Satellite task scheduling solution space expansion method based on double-layer particle size optimization

Abstract

The application relates to a satellite task scheduling solution space expansion method based on double-layer particle size optimization, which comprises the following steps: obtaining an initial task scheduling scheme; pre-processing the initial task scheduling scheme to obtain a potential optional arc segment set of each task, an unfinished task set and a completed task set; dynamically adjusting the unfinished task set based on a double-layer particle size optimization method to generate a new scheduling solution; obtaining a task scheduling result based on the new scheduling solution and evaluating the task scheduling result. Through the replacement strategy of the double-layer particle size, the solution space is effectively expanded on the basis of the original scheduling solution by layer-by-layer replacement optimization without increasing resources or prolonging the task time, so that more tasks can find suitable scheduling time periods, and the task completion rate is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite task planning and resource scheduling, in particular to a satellite task scheduling solution space expansion method based on double-layer granularity optimization. BACKGROUND

[0002] Nowadays, satellites of different missions such as communication, navigation, remote sensing, etc. have brought major changes to various industries. When a satellite along the orbit enters the visible space of a ground station, the ground station sends measurement and control instructions to the satellite through a measurement and control antenna, and the satellite performs work in different fields such as communication, navigation, remote sensing, etc. based on the instructions to serve mankind. In actual scenarios, ground stations can only be established within a certain geographical range, resulting in a sharp narrowing of the measurement and control bandwidth, especially in the case of ground station failures, which will inevitably deepen the contradiction between massive measurement and control tasks and scarce ground stations.

[0003] Satellite measurement and control task scheduling is to establish a communication link between a ground station and a satellite within the visible range of the satellite in orbit, complete telemetry tracking, obtain position information, and measure and analyze the working conditions of satellite-borne ground equipment. The specific process is as follows: first, the user transmits the satellite measurement and control task demand to the ground scheduling center, the scheduling center plans the satellite measurement and control scheme by analyzing and calculating the satellite task, then transmits the generated scheduling scheme to the corresponding ground measurement and control ground equipment, and finally executes the satellite measurement and control task through the measurement and control ground equipment, and completes the uploading and injection of satellite task instructions, and feeds back data information to the scheduling center.

[0004] In complex task scheduling problems, especially in the field of measurement and control ground equipment task scheduling, the task scheduling system usually generates an initial scheduling scheme using heuristic algorithms, rule algorithms or algorithms based on global optimization. However, due to constraints such as ground equipment resources, time windows, task priorities, etc., the scheduling system often cannot find an ideal scheduling time period for all tasks, resulting in the overall completion rate of the tasks failing to reach the theoretical optimal level.

[0005] When facing high complexity scheduling problems, the current scheduling system often falls into a local optimal solution, i.e. only optimizing the scheduling of the current task under limited resources and time conditions, while ignoring the global optimization of the overall task. The generation of such a local optimal solution often leads to the fact that part of the tasks in the system cannot be completed on time due to resource conflicts, tight time windows, etc., thereby affecting the task execution efficiency of the system.

[0006] The limitations of the existing optimization methods mainly manifest in two aspects: one is to rely on increasing ground equipment resources or adjusting the time window of task execution to improve the task completion rate. Although this method can solve the problem of resource shortage to a certain extent, its applicability is poor in the case of limited ground equipment resources or strict requirements for task completion time limit of the system, and it cannot meet the demand of efficient scheduling; the second is that the existing optimization methods are mostly limited to local adjustment of the initial scheduling scheme, such as trying to improve resource utilization through task reordering or task gap filling. This method can only have an effect on local tasks and cannot fundamentally solve the problem of low global task completion rate, and due to the limitation of the solution space by the existing resources, it cannot break through the limitation of the initial solution.

[0007] The task scheduling problem belongs to a typical NP-hard problem, and with the expansion of the task scale, the solution space increases exponentially, resulting in that the optimization algorithm is difficult to efficiently search for the global optimal solution in the complex solution space. In view of this problem, how to optimize the scheduling solution without increasing resources or prolonging the task time, expand the task solution space of the system, and improve the overall completion rate of task scheduling has become a key problem to be solved in the current task scheduling field. SUMMARY

[0008] The purpose of the present application is to provide a satellite task scheduling solution space expansion method based on double-layer granularity optimization, which can expand the task solution space of the system without increasing resources or prolonging the task time, solve the problem of difficulty in optimizing the initial solution in the task scheduling process, and improve the overall completion rate of the task.

