the adaptive large neighborhood algorithm, and further proposes an improved adaptive large neighborhood search algorithm fused with the tabu search to realize the allocation of large-scale constellation observation tasks among star clusters, which can effectively improve the search efficiency and search quality of the solution.

[0161] S102: For any star cluster, based on the star cluster autonomous task planning algorithm of the extended contract network mechanism, the tasks in the task set to be planned are allocated to the satellites in the star cluster to obtain a star cluster task planning scheme of the star cluster.

[0162] It should be noted that the contract network mechanism (ContractNetMechanism) is a protocol for task allocation and resource management, mainly used for task allocation and production scheduling in distributed systems. It enables the entire system to complete task allocation at a lower cost and higher quality through the "tender-bid-winning" method.

[0163] In this embodiment, in order to solve the problem that the traditional method cannot be applied to the satellite damage and interference scenario, the present embodiment proposes a star cluster autonomous task planning algorithm based on an extended contract network mechanism, which can allocate a large number of tasks to different star cluster groups, and perform multi-satellite distributed collaborative autonomous task planning within the star cluster. This method can share satellite resources within the same star cluster group, and when some satellites within the same star cluster group cannot work normally, other satellites can cooperate to complete task planning. Compared with the traditional method of task allocation on the ground and then uploading tasks to a single satellite for autonomous task planning, the present embodiment can fully adapt to the satellite damage and interference scenario, and when the satellite is damaged and interfered, the task can be completed by other satellites in cooperation, thereby ensuring the robustness and reliability of task planning.

[0164] S103: Based on the star cluster task planning scheme corresponding to each of the plurality of star clusters, a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation task is generated.

[0165] In this embodiment, referring to Figure 2 , a large-scale constellation resource earth observation task planning solution framework diagram is shown. The framework is divided into an upper layer (task allocation layer) and a lower layer (star cluster multi-satellite autonomous task planning layer). The upper layer is a task allocation problem, and the lower layer is a multi-satellite task planning problem within a star cluster. The upper layer uses an improved adaptive large neighborhood search algorithm based on fusion tabu search to implement task allocation to different star clusters, and divides the large-scale constellation earth observation task planning problem into a single-star cluster earth observation task planning sub-problem; for the single-star cluster earth observation task planning sub-problem, the lower layer uses a star cluster autonomous task planning algorithm based on an extended contract network mechanism to implement multi-satellite task planning within the star cluster. Each star cluster corresponds to a sub-problem, and the star cluster task planning scheme of each star cluster is the solution to the corresponding sub-problem. Combining the solutions of all sub-problems, the solution to the original problem, i.e., the large-scale constellation earth observation task planning scheme, can be obtained.

[0166] The method for large-scale constellation earth observation task planning provided in the embodiments of the present application can reduce the dimension of the problem by dividing the large-scale constellation earth observation task problem into a multi-star cluster group allocation problem and a single-star cluster autonomous task planning problem, thereby greatly reducing the search space and improving the solving efficiency. On the other hand, by allocating a large number of tasks to different star clusters and performing collaborative autonomous task planning in the star clusters, the collaborative capability between satellites in the star clusters can be improved, especially in the case of satellite damage and interference, the anti-interference capability and collaborative completion efficiency of the task can be improved, and the robustness and reliability of the task planning can be ensured.

[0167] In a feasible implementation, S101 can specifically include the following sub-steps:

[0168] S101-1: generate an initial solution of the algorithm, and take the initial solution as a current solution and a current optimal solution.

[0169] In the embodiments, all algorithm parameters of the algorithm can be initialized first, and then the initial solution of the algorithm is generated based on the initialized algorithm parameters. The algorithm refers to an improved adaptive large neighborhood search algorithm based on fusion tabu search; and the initial solution refers to an initial allocation scheme.

[0170] In the embodiments, it is considered that in the search process, unconditionally accepting the optimal solution is easy to fall into local optimization. In order to jump out of the local optimization, in the algorithm search process, the inferior solution will be accepted with a certain probability. Therefore, the simulated annealing algorithm can be used for solving, so as to accept the superior solution in the process of temperature annealing and neighborhood search, and accept the inferior solution with a certain probability according to the Metropolis criterion (a criterion for judging whether a new state is accepted), and at the same time, with the decrease of the temperature, the acceptance probability of the inferior solution is also decreasing.

[0171] Specifically, the algorithm parameters can specifically include an initial temperature T0, a minimum temperature T min , a temperature cooling rate τ, an initial weight ρ and a score w of the operator, and a maximum iteration number Iter max .

[0172] In the embodiments, it is considered that the probability of accepting the inferior solution is related to the initial temperature setting. If the initial temperature setting is too high, the probability of accepting the inferior solution will be too low, and the local optimization cannot be jumped out. If the initial temperature setting is too low, the probability of accepting the inferior solution will be too high, and the quality of the solution will be poor. Therefore, the initial temperature T init can be calculated according to the return value of the initial allocation scheme, and the calculation formula is as follows:

[0173]

[0174] Wherein, f(s init) represents a total revenue value of the initial allocation scheme, the total revenue value is calculated according to formula (1); T init is an initial temperature.

[0175] In the embodiment, after obtaining the initial solution, the initial solution can be taken as the current solution and the current optimal solution. The current solution and the current optimal solution will be updated in the process of subsequent iterative solving.

[0176] S101-2: Repeat the iterative solving step until the algorithm meets the preset iteration stopping condition.

[0177] In the embodiment, the iterative solving step is: determining a target destruction operator and a target repair operator in the preset at least one destruction operator and at least one repair operator respectively, and deleting a preset number of scheduled tasks from the current solution by the target destruction operator and inserting a preset number of unscheduled tasks by the target repair operator based on the taboo attribute of each task to generate a new solution; updating the current solution and the current optimal solution based on the new solution.

[0178] It should be noted that the adaptive large neighborhood search algorithm reconstructs the solution by dynamically selecting the destruction operator and the repair operator, and the performance of the operator directly determines the performance of the adaptive large neighborhood search algorithm.

[0179] In the embodiment, in order to ensure the effect of the adaptive large neighborhood search algorithm, a plurality of destruction operators and a plurality of repair operators with different functions are provided. After the target repair operator is executed to obtain a destroyed solution, the destroyed solution can be repaired to a feasible solution by selecting the target repair operator, and then a new solution can be generated in each iteration solving, the current solution and the current optimal solution are updated by the new solution, and the current optimal solution of maximizing the total revenue of the observation task can be output when the algorithm meets the preset iteration stopping condition.

[0180] In the embodiment, the current solution and the current optimal solution are updated based on the new solution, specifically including: in the case that the revenue of the new solution s n is greater than the revenue of the current solution s c and greater than the revenue of the current optimal solution s b , the new solution s n is updated as the current optimal solution s b and the current solution s c ; in the case that the revenue of the new solution s n is greater than the revenue of the current solution s c and less than the revenue of the current optimal solution s b , the new solution s n is updated as the current solution s c ; in the case that the revenue of the new solution s n is less than the revenue of the current solution s cthe yield of the new solution s c and the yield of the current solution s c , and the current annealing temperature, to determine a target probability, and based on the target probability, to determine whether to accept the new solution as the current solution.

[0181] In a specific implementation, if the yield of the new solution s n is not higher than the yield of the current solution s c , it can be determined whether to accept the new solution by using a Metropolis acceptance criterion, and the Metropolis acceptance criterion works as follows:

[0182] The Metropolis acceptance criterion is used to determine whether to accept a neighborhood solution (i.e., a new solution) as a current solution. In a search process, unconditionally accepting a new solution can easily fall into a local optimum. In order to jump out of the local optimum, a poor solution is accepted with a certain probability in the algorithm search process. However, as the search time increases and the search range expands, the probability of accepting a poor solution should gradually decrease. Since the simulated annealing algorithm can accept both a good solution and a poor solution according to the Metropolis criterion with a target probability during the cooling annealing and neighborhood search processes, and as the temperature decreases, the acceptance probability of a poor solution also decreases. The calculation formula of the target probability is as follows:

[0183]

[0184] wherein p(s n ) represents the target probability of the current solution s c ; f(s c ) and f(s n ) are the yield values of the current solution s c and the new solution s n respectively; and T is the current annealing temperature.

[0185] S101-3: In a case where it is determined that the algorithm satisfies an iteration stop condition, the current optimal solution is taken as the target star cluster allocation scheme.

[0186] In this embodiment, the iteration stop condition can be any one of a first iteration stop condition and a second iteration stop condition. The first iteration stop condition is that the number of iterations of the algorithm reaches a maximum iteration number Iter max . The second iteration stop condition is that the temperature of the algorithm is annealed to a minimum temperature T min .

[0187] In this embodiment, based on a simple and intuitive principle, the annealing process can adopt a linear annealing manner, and the temperature update is as shown in the formula:

[0188] T n+1 = τT n (11).

[0189] wherein τ represents the temperature cooling rate; T n represents the annealing temperature of the current round; T n+1 represents the annealing temperature of the next round of the current round.

[0190] In the present embodiment, if the number of iterative solutions of the algorithm does not reach the maximum number of iterations Iter max , or the annealing temperature of the algorithm does not reach the minimum temperature T min , the iterative solution step is repeated; otherwise, the current optimal solution is output as the target star cluster allocation scheme.

[0191] In a feasible embodiment, the step of determining the target destruction operator and the target repair operator in the preset at least one destruction operator and at least one repair operator in S101-2 can specifically include the following sub-steps:

[0192] S101-2-1: determining the target destruction operator from the at least one destruction operator based on the respective current operator weight of each of the at least one destruction operator.

[0193] In the present embodiment, the at least one destruction operator includes one or more of a random deletion operator, a minimum revenue deletion operator, a maximum attitude adjustment conflict deletion operator, a maximum imaging opportunity deletion operator, and a maximum time window conflict deletion operator.

[0194] In the present embodiment, in order to ensure the quality of the solution, the above-mentioned five destruction operators can be configured at the same time.

[0195] Random deletion operator: randomly deleting n scheduled tasks on the orbit, and putting the deleted tasks into the unscheduled task set. If the number of scheduled tasks on the orbit is less than n, deleting all scheduled tasks on the orbit. The random deletion operator introduces randomness into the search process, reducing the risk of falling into a local optimum.

[0196] Minimum revenue deletion operator: preferentially deleting n scheduled tasks with the smallest revenue on the orbit, and putting the deleted tasks into the unscheduled task set. The minimum revenue deletion operator selects to delete tasks with smaller contribution to the current scheduling scheme, i.e., tasks with smaller revenue, minimizes the destruction to the current solution, and controls the search direction to be in a direction conducive to optimizing the scheduling scheme. Referring to Figure 3 , a working principle schematic diagram of the minimum revenue deletion operator is shown. There are scheduled tasks i, j, k, and m on the orbit g, with revenue values of 1, 2, 9, and 7, respectively. When the minimum revenue deletion operator is executed, task i will be deleted from the original scheduling scheme, and task i will be added to the unscheduled task set.

[0197] The maximum attitude adjustment conflict removal operator prioritizes removing the n assigned tasks with the highest attitude adjustment conflict on the track and adds the removed tasks to the unscheduled task set. By removing the tasks with the highest attitude adjustment conflict in the current scheduling scheme, the maximum attitude adjustment conflict removal operator improves the quality of the solution and creates better conditions for repairing subsequent solutions.

[0198] Maximum Imaging Opportunity Deletion Operator: Prioritizes deleting the task with the largest total imaging time window (i.e., the nth allocated task on the track), and adds the deleted task to the unscheduled task set. (See reference...) Figure 4 The diagram illustrates the working principle of the maximum imaging opportunity deletion operator. On orbit g, there are scheduled tasks i, j, k, and m. The shadow length represents the observable time window length. According to the principle of the maximum imaging opportunity deletion operator, tasks with the highest imaging opportunity (i.e., the longest observable time window) are prioritized for deletion. Therefore, on orbit g, the maximum imaging opportunity deletion operator will delete task k. The maximum imaging opportunity deletion operator improves the solution quality by deleting tasks with the highest imaging opportunity, allowing more tasks to be processed and increasing task processing efficiency.