[0009] The technical scheme adopted by the present application is a satellite task scheduling solution space expansion method based on double-layer granularity optimization, comprising the following steps:

[0010] S1, obtaining an initial task scheduling scheme;

[0011] S2, preprocessing the initial task scheduling scheme to obtain a set of potential optional arc segments of each task, a set of unfinished tasks, and a set of completed tasks;

[0012] S3, dynamically adjusting the set of unfinished tasks based on a double-layer granularity optimization method to generate a new scheduling solution; specifically comprising:

[0013] S31, setting a first layer granularity parameter x, and randomly selecting x unfinished tasks from the set of unfinished tasks; for each selected unfinished task, performing a 1:1 replacement mechanism to replace the completed tasks in the set of completed tasks, and generating a new set of unfinished tasks by taking the successfully replaced completed tasks as new unfinished tasks;

[0014] S32, setting a second layer granularity parameter for each of the new unfinished tasks in the new unfinished task set, performing the 1:1 replacement mechanism with the finished tasks in the finished task set to replace the finished tasks in multiple layers recursively until reaching a replacement depth or the new unfinished task set is empty, the replacement depth being the second layer granularity parameter ;

[0015] S33, iteratively performing steps S31 to S32 until reaching a maximum iteration number or the unfinished task set is empty;

[0016] S4, obtaining a task scheduling result based on the new scheduling solution, and evaluating the task scheduling result.

[0017] The satellite task scheduling solution space expansion method based on double-layer granularity optimization has the advantages that: through the replacement strategy of double-layer granularity, on the basis of the original scheduling solution, the solution space is effectively expanded without increasing resources or prolonging the task time, so that more tasks can find suitable scheduling time periods, the task completion rate is improved, the task replacement optimization strategy can make the idle time periods of the ground equipment be more fully utilized, resource waste is reduced, meanwhile, the design of double-layer granularity optimization enables the method to be self-adaptively adjusted according to the task demand, is applicable to task scheduling problems of different scales, and has a wide application prospect.

[0018] Preferably, in step S3, the specific process of performing the 1:1 replacement mechanism with the finished tasks in the finished task set to replace the finished tasks is as follows:

[0019] For the current unfinished task, all the potential optional arc segments in the potential optional arc segment set are sorted in ascending order of arc segment length to obtain a sorted potential optional arc segment set;

[0020] The sorted potential optional arc segment set is checked in sequence, and once it is found that a potential optional arc segment is occupied by a finished task, the finished task occupying the arc segment is removed, and it is verified whether the current unfinished task can be arranged in the arc segment; if the verification is feasible, the current unfinished task is arranged in the arc segment, the state of the removed finished task is set as unfinished, and the removed finished task is put into the unfinished task set as a new unfinished task; if the verification is not feasible, the finished task occupying the arc segment is not removed, and the next potential optional arc segment is checked.

[0021] As preferred, the specific process of verifying whether the current unfinished task can be arranged in the arc segment is: if the arc segment does not exist satellite orbit number conflict with all the completed tasks, and the arc segment does not exist ground equipment occupation time conflict with all the completed tasks, the current unfinished task can be arranged in the arc segment; if the arc segment exists satellite orbit number conflict with any completed task or exists ground equipment occupation time conflict with any completed task, the current unfinished task cannot be arranged in the arc segment; the satellite orbit number conflict is defined as not allowing to execute multiple same type tasks under the same satellite and the same orbit number.

[0022] As preferred, before step S32, further comprising:

[0023] S311, for the new unfinished task set, firstly, it is judged whether there exists an unoccupied potential optional arc segment in the potential optional arc segment set corresponding to each unfinished task in the new unfinished task set, if yes, the unoccupied potential optional arc segment of the unfinished task is sorted in ascending order according to the arc segment length, the shortest unoccupied potential optional arc segment is selected based on the minimum arc segment principle and is assigned to the unfinished task, and the unfinished task is deleted from the new unfinished task set, until all the unfinished tasks are traversed, and a new unfinished task set is regenerated.

[0024] As preferred, in step S32, for each layer replacement in the multi-layer recursive replacement, the new unfinished task replaced out is judged whether there exists an unoccupied potential optional arc segment in the potential optional arc segment set corresponding to the new unfinished task according to the way of step S311, if yes, the unoccupied potential optional arc segment of the unfinished task is sorted in ascending order according to the arc segment length, the shortest unoccupied potential optional arc segment is selected based on the minimum arc segment principle and is assigned to the unfinished task, and the new unfinished task is deleted from the current new unfinished task set; if no, the next layer replacement is performed on the new unfinished task.

[0025] As preferred, in step S1, further comprising: reading satellite visibility forecast, task demand and ground equipment constraint information;

[0026] The satellite visibility forecast contains M pieces of visibility forecast of X satellites for N days, each piece of the visibility forecast contains an arc segment of a satellite and a highest elevation angle of the satellite, and the satellite visibility forecast is stored in a multidimensional array Acc.

[0027] The task demand contains SN total tasks, each total task contains tasks for N days, each day contains K tasks, each task contains a satellite number, a task number, an expansion chain time before a forecast start time and a delay disassembly chain time after a forecast end time, and the task demand is stored in a multidimensional array req.

[0028] The ground equipment constraint information contains ground equipment codes of all available ground equipment and ground equipment disable time periods corresponding to the ground equipment, and the ground equipment constraint information is stored in a multi-dimensional array FB.

[0029] Preferably, in step S2, the preprocessing of the initial task scheduling scheme specifically includes:

[0030] S21: According to the obtained satellite visibility forecast, task requirements and ground equipment constraint information, a set of potential optional arc segments for each task is screened out;

[0031] S22: According to the initial task scheduling scheme, each task in the initial task scheduling scheme is marked to mark the completed tasks and the uncompleted tasks, wherein the completed tasks are tasks that have been allocated to arc segments, the uncompleted tasks are tasks that have not been allocated to arc segments, a set of completed tasks and a set of uncompleted tasks are obtained, and a task completion rate is calculated.

[0032] Preferably, in step S21, the specific process of screening out a set of potential optional arc segments for each task includes the following steps:

[0033] S211: According to the satellite number and the task number of each task, all visible forecasts of the corresponding satellite for the day are extracted from the multi-dimensional array Acc for each task;

[0034] S212: From all the visible forecasts for the day, the forecasts whose highest elevation angles of the satellite are greater than or equal to the highest elevation angle value required by the task are screened out to form a set of visible forecasts meeting the highest elevation angle requirement;

[0035] S213: From the set of visible forecasts meeting the highest elevation angle requirement, the forecasts whose starting time is within the allowed offset range are screened out to form a set of visible forecasts meeting the time window constraint;

[0036] S214: For each visible forecast in the set of visible forecasts meeting the time window constraint, it is checked whether the ground equipment occupation time of each visible forecast overlaps with the ground equipment disable time period, if there is an overlap, the visible forecast is unavailable, if there is no overlap, the arc segment of the visible forecast is marked as a potential optional arc segment for the task; The ground equipment occupation time = the time after the end of the forecast is delayed to the time before the start of the forecast is expanded to the chain time.

[0037] Preferably, the specific process of step S4 includes:

[0038] S41: Based on the new scheduling solution generated in step S3, a task scheduling matrix is established, and each row of the task scheduling matrix represents a successfully scheduled task and records the arc segment information corresponding to the task;

[0039] S42, write the task scheduling matrix into an xml file and format the output by a tinyxml2 library function to obtain a task scheduling result;

[0040] S43, analyze the task completion rate and calculation time consumption of the task scheduling result. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A flowchart of the satellite task scheduling solution space expansion method based on double-layer granularity optimization of the application;

[0042] Figure 2 A flowchart of generating a new scheduling solution for the unfinished task set based on the dynamic adjustment of the double-layer granularity optimization method in the application;

[0043] Figure 3 A simulation diagram of the task scheduling success rate of TT&C and data transmission according to the initial scheduling scheme in the experiment of the application;

[0044] Figure 4 A simulation diagram of the task scheduling success rate of TT&C and data transmission according to the scheduling scheme proposed in the application in the experiment of the application;

[0045] Figure 5 A final task scheduling result Gantt chart obtained according to the scheduling scheme proposed in the application in the experiment of the application. DETAILED DESCRIPTION

[0046] The application will be further described below with reference to the accompanying drawings and in conjunction with specific embodiments, and the protection scope of the application is not limited to the specific embodiments.