[0199] Maximum Time Window Conflict Deletion Operator: Prioritizes deleting tasks with high time window conflict on the track (n already allocated time windows), and adds the deleted tasks to the unscheduled task set. (See reference...) Figure 5 The diagram illustrates the working principle of the maximum time window conflict removal operator. There are scheduled tasks i, j, and m on track g, with observable time window durations of 15s, 10s, and 25s respectively. The conflict time between tasks i and j is 4s, between tasks i and m is 6s, and between tasks j and m is 10s. That is, the conflict coefficient of task j on orbit g. It is 1.4. That is, the conflict coefficient of task m on orbit g. If the value is 0.64, then the maximum time window conflict deletion operator will prioritize deleting task j with the highest conflict coefficient.

[0200] S101-2-1: Based on the current operator weights corresponding to at least one repair operator, determine the target repair operator among at least one repair operator.

[0201] In this embodiment, at least one repair operator includes one or more of the following: maximum gain insertion operator, minimum time window conflict insertion operator, minimum attitude adjustment conflict insertion operator, and minimum imaging opportunity insertion operator.

[0202] In this embodiment, to ensure the quality of the solution, the above four repair operators can be configured simultaneously.

[0203] The maximum benefit insertion operator: preferentially selects n tasks with the highest benefit from the track in the unscheduled task set for insertion. The maximum benefit insertion operator will keep inserting the selected tasks until a feasible observable time window is found. If the task has no feasible observable time window, the maximum benefit insertion operator will give up inserting the task and select the task with the highest benefit other than the task for insertion until a task that can be inserted is found. The maximum benefit insertion operator selects the task with the highest benefit for insertion to the current scheduling scheme to maximize the quality of the solution.

[0204] The minimum time window conflict insertion operator: preferentially selects n tasks with the minimum time window conflict from the track in the unscheduled task set for insertion.

[0205] The minimum attitude adjustment conflict insertion operator: preferentially selects n tasks with the minimum attitude adjustment conflict from the track in the unscheduled task set for insertion.

[0206] The minimum imaging opportunity insertion operator: preferentially selects n tasks with the minimum imaging opportunity from the track in the unscheduled task set for insertion.

[0207] In the embodiment, the greater the operator weight of the operator, the greater the probability that the operator is selected. The probability that the destruction operator or the repair operator is selected is calculated according to the following formula:

[0208]

[0209] wherein: φ i represents the probability that the i-th operator is selected; p i represents the weight of the i-th operator; and Ω represents the number of operators.

[0210] In the embodiment, when the operator is a destruction operator, Ω takes 5, and when the operator is a repair operator, Ω takes 4.

[0211] In the embodiment, the operator score and the operator weight and the taboo attribute value of the taboo attribute of the task can also be updated once every n generation. Considering that the basic idea of the adaptive process is to assign different scores to the destruction operator and the repair operator according to their performance, the operator score and the operator weight can be updated in the following manner:

[0212] For any one of the at least one destruction operator and the at least one repair operator, a new solution s n generated based on the operator is obtained.

[0213] In the case where the new solution s n is updated to the current optimal solution s b , that is, when the new solution s n generated after the operation of the operator is the optimal solution sb a first score w1 is assigned to the operator;

[0214] in the case that the benefit of the new solution s n is greater than the benefit of the current solution s c and less than the benefit of the current optimal solution, i.e., when the new solution s n generated after the operator operates is better than the current solution s c but worse than the current optimal solution s b , a second score w2 is assigned to the operator;

[0215] in the case that the benefit of the new solution s n is less than the benefit of the current solution s c and the new solution s n is accepted as the current solution s c , a third score w3 is assigned to the operator;

[0216] in the case that the benefit of the new solution s n is less than the benefit of the current solution s c and the new solution s n is not accepted as the current solution s c , a fourth score w4 is assigned to the operator; wherein w1 > w2 > w3 > w4.

[0217] The total score sum of the operator in the preset number of iterations is determined every interval, and the current operator weight of the operator is updated based on the total score sum and the historical operator weight of the operator. The total score sum represents the sum of the scores assigned to the operator in the preset number of iterations.

[0218] It should be noted that the higher the total score sum of the operator, the better the solving performance of the operator. The operator weight is updated according to the historical weight and the total score sum of the operator, and the update formula of the operator weight is as follows:

[0219]

[0220] wherein, ρ i represents the weight of the i-th operator; ρ' i represents the historical weight of the i-th operator; n i represents the number of times the i-th operator is called; Q i represents the total score sum of the i-th operator in the preset number of iterations; λ is a weight parameter, ranging from 0 to 1, representing the importance of the operator score sum and the historical operator weight, for example, λ can be 0.5.

[0221] In a feasible implementation, the tabu attribute includes a deletion tabu attribute and an insertion tabu attribute; the large-scale constellation earth observation task planning method can further include the following steps:

[0222] S201: For any task, if the new solution containing the task is accepted as the current solution, or if the new solution after deleting the task is not accepted as the current solution, the deletion tabu attribute of the task is set to the first attribute value.

[0223] In the embodiment, the first attribute value is used to indicate that the task is prohibited from being deleted in the subsequent preset number of iterations of solving.

[0224] S202: For any task, if the new solution after deleting the task is accepted as the current solution, or if the new solution containing the task is not accepted as the current solution, the insertion tabu attribute of the task is set to the second attribute value.

[0225] In the embodiment, the second attribute value is used to indicate that the task is prohibited from being inserted in the subsequent preset number of iterations of solving.