[0047] The application provides a satellite task scheduling solution space expansion method based on double-layer granularity optimization, as shown in the figure, comprising the following steps: Figure 1

[0048] Step one, obtain an initial task scheduling scheme;

[0049] In specific embodiments, the initial task scheduling scheme is a ground equipment work plan xml file, which contains arc segment allocation information of all completed tasks, each completed task corresponds to a PLAN node, and each PLAN node (plan node) contains a spacecraft number, a ground equipment number, a task preparation start time, a ground equipment tracking start time, a ground equipment tracking end time, a task end time, a task code number corresponding to the plan, a day number corresponding to the plan, and a task number corresponding to the plan.

[0050] ​In addition, satellite visibility prediction, task requirements and ground equipment constraint information are read;

[0051] The satellite visibility prediction is a satellite orbit prediction xml file, containing M pieces of visibility prediction of X satellites for N days, wherein X, N and M are defined by the data given in the file, and each satellite visibility prediction contains ground equipment code, satellite total orbit number, transit orbit number, total satellite transit orbit number, satellite entry time, entry pitch angle, satellite maximum elevation time, maximum elevation, satellite exit time, and exit pitch angle. All prediction information is written into a multidimensional array Acc;

[0052] The satellite total orbit number refers to the total number of complete orbit periods accumulated since the satellite was launched;

[0053] The transit orbit number is a unique identifier for a single transit of a satellite to a particular ground station, used to constrain task allocation conflicts within the same orbit;

[0054] The satellite entry time refers to the starting time when the satellite enters the effective coverage range of the ground station's TT&C antenna;

[0055] The satellite exit time refers to the end time when the satellite leaves the effective coverage range of the ground station's TT&C antenna;

[0056] The time period between the satellite entry time and the satellite exit time is the arc segment;

[0057] The maximum elevation is the elevation angle corresponding to the highest position point of the satellite during transit relative to the observation point of the ground station;

[0058] The task requirement is a high-low-orbit weekly task requirement xml file, containing SN total tasks, each total task containing N days of tasks, each day containing K tasks, each task containing task number, task type (TT&C or data transmission), satellite parameters, arc segment selection constraints and ground available ground equipment set, and all requirement information is written into a multidimensional array req;

[0059] The satellite parameters include satellite code, pre-expansion chain time before the prediction start time (such as 10 minutes in advance), and post-delay disassembly chain time after the prediction end time (such as 5 minutes later to end);

[0060] The arc segment selection constraint includes tracking maximum elevation requirement and task window constraint, and the task window constraint includes constraint mode definition, allowed range offset, most desired time / orbit number and allowed range offset;

[0061] The ground available ground equipment set contains ground equipment code and ground equipment support function, and the ground equipment support function includes only data transmission, only TT&C or both data and TT&C.

[0062] The ground equipment constraint information is a ground equipment function XML file containing ground equipment codes of all available ground equipment on the ground and a plurality of ground equipment disabled time periods corresponding to the ground equipment, each ground equipment disabled time period containing a ground equipment disabled start time and a ground equipment disabled end time, and all ground equipment disabled information is written into a multi-dimensional array FB.

[0063] The step one imports information through the read-write XML module, and the imported information is stored in a corresponding matrix for subsequent algorithm calling.

[0064] Step two, preprocessing the initial task scheduling scheme to obtain a potential optional arc segment set of each task, an unfinished task set and a completed task set.

[0065] The specific process of preprocessing the initial task scheduling scheme includes the following steps:

[0066] S21: According to the obtained satellite visibility forecast, task requirements and ground equipment constraint information, the potential optional arc segment set of each task is screened out.

[0067] S22: According to the initial task scheduling scheme, each task in the initial task scheduling scheme is marked to mark the completed tasks and the unfinished tasks, wherein the completed tasks are tasks allocated to the arc segments, the unfinished tasks are tasks not allocated to the arc segments, the completed task set and the unfinished task set are obtained, and the task completion rate is calculated.

[0068] The specific process of screening out the potential optional arc segment set of each task is as follows:

[0069] S211: According to the satellite number and the task number of each task, all visible forecasts of the corresponding satellite on the day are extracted from the multi-dimensional array Acc under each task.

[0070] S212: From all the visible forecasts on the day, the forecasts with the highest elevation angle ≥ the highest elevation angle value required by the task are screened out to form a visible forecast set meeting the highest elevation angle requirement.

[0071] S213: From the visible forecast set meeting the highest elevation angle requirement, the forecasts with the forecast start time within the allowed offset range are screened out to form a visible forecast set meeting the time window constraint.

[0072] S214, for each visible prediction in the visible prediction set of the time window constraint, check whether the ground equipment occupation time of each visible prediction overlaps with the ground equipment disable time period, if there is overlap, the visible prediction is unavailable, if there is no overlap, mark the arc segment of the visible prediction as a potential optional arc segment of the task; the ground equipment occupation time = the time after the prediction end time is delayed to the time when the chain is cut + the time before the prediction start time is expanded to the time when the chain is built.

[0073] wherein the specific process of marking each task in the initial task scheduling scheme is: defining a vector <int>An array Iteration_path of type is initialized with all elements having a value of 0, and the array Iteration_path contains as many vectors as the total number of tasks, and the vectors are used to store the completion of each task, such as Iteration_path[0] represents the completion of the first task, Iteration_path[1] represents the completion of the second task, according to the initial task scheduling scheme data read in step S1, according to the task number corresponding to the plan in each PLAN node, the day number corresponding to the plan, and the task number corresponding to the plan, the task is calculated as the mth demand (i.e. demand index), and the calculation method of the demand index is: task number x total number of task codes + plan corresponding day number x total number of days + plan corresponding subtask number; the spacecraft number (i.e. satellite number) in each PLAN node and the day number corresponding to the plan are used to find all visible predictions of the satellite on the day in the satellite visibility prediction, and the arc segment start time, arc segment end time and device number of the visible prediction are compared with the device tracking start time, device tracking end time and device number in the PLAN node, to determine whether they are the same, if they are the same, the value of the demand index (the mth demand) in the array Iteration_path is changed to the index corresponding to the value of the visible prediction of the satellite on the day in the traversal process, such as Iteration_path[8]=216 represents that the 9th demand has been completed (the index starts from 0), and the arc segment corresponding to the demand is the 217th visible prediction (the index starts from 0) of the satellite on the day corresponding to the task satellite day number constraint. Traverse all the PLAN nodes of the task scheduling scheme data, and record the demand number completed by all the PLAN nodes and the arc segment number of the demand in the array Iteration_path by the above-mentioned method, the number of elements with a value of 0 in the array Iteration_path is the number of uncompleted tasks, and the number of elements with a value other than 0 is the number of completed tasks, the ratio of the number of completed tasks to the total number of tasks is calculated to obtain the current scheduling solution and calculate the task completion rate, and the uncompleted task set is recorded, which is the index set of all elements with a value of 0 in the array Iteration_path.

[0074] Step three, dynamically adjusting the uncompleted task set based on the double-layer granularity optimization method to generate a new scheduling solution; dynamically adjusting the task scheduling scheme by double-layer granularity optimization to find more feasible execution time windows for uncompleted tasks, and finally improve the overall task completion rate; as shown in the figure, specifically comprising: Figure 2

[0075] ​S31, set a first layer granularity parameter x, randomly select x unfinished tasks from the unfinished task set; for each selected unfinished task, perform a 1:1 replacement mechanism to replace a completed task in the completed task set, and generate a new unfinished task set by taking the successfully replaced completed task as a new unfinished task;

[0076] S311, for the new unfinished task set, first judge whether there is an unoccupied potential optional arc segment in the potential optional arc segment set corresponding to each unfinished task in the new unfinished task set, if there is, sort the unoccupied potential optional arc segment of the unfinished task from small to large according to the arc segment length, select the shortest unoccupied potential optional arc segment based on the minimum arc segment principle and assign it to the unfinished task, and delete the unfinished task from the current new unfinished task set, until all unfinished tasks are traversed, and a new unfinished task set is generated;

[0077] S32, set a second layer granularity parameter , for each new unfinished task in the new unfinished task set, perform a 1:1 replacement mechanism to replace a completed task in the completed task set in multiple layers of recursive replacement until the replacement depth or the new unfinished task set is empty, the replacement depth is the second layer granularity parameter ;

[0078] S33, iteratively perform steps S31 to S32 until the maximum iteration number or the unfinished task set is empty.