[0226] It should be noted that in the tabu search algorithm, some simple local search operations are usually defined for solution updating, in order to prevent short-term cycles, the recently accessed solutions or the recently used local search operations are stored in a tabu list, the solutions or operations in the list will not be selected again, and the tabu solutions or operations are cancelled after a period of time or a number of iterations.

[0227] In the embodiment, a close hybrid strategy of ALNS (adaptive large neighborhood search algorithm) and TS (tabu search algorithm) is proposed, which sets two tabu attributes for the task: deletion tabu attribute and insertion tabu attribute. That is, for each task in the large-scale constellation earth observation task, a deletion tabu attribute and an insertion tabu attribute will be generated.

[0228] The two tabu attributes are used in the solution updating phase. The deletion tabu attribute and the insertion tabu attribute of each task are updated every n iterations. For the determination of n value, n can be set as a random number in (1, m), where m is the number of tasks.

[0229] In the embodiment, for the large-scale constellation observation task allocation problem, an improved adaptive large neighborhood search algorithm combined with tabu search is proposed, which can realize the effective allocation of large-scale constellation observation tasks in multiple star clusters. The adaptive large neighborhood search algorithm (ALNS) can effectively jump out of the local optimal solution, explore a larger range of solution space, and the adaptive operator update can also adjust the search strategy according to the current search process. At the same time, combined with the tabu search strategy, it can well avoid the repeated search of recent solutions, and greatly improve the search efficiency.

[0230] In one possible implementation, S102 can specifically include the following sub-steps:

[0231] S102-1: For any star cluster, determine the master star and the slave stars of the star cluster.

[0232] It should be noted that the contract net mechanism is an effective method for solving distributed task planning, and the planning of tasks is completed by simulating the "bidding-tender-evaluation" process in the market mechanism.

[0233] Specifically, when the contract net protocol is used to solve the problem of distributed satellite task planning, the satellites can be divided into master stars and slave stars to act as bidders and tenderers in the bidding process, respectively. Any star in the star cluster can be a master star, and other satellites are slave stars. For example, the satellite that receives the task information in the star cluster can be the master star of the star cluster.

[0234] In the present embodiment, reference is made to Figure 6 , which shows a task planning schematic diagram based on the contract net mechanism. First, the master star broadcasts the tasks in the task set to be planned to the slave stars for bidding as tender information; then, each slave star autonomously plans according to the information information after receiving the tender information, generates a bidding scheme for bidding; finally, the master star evaluates all the returned bidding schemes, determines the winning scheme, and broadcasts the winning information to the slave stars. The whole planning process needs multiple bidding, and finally outputs a complete task planning scheme.

[0235] In the present embodiment, by introducing the contract net mechanism, the star cluster autonomous task planning algorithm based on the extended contract net mechanism can realize the star cluster autonomous task planning.

[0236] S102-2: Publish the tasks in the task set to be planned to the slave stars for bidding by the master star.

[0237] In the present embodiment, the traditional contract net protocol adopts the "single task bidding" mode, and the master star and the slave star perform "bidding-tender-evaluation-winner" for each task in the task set T to determine the winning slave star of each task. This mode needs constant interstellar negotiation, and the communication volume is huge in the large-scale task scenario. Therefore, the present embodiment will adopt the centralized bidding strategy, i.e., the master star will conduct centralized bidding for all the tasks in T in each bidding process to reduce the negotiation times in the bidding process, and thus reduce the interstellar communication volume.

[0238] S102-3: Generate a bidding scheme by each slave star, and deliver the bidding scheme to the master star.

[0239] In the present embodiment, each slave star can generate a bidding scheme based on the heuristic algorithm of the immediately preceding arrangement strategy, and deliver the bidding scheme to the master star.

[0240] Specifically, the heuristic algorithm based on the immediate precedence strategy sorts the tasks according to the start time of the visible time window, and then plans as many tasks as possible into the bidding scheme according to the constraint conditions. Each task in the bidding scheme adopts the immediate precedence strategy, that is, each task starts as early as possible.

[0241] S102-4: Determine the task planning scheme corresponding to the star cluster based on the bidding scheme of each slave star by the master star.

[0242] In this embodiment, the master star receives the bidding scheme of each slave star, directly bids for the tasks without repeated planning, and needs to bid for the tasks with repeated planning to determine the bidding scheme.

[0243] It should be noted that under the centralized bidding strategy, each slave star may bid for multiple tasks, and there may be different tasks or the same tasks in the bidding schemes of multiple slave stars, and it is difficult to reasonably determine the bidding slave star for each task by using the single-agent bidding strategy. Therefore, the task group bidding strategy is proposed on the basis of the centralized bidding strategy.

[0244] Specifically, the task group bidding strategy includes: in the case that the bidding task of any slave star is a non-repeated planning task, determining the bidding scheme of the slave star as the bidding scheme; in the case that the bidding task of any slave star is a repeated planning task, bidding for the repeated planning task, and based on the bidding result, correcting the bidding scheme of the slave star to obtain the bidding scheme of the slave star. Wherein, the non-repeated planning task refers to a task planned by only one slave star; the repeated planning task refers to a task planned by at least two slave stars.

[0245] In this embodiment, based on the task group bidding strategy, the master star only needs to bid for the repeated planning task, and the non-repeated planning task is directly bid by the bidding slave star. In this way, it can not only ensure that the time window resources are not wasted due to repeated planning tasks, but also fully utilize the computing resources of the slave stars, avoiding the situation that most slave stars fail to bid in the "single-agent bidding" mode and the computing resources are wasted.

[0246] In this embodiment, based on the improved contract net protocol, the master star can correct the bidding scheme of the slave star according to the bidding result, and the slave star will accept the bidding scheme corrected by the master star based on its bidding scheme.

[0247] In this embodiment, for the repeated planning task, the first bidding strategy and the second bidding strategy can be used for planning. The first bidding strategy is to preferentially select the bidding scheme with the earliest execution time to bid. The second bidding strategy is to preferentially select the bidding scheme with the least planning quantity to bid. Wherein, the priority of the first bidding strategy is higher than that of the second bidding strategy.