[0079] The specific principle of the 1:1 replacement mechanism is as follows:

[0080] For the current unfinished task, traverse its potential optional arc segment set, sort all potential optional arc segments in the potential optional arc segment set from small to large according to the arc segment length, and obtain the sorted potential optional arc segment set;

[0081] Check each potential optional arc segment in its sorted potential optional arc segment set in order, or check each potential optional arc segment in the potential optional arc segment set in random order. Once it is found that a potential optional arc segment is occupied by a completed task, remove the completed task occupying the arc segment, and verify whether the current unfinished task can be arranged in the arc segment; if it is feasible, arrange the current unfinished task in the arc segment, set the state of the removed completed task to unfinished, and put the successfully replaced completed task into the unfinished task set as a new unfinished task; if it is not feasible, do not remove the completed task occupying the arc segment, and check the next potential optional arc segment.

[0082] In the 1:1 replacement mechanism, the specific principle of verifying whether the current unfinished task can be arranged in the arc segment is as follows:

[0083] If the arc segment does not exist satellite ring number conflict with all completed tasks, and the arc segment does not exist device occupation time conflict with all completed tasks, the current unfinished task can be arranged in the arc segment; if the arc segment exists satellite ring number conflict with any completed task or exists device occupation time conflict with any completed task, the current unfinished task cannot be arranged in the arc segment; the satellite ring number conflict is defined as that multiple same type tasks are not allowed to be executed under the same satellite and the same ring number.

[0084] In step S32, for each layer replacement in the multi-layer recursive replacement, the new unfinished task replaced is judged whether there is an unoccupied potential optional arc segment in the corresponding potential optional arc segment set in the manner of step S311, if there is, the unoccupied potential optional arc segment of the unfinished task is sorted from small to large according to the arc segment length, the shortest unoccupied potential optional arc segment is selected and distributed to the unfinished task based on the minimum arc segment principle, and the new unfinished task is deleted from the current new unfinished task set; if not, the next layer replacement is performed on the new unfinished task.

[0085] In the application, before performing each layer 1:1 replacement mechanism, it is necessary to judge whether there is an unoccupied potential optional arc segment in the corresponding potential optional arc segment set of each unfinished task in the new unfinished task set, if there is, the unoccupied potential optional arc segment of the unfinished task is sorted from small to large according to the arc segment length, the shortest unoccupied potential optional arc segment is selected and distributed to the unfinished task based on the minimum arc segment principle, and the unfinished task is deleted from the current new unfinished task set, until all unfinished tasks are traversed, and a new unfinished task set is regenerated.

[0086] In the application, through the 1:1 replacement mechanism between the unfinished task and the completed task in step S32, the potential optional arc segment of the new unfinished task replaced is used as a new solution space, so that the solution space is expanded and the possibility of task completion is improved without reducing the task completion rate.

[0087] Through the multi-layer recursive 1:1 replacement between the unfinished task and the completed task in step S33, the solution space is further expanded in each layer 1:1 replacement process, and the replacement depth can be adjusted by controlling the size of the second layer granularity, so as to control the calculation efficiency, and the size of the second layer granularity can be adjusted adaptively according to the task size.

[0088] Step four, based on the new scheduling solution generated in step S3, obtain the task scheduling result, and evaluate the task scheduling result, the specific process includes:

[0089] S41, based on the new scheduling solution generated in step S3, a task scheduling matrix is established, each row of the task scheduling matrix represents a successfully scheduled task, and records the corresponding arc segment of the task, and also records the satellite number, device number, satellite circle number, preparation start time, tracking start time, tracking end time, chain disassembly end time, demand index number and other information;

[0090] S42, the task scheduling matrix is written into an xml file and formatted output by using the tinyxml2 library function, and the task scheduling result is obtained;

[0091] S43, the task completion rate and calculation time consumption of the task scheduling result are analyzed.

[0092] The effect of the algorithm is evaluated, S represents the measurement and control task satisfaction degree, and D represents the data transmission task satisfaction degree, and the evaluation index form of the measurement and control task satisfaction degree is:

[0093] S43, the task completion rate and calculation time consumption of the task scheduling result are analyzed.

[0094] The effect of the algorithm is evaluated, S represents the measurement and control task satisfaction degree, and D represents the data transmission task satisfaction degree, and the evaluation index form of the measurement and control task satisfaction degree is: ; wherein, is the number of completed measurement and control tasks, is the total number of measurement and control tasks.

[0095] The evaluation index form of the data transmission task satisfaction degree is: ; wherein, is the number of completed measurement and control tasks, is the total number of measurement and control tasks.

[0096] Based on the above technical scheme, the satellite task scheduling solution space expansion method based on double-layer particle size optimization provided by the application evaluates the planning result after completing the task planning. The method of the application dynamically adjusts the scheduling scheme by double-layer task replacement strategy (particle size optimization) to expand the solution space, instead of relying on local adjustment only. In the case of not increasing resources, the device idle time utilization rate is improved, the task conflict is reduced, and the overall task completion rate is improved. It is suitable for scheduling scenes with limited resources such as satellite measurement and control and communication, and is especially suitable for high complexity and multi-constrained task planning problems.

[0097] The following experiment proves the superiority of the satellite task scheduling solution space expansion method based on double-layer granularity optimization provided by the present application.

[0098] The simulation data set used in the experiment is from the data set 1 in the second subject of the fourth "Sky Cup" artificial intelligence challenge in spacecraft management and control field. The basic situation of the measurement and control equipment and task requirements is shown in Table 1:

[0099] Table 1 Basic situation of measurement and control scene

[0100] .

[0101] The experimental verification environment is Win11 64-bit system, the CPU is R5-7530, 2.4GHz six-core, 16G RAM, and the experimental tool is VS2019. In order to more objectively compare the performance of the algorithm, the task scheduling success rate is used to measure the quality of the algorithm. According to the initial scheduling scheme obtained by the above scene experiment, the measurement and control and data transmission task scheduling success rate is as shown in Figure 3 .

[0102] The double-layer granularity expansion solution space algorithm is used, based on the initial scheduling scheme and initial scheduling plan, satellite visibility prediction, task demand and equipment constraint information, the first layer granularity (the number of replaced unfinished demands) is set to 10, the second layer granularity (the maximum replacement depth) is set to 50, and the maximum number of iterations is set to 100, and the task scheduling success rate obtained is as shown in Figure 4 , Figure 3 and Figure 4 As can be seen from the above, the satellite task scheduling solution space expansion method based on double-layer granularity optimization provided by the present application can improve the data transmission completion rate and the measurement and control completion rate to 100% completion rate. As shown in Figure 5 , the final scheduling result Gantt chart is given, which more directly shows the scheduling situation of the ground station resources.< / int>

Claims

1. A satellite mission scheduling solution space expansion method based on two-layer granularity optimization, characterized in that: The method comprises the following steps: S1, obtaining an initial task scheduling scheme; S2, preprocessing the initial task scheduling scheme to obtain a potential optional arc segment set of each task, an unfinished task set and a completed task set; specifically comprising: S21: according to the obtained satellite visibility forecast, task requirements and ground equipment constraint information, screening out the potential optional arc segment set of each task; S22: according to the initial task scheduling scheme, marking each task in the initial task scheduling scheme to mark the completed tasks and the unfinished tasks, wherein the completed tasks are tasks that have been allocated to arc segments, the unfinished tasks are tasks that have not been allocated to arc segments, obtaining the completed task set and the unfinished task set, and calculating the task completion rate; S3, dynamically adjusting the unfinished task set based on a double-layer granularity optimization method to generate a new scheduling solution; specifically comprising: S31, setting a first layer granularity parameter x, and randomly selecting x unfinished tasks from the unfinished task set; for each selected unfinished task, performing a 1:1 replacement mechanism to replace the completed tasks in the completed task set, and taking the successfully replaced completed tasks as new unfinished tasks to generate a new unfinished task set; S32, setting a second layer granularity parameter , for each of the new unfinished tasks in the new unfinished task set, performing the 1:1 replacement mechanism with the completed tasks in the completed task set for multi-layer recursive replacement until reaching a replacement depth or the new unfinished task set being empty, the replacement depth being the second layer granularity parameter ; wherein, for each layer of replacement in the multi-layer recursive replacement, the new unfinished task replaced out judges whether there is an unoccupied potential optional arc segment in the corresponding potential optional arc segment set, if there is, then the unoccupied potential optional arc segment of the unfinished task is sorted in ascending order of arc segment length, the shortest unoccupied potential optional arc segment is selected based on the minimum arc segment principle and assigned to the unfinished task, and the new unfinished task is deleted from the current new unfinished task set; if not, the next layer of replacement is performed on the new unfinished task; S33, iteratively executing steps S31 to S32 until the maximum iteration number is reached or the unfinished task set is empty; S4, obtaining a task scheduling result based on the new scheduling solution, and evaluating the task scheduling result.