[0248] In one possible implementation, S102-3 can specifically include the following sub-steps:

[0249] S102-3-1: For any slave star, sort the tasks published by the master star in descending order of priority to obtain an observation task set.

[0250] In this embodiment, after the slave star receives the task set T to be planned of the master star, the tasks are sorted in descending order of priority to obtain a new observation task set New_OT.

[0251] S102-3-2: In order of priority from high to low, the tasks in the observation task set are sequentially planned. Task planning includes: in the case where there is at least one observable time window that meets the task requirements of the task in the current observable time window set of the slave star, the earliest observable time window is determined as the target window of the task; the task is planned into the task planning set, and the current observable time window is updated based on the target window.

[0252] In this embodiment, for each task planning, let the first task with the highest priority in the observation task set New_OT be g i , find all visible time windows time_win in the current observable time window set VAR of the slave star. Then, from time_win, select the observable time window remain_win that meets the task requirements of the observation task g i , which can include resolution requirements, payload type requirements, imaging deadline requirements, etc. Considering that there is more than one observable time window that meets the task requirements, in order to ensure that high-priority tasks can be completed first, the earliest observable time window is determined as the target window of the task g i . Finally, the task g i is planned into the task planning set NP, and the target window occupied by the task g i is deleted from the current observable time window set VAR to obtain the updated current observable time window, and then the next task g i in the observation task set New_OT is planned based on the updated current observable time window. In this way, by repeatedly repeating the above process, the final task planning set can be obtained.

[0253] It should be noted that the execution start time and the execution end time of the current task g i need to meet the constraints of formulas (2)-(8).

[0254] In a specific implementation, the execution start time and the execution end time of the current task can be calculated according to the preceding task based on the following formula:

[0255]

[0256]

[0257] wherein, denotes the execution start time of the current task g i ; denotes the execution end time of the current task g i ; denotes the start time of the w i th observable time window of the current task g i ; denotes the observation end time of the preceding task pre i ; denotes the pose conversion time between the preceding task pre i and the current task g i ; i denotes the duration of the task g i .

[0258] S102-3-3: In the case where the task planning of the last task in the observation task set is completed, a bidding scheme is obtained based on the task planning set.

[0259] In the present embodiment, the satellite cluster can obtain the final task planning set after repeatedly performing the above task planning process and completing the task planning of the last task in the observation task set, and finally obtain the bidding scheme of the satellite cluster according to the tasks that have been planned in the task planning set.

[0260] In the present embodiment, for the multi-satellite task planning problem in a satellite cluster, the present embodiment proposes a satellite cluster autonomous task planning algorithm based on an extended contract net mechanism. In the bidding process, a centralized bidding strategy is adopted in the all-task set, and a task group bidding strategy is adopted in the multiple bidders. Compared with the traditional single bidding mode, the centralized bidding strategy can greatly reduce the inter-satellite communication. The task group bidding strategy can not only ensure that the time window resources are not wasted due to repeated planning of tasks, but also fully utilize the computing resources of the slave satellites, thereby avoiding the situation that most slave satellites fail to bid and computing resources are wasted in the "single agent bidding" mode.

[0261] The large-scale constellation earth observation task planning method (hereinafter referred to as ALNS-CN) provided by the embodiment of the application can avoid repeated search and repeated solution, and has higher search efficiency. Compared with the traditional GA (Genetic Algorithm) and VNS (Variable Neighborhood Search), the ALNS-CN can be superior to other traditional methods in terms of solution quality and solution efficiency.

[0262] In order to verify the superiority of the embodiment of the application in terms of solution quality and solution efficiency, experiments and simulations are performed in two scenarios of global uniform mode and regional local concentration mode. The simulation experiment scene is designed as follows.

[0263] Scenario A: the task size is 1500, the number of satellites is 500, and the tasks are randomly generated in the global range.

[0264] Scenario B: the task size is 1500, the number of satellites is 500, and the tasks are randomly generated in the range of 21.85°-25.33° north latitude and 119.91°-122.25° east longitude.

[0265] 500 satellites are designed to form 20 orbital planes, and each orbital plane has 25 satellites. The simulation scene time is 24 hours, and the CPU in the experimental environment is Intel(R) Core(TM) i5-12600K.

[0266] As shown in Table 2, the comparison results of different algorithms in different scenarios are shown.

[0267] Table 2 Comparison results of different algorithms

[0268]

[0269] According to the experimental simulation results, the embodiment of the application has higher superiority in terms of solution quality and solution efficiency compared with the traditional algorithm.

[0270] When facing partial satellite damage or interference, other satellites can cooperate to complete the task planning of the damaged satellites, and can play the cooperative advantage of the star cluster. Compared with the traditional method of task allocation on the ground and then uploading the task to a single satellite for autonomous task planning, the problem that the task on the damaged satellite cannot be effectively executed when some satellites are damaged can be effectively avoided.

[0271] In order to verify that the embodiment of the application can adapt to the scene of partial satellite damage and interference compared with the traditional method, the allocation result of a star cluster is selected, and 1-4 satellites are randomly selected for simulation experiment. As shown in Table 3, the experimental simulation results of damaging different number of satellites are shown.

[0272] Table 3 Damage test experiment results

[0273] Number Number of satellites destroyed Number of missions Number of missions completed Mission completion rate A1 1 14 14 100％ A2 2 14 14 100％ A3 3 14 14 100％ A4 4 14 12 85.7％

[0274] According to the experimental simulation result, it can be known that the embodiment of the application supports a certain number of satellite damages, and when a satellite is damaged, other satellites can cooperatively complete task planning.

[0275] In a second aspect, referring to Figure 7 The embodiment of the application provides a large-scale constellation earth observation task planning device 700, which comprises:

[0276] The upper-layer task allocation module 701 is configured to perform optimization solving on a preset large-scale constellation earth observation task planning model based on an improved adaptive large neighborhood search algorithm of fusion tabu search, so as to allocate the large-scale constellation earth observation task to a plurality of star clusters and obtain a target star cluster allocation scheme; wherein the target star cluster allocation scheme represents that the plurality of star clusters are allocated with respective corresponding task set to be planned.

[0277] The lower-layer task allocation module 702 is configured to, for any star cluster, allocate tasks in the task set to be planned to satellites in the star cluster based on a star cluster autonomous task planning algorithm of an extended contract net mechanism, and obtain a star cluster task planning scheme of the star cluster.

[0278] The planning scheme generation module 703 is configured to generate a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation task based on the respective corresponding star cluster task planning schemes of the plurality of star clusters.

[0279] In an embodiment of the application, the large-scale constellation earth observation task planning model comprises a target function and a constraint condition set; wherein,

[0280] The target function aims to maximize the total revenue of observation tasks.

[0281] The constraint condition set comprises:

[0282] A single-orbit fixed storage constraint condition is configured to constrain that the storage space required by all tasks observed by each orbit of each satellite must be less than or equal to the fixed storage capacity limit of each orbit of the satellite.

[0283] A single-orbit energy constraint condition is configured to constrain that the energy consumed by all tasks observed by each orbit of each satellite must be less than or equal to the energy limit of each orbit of the satellite.

[0284] A task execution times constraint condition is configured to constrain that each task needs to be observed only once on one orbit to be considered as completing observation.

[0285] a task uniqueness constraint condition for constraining each satellite to perform only one task at the same time;

[0286] a task completion time constraint condition for constraining each task to be completed before a preset task completion time;

[0287] an observation start and end time constraint condition for constraining actual observation start time and actual observation end time of all tasks to be within an observable time window of the satellite;

[0288] a satellite attitude adjustment constraint condition for constraining a length of time for which the satellite is used for attitude adjustment and a length of time for which the satellite payload is switched on to be less than a time interval between a start time of a task immediately following a current task and an observation activity completion time of the current task.

[0289] In an embodiment of the present application, the upper-layer task allocation module 701 comprises:

[0290] an initial solution generation submodule for generating an initial solution of the algorithm and taking the initial solution as a current solution and a current optimal solution;

[0291] an iterative solution submodule for repeatedly performing an iterative solution step until the algorithm satisfies a preset iteration stopping condition; wherein the iterative solution step comprises: determining a target destruction algorithm and a target repair algorithm from at least one destruction algorithm and at least one repair algorithm respectively, and based on the taboo attribute of each task, deleting a preset number of scheduled tasks from the current solution by the target destruction algorithm and inserting a preset number of unscheduled tasks by the target repair algorithm to generate a new solution; and updating the current solution and the current optimal solution based on the new solution;

[0292] a star cluster allocation scheme determination submodule for taking the current optimal solution as a target star cluster allocation scheme when it is determined that the algorithm satisfies the iteration stopping condition.

[0293] In an embodiment of the present application, the iterative solution submodule comprises:

[0294] a destruction algorithm determination unit for determining the target destruction algorithm from the at least one destruction algorithm based on respective current algorithm weights of the at least one destruction algorithm; wherein the at least one destruction algorithm comprises one or more of a random deletion algorithm, a lowest revenue deletion algorithm, a maximum attitude adjustment conflict deletion algorithm, a maximum imaging opportunity deletion algorithm and a maximum time window conflict deletion algorithm;

[0295] The repair operator determination unit is configured to determine a target repair operator from the at least one repair operator based on respective current operator weights of the at least one repair operator; and the at least one repair operator comprises one or more of a maximum benefit insertion operator, a minimum time window conflict insertion operator, a minimum attitude adjustment conflict insertion operator, and a minimum imaging opportunity insertion operator.

[0296] In an embodiment of the present application, the iteration solving submodule comprises:

[0297] The first updating unit is configured to update the new solution as the current optimal solution and the current solution in a case where a benefit of the new solution is greater than a benefit of the current solution and greater than a benefit of the current optimal solution.

[0298] The second updating unit is configured to update the new solution as the current solution in a case where the benefit of the new solution is greater than the benefit of the current solution and less than the benefit of the current optimal solution.

[0299] The third updating unit is configured to determine a target probability based on the benefits of the new solution and the current solution and a current annealing temperature in a case where the benefit of the new solution is less than the benefit of the current solution; and determine whether to accept the new solution as the current solution based on the target probability.

[0300] In an embodiment of the present application, the tabu attribute comprises a deletion tabu attribute and an insertion tabu attribute; and the large-scale constellation earth observation task planning apparatus further comprises:

[0301] The deletion tabu attribute determination module is configured to set a deletion tabu attribute of a task to a first attribute value in a case where a new solution containing the task is accepted as a current solution, or in a case where a new solution after deletion of the task is not accepted as the current solution; and the first attribute value is used to indicate that the task is prohibited from being deleted in subsequent iteration solving for a preset number of times.

[0302] The insertion tabu attribute determination module is configured to set an insertion tabu attribute of a task to a second attribute value in a case where a new solution after deletion of the task is accepted as a current solution, or in a case where a new solution containing the task is not accepted as the current solution; and the second attribute value is used to indicate that the task is prohibited from being inserted in subsequent iteration solving for a preset number of times.

[0303] In an embodiment of the present application, the large-scale constellation earth observation task planning apparatus further comprises:

[0304] The new solution acquisition module is configured to acquire a new solution generated based on any one of the at least one destruction operator and the at least one repair operator;

[0305] The first scoring module is configured to assign a first score to the operator in a case where the new solution is updated as the current optimal solution.

[0306] The second scoring module is configured to assign a second score to the operator in a case where the benefit of the new solution is greater than the benefit of the current solution and less than the benefit of the current optimal solution.

[0307] The third scoring module is configured to assign a third score to the operator in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is accepted as the current solution.

[0308] The fourth scoring module is configured to assign a fourth score to the operator in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is not accepted as the current solution; and the first score, the second score, the third score and the fourth score decrease in turn.

[0309] The operator weight updating module is configured to determine a total score of the operator in the preset number of iterations every interval, and update a current operator weight of the operator based on the total score and a historical operator weight of the operator.

[0310] In an embodiment of the present application, the lower-layer task allocation module 702 comprises:

[0311] The satellite type determination sub-module is configured to determine a master satellite and a slave satellite of the star cluster for any star cluster.

[0312] The bidding sub-module is configured to publish tasks in the task set to be planned to the slave satellites for bidding through the master satellite.

[0313] The bidding sub-module is configured to generate a bidding scheme through each slave satellite and deliver the bidding scheme to the master satellite.

[0314] The scheme determination sub-module is configured to determine a task planning scheme corresponding to the star cluster based on the bidding scheme of each slave satellite through the master satellite.

[0315] In an embodiment of the present application, the bidding sub-module comprises:

[0316] The sorting unit is configured to sort the tasks in descending order based on the priority of the tasks published by the master satellite to obtain an observation task set for any slave satellite.

[0317] The task planning unit is configured to sequentially plan the tasks in the observation task set in order of priority from high to low; the task planning comprises: in a case where there is at least one observable time window that meets the task requirements of the task in the current observable time window set of the slave satellite, determining the observable time window with the earliest time as a target window of the task; planning the task into a task planning set, and updating the current observable time window based on the target window.

[0318] The bidding scheme generation unit is configured to obtain a bidding scheme based on the task planning set in a case where the task planning of the last task in the observation task set is completed. It should be noted that the specific implementation of the large-scale constellation earth observation task planning apparatus 700 in the embodiments of the present application is described with reference to the specific implementation of the large-scale constellation earth observation task planning method in the first aspect of the embodiments of the present application, which will not be described herein again.

[0319] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, apparatus, or computer program product. Therefore, the embodiments of the present application can be in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0320] The embodiments of the present application are described with reference to flowcharts and / or block diagrams according to the method, terminal device (system), and computer program product of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the computer or other programmable data processing terminal device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0321] These computer program instructions can also be stored in a computer-readable memory that can guide the computer or other programmable data processing terminal device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0322] These computer program instructions can also be loaded into a computer or other programmable data processing terminal device, so that a series of operation steps are performed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0323] While preferred embodiments of the application have been described, those skilled in the art will appreciate that other modifications than those specifically described can be made within the scope of the application. Accordingly, the appended claims are intended to embrace all such alternatives as well as the embodiments specifically described.

[0324] Finally, it should be noted that, in the present document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily implying any actual relationship or order between such entities or actions. Also, the terms "comprises", "comprising", or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0325] The above describes in detail the large-scale constellation earth observation task planning method and device provided by the present application, and the principles and implementation manners of the present application are described by using specific examples. The above description of the embodiments is only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges can be changed. In summary, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A method for large-scale constellation earth observation mission planning, characterized in that, The method comprises: An improved adaptive large neighborhood search algorithm based on fusion tabu search is used to optimize and solve a preset large-scale constellation earth observation task planning model, so that large-scale constellation earth observation tasks are allocated to multiple star clusters, and a target star cluster allocation scheme is obtained; wherein the target star cluster allocation scheme represents that multiple star clusters are allocated with respective corresponding to-be-planned task sets; the star cluster represents a set of n satellites with similar spatial distances in the same orbital plane, and all satellites in the same star cluster share the same task information, wherein n is a positive integer greater than 1; For any star cluster, a star cluster autonomous task planning algorithm based on an extended contract net mechanism is used to allocate tasks in the to-be-planned task set to the satellites in the star cluster, and a star cluster task planning scheme of the star cluster is obtained; Based on the respective star cluster task planning schemes of multiple star clusters, a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation tasks is generated; An improved adaptive large neighborhood search algorithm based on fusion tabu search is used to optimize and solve a preset large-scale constellation earth observation task planning model, so that large-scale constellation earth observation tasks are allocated to multiple star clusters, and a target star cluster allocation scheme is obtained, comprising: An initial solution of the algorithm is generated, and the initial solution is taken as a current solution and a current optimal solution; The iteration solving step is repeated until the algorithm meets a preset iteration stopping condition; wherein the iteration solving step is: in at least one destruction operator and at least one repair operator, a target destruction operator and a target repair operator are determined respectively, and based on the tabu attribute of each task, a preset number of scheduled tasks are deleted from the current solution by the target destruction operator, and a preset number of unscheduled tasks are inserted by the target repair operator to generate a new solution; based on the new solution, the current solution and the current optimal solution are updated; In a case where it is determined that the algorithm meets the iteration stopping condition, the current optimal solution is taken as the target star cluster allocation scheme.

2. The method of claim 1, wherein, The large-scale constellation earth observation task planning model comprises a target function and a constraint condition set; wherein The target function aims to maximize the total revenue of observation tasks; The constraint condition set comprises: A single-orbit fixed storage constraint condition is used to constrain that the storage space required by all tasks of each orbit observation of each satellite must be less than or equal to the fixed storage capacity limit of each orbit of the satellite; A single-orbit energy constraint condition is used to constrain that the energy consumed by all tasks of each orbit observation of each satellite must be less than or equal to the energy limit of each orbit of the satellite; A task execution times constraint condition is used to constrain that each task needs to be successfully observed on one orbit only to be considered as completed observation; A task uniqueness constraint condition is used to constrain that each satellite can only execute one task at the same time; A task completion time constraint condition is used to constrain that each task needs to be completed before a preset task completion time; An observation start and end time constraint condition is used to constrain that the actual observation start time and the actual observation end time of all tasks are within the observable time window of the satellite; Satellite attitude adjustment constraint conditions are used to constrain the time length of satellite attitude adjustment and the time length of satellite payload switch-on must be less than the time interval between the start time of the next task of the current task and the completion time of the observation activity of the current task.

3. The method of claim 1, wherein, In the preset at least one destruction operator and at least one repair operator, a target destruction operator and a target repair operator are determined respectively, including: The target destruction operator is determined from the at least one destruction operator based on the current operator weight corresponding to each of the at least one destruction operator; wherein the at least one destruction operator includes one or more of a random deletion operator, a minimum revenue deletion operator, a maximum attitude adjustment conflict deletion operator, a maximum imaging opportunity deletion operator and a maximum time window conflict deletion operator; The target repair operator is determined from the at least one repair operator based on the current operator weight corresponding to each of the at least one repair operator; wherein the at least one repair operator includes one or more of a maximum revenue insertion operator, a minimum time window conflict insertion operator, a minimum attitude adjustment conflict insertion operator and a minimum imaging opportunity insertion operator.

4. The method of claim 1, wherein, Based on the new solution, the current solution and the current optimal solution are updated, including: In the case that the revenue of the new solution is greater than the revenue of the current solution and is greater than the revenue of the current optimal solution, the new solution is updated as the current optimal solution and the current solution; In the case that the revenue of the new solution is greater than the revenue of the current solution and is less than the revenue of the current optimal solution, the new solution is updated as the current solution; In the case that the revenue of the new solution is less than the revenue of the current solution, a target probability is determined based on the revenues of the new solution and the current solution and the current annealing temperature, and whether to accept the new solution as the current solution is determined based on the target probability.

5. The method of claim 1, wherein, The taboo attribute includes a deletion taboo attribute and an insertion taboo attribute; the method further includes: For any task, in the case that a new solution containing the task is accepted as a current solution, or in the case that a new solution after deleting the task is not accepted as a current solution, the deletion taboo attribute of the task is set to a first attribute value; the first attribute value is used to indicate that the task is prohibited from being deleted in subsequent preset number of iterations of solving; For any task, in the case that a new solution after deleting the task is accepted as a current solution, or in the case that a new solution containing the task is not accepted as a current solution, the insertion taboo attribute of the task is set to a second attribute value; the second attribute value is used to indicate that the task is prohibited from being inserted in subsequent preset number of iterations of solving.

6. The method of claim 3, wherein, The method further includes: For any operator in the at least one destruction operator and the at least one repair operator, a new solution generated based on the operator is obtained; In the case that the new solution is updated as the current optimal solution, the operator is given a first score; In the case that the revenue of the new solution is greater than the revenue of the current solution and is less than the revenue of the current optimal solution, the operator is given a second score; and In the case that the revenue of the new solution is less than the revenue of the current solution, the operator is given a third score. in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is accepted as the current solution, a third score is assigned to the operator; in a case where the benefit of the new solution is less than the benefit of the current solution and the new solution is not accepted as the current solution, a fourth score is assigned to the operator; wherein the first score, the second score, the third score and the fourth score decrease in turn; every interval of a preset number of iterations of solving, determining a total score of the operator in the preset number of iterations of solving, and updating the current operator weight of the operator based on the total score and a historical operator weight of the operator.

7. The method of claim 1, wherein, For any of the star clusters, a star cluster autonomous task planning algorithm based on an extended contract net mechanism is used to allocate tasks in the set of tasks to be planned to satellites in the star cluster, to obtain a star cluster task planning scheme of the star cluster, including: For any of the star clusters, a master star and a slave star of the star cluster are determined; the master star publishes tasks in the set of tasks to be planned to the slave star for bidding; each slave star generates a bidding scheme and delivers the bidding scheme to the master star; the master star determines the task planning scheme corresponding to the star cluster based on the bidding scheme of each slave star.

8. The method of claim 7, wherein, each slave star generates a bidding scheme, including: For any slave star, the tasks are sorted in descending order based on the priority of the tasks published by the master star to obtain a set of observation tasks; In order to obtain the bidding scheme, the tasks in the set of observation tasks are sequentially planned in order from high to low according to the priority; the task planning includes: in a case where there is at least one observable time window that meets the task requirements of the task in the set of current observable time windows of the slave star, the earliest observable time window is determined as the target window of the task; the task is planned into a task planning set, and the current observable time window is updated based on the target window; in a case where the task planning of the last task in the set of observation tasks is completed, the bidding scheme is obtained based on the task planning set.

9. A large-scale constellation earth observation mission planning apparatus, characterized by comprising: The device comprises: An upper layer task allocation module is configured to optimize and solve a preset large-scale constellation earth observation task planning model based on an improved adaptive large neighborhood search algorithm of fusion tabu search, so as to allocate large-scale constellation earth observation tasks to multiple star clusters and obtain a target star cluster allocation scheme; wherein the target star cluster allocation scheme represents that multiple star clusters are allocated with respective sets of tasks to be planned; the star cluster represents a set of n satellites in the same orbital plane with similar spatial distances, and all satellites in the same star cluster share the same task information, wherein n is a positive integer greater than 1; A lower layer task allocation module is configured to, for any of the star clusters, allocate tasks in the set of tasks to be planned to satellites in the star cluster based on a star cluster autonomous task planning algorithm based on an extended contract net mechanism, to obtain a star cluster task planning scheme of the star cluster; The planning scheme generation module is configured to generate a large-scale constellation earth observation task planning scheme corresponding to the large-scale constellation earth observation task based on the respective star cluster task planning schemes of the multiple star clusters. The upper-layer task allocation module comprises: An initial solution generation submodule is configured to generate an initial solution of the algorithm, and take the initial solution as a current solution and a current optimal solution; An iterative solution submodule is configured to repeatedly perform an iterative solution step until the algorithm meets a preset iteration stop condition; the iterative solution step comprises: determining a target destruction operator and a target repair operator from at least one destruction operator and at least one repair operator respectively, and generating a new solution by deleting a preset number of scheduled tasks from the current solution through the target destruction operator and inserting a preset number of unscheduled tasks through the target repair operator based on the taboo attribute of each task; and updating the current solution and the current optimal solution based on the new solution; A star cluster allocation scheme determination submodule is configured to take the current optimal solution as the target star cluster allocation scheme when it is determined that the algorithm meets the iteration stop condition.

Images