2. The method of claim 1, wherein the method is based on a two-layer particle swarm optimization (PSO) algorithm. In step S3, the specific process of performing the 1:1 replacement mechanism to replace the completed tasks in the completed task set and taking the successfully replaced completed tasks as new unfinished tasks is as follows: for the current unfinished task, traversing its potential optional arc segment set, sorting all potential optional arc segments in the potential optional arc segment set in ascending order of arc segment length to obtain a sorted potential optional arc segment set; checking each potential optional arc segment in the sorted potential optional arc segment set in order, and once it is found that a potential optional arc segment is occupied by a completed task, removing the completed task occupying the arc segment and verifying whether the current unfinished task can be arranged in the arc segment at the same time; if the verification is feasible, arranging the current unfinished task in the arc segment, setting the state of the removed completed task as unfinished, and putting the removed completed task into the unfinished task set as a new unfinished task; if the verification is not feasible, traversing the next potential optional arc segment.

3. The method of claim 2, wherein: The specific process of verifying whether the current unfinished task can be arranged in the arc segment is as follows: if the arc segment does not have satellite orbit number conflicts with all completed tasks, and the arc segment does not have ground equipment occupation time conflicts with all completed tasks, the current unfinished task can be arranged in the arc segment; if the arc segment has satellite orbit number conflicts with any completed task or has ground equipment occupation time conflicts with any completed task, the current unfinished task cannot be arranged in the arc segment; The satellite orbit number conflict is defined as that multiple same type tasks are not allowed to be executed under the same satellite orbit number.

4. The method of claim 1 or 3, wherein: Before step S32, further comprising: S311, for the new set of unfinished tasks, first judge whether there is an unoccupied potential optional arc segment in the potential optional arc segment set corresponding to each unfinished task in the new set of unfinished tasks, if there is, then sort the unoccupied potential optional arc segment of the unfinished task according to the arc segment length from small to large, select the shortest unoccupied potential optional arc segment based on the minimum arc segment principle and assign it to the unfinished task, and delete the unfinished task from the current new set of unfinished tasks until all unfinished tasks are traversed, and a new set of unfinished tasks is regenerated.

5. The method of claim 1, wherein: In the step S1, further comprising: reading satellite visibility forecast, task requirements and ground equipment constraint information; Wherein, the satellite visibility forecast contains M visibility forecasts of X satellites for N days, each visibility forecast contains an arc segment of a satellite and the highest elevation angle of the satellite; the satellite visibility forecast is stored in a multidimensional array Acc; The task requirements contain SN total tasks, each total task contains N days of tasks, each day contains K tasks, each task contains satellite number, task number, pre-forecast start time extension chain time and post-forecast end time delay chain time; the task requirements are stored in a multidimensional array req; The ground equipment constraint information contains the equipment code of all available ground equipment and the corresponding equipment disabled time period, and the equipment constraint information is stored in a multidimensional array FB.

6. The method of claim 5, wherein: In step S21, the specific process of screening out the potential optional arc segment set of each task includes the following steps: S211, according to the satellite number and task number of each task, extract all visibility forecasts of the corresponding satellite on the same day from the multidimensional array Acc; S212, from all the visibility forecasts on the same day, select the forecasts with the highest elevation angle ≥ the highest elevation angle value required by the task to form a set of visibility forecasts meeting the highest elevation angle requirement; S213, from the set of visibility forecasts meeting the highest elevation angle requirement, select the forecasts with the forecast start time within the allowed offset range to form a set of visibility forecasts meeting the time window constraint; S214, for each visibility forecast in the set of visibility forecasts meeting the time window constraint, check whether the ground equipment occupation time of each visibility forecast overlaps with the ground equipment disabled time period, if there is overlap, the visibility forecast is unavailable, if there is no overlap, mark the arc segment of the visibility forecast as the potential optional arc segment of the task; the ground equipment occupation time = the post-forecast end time delay chain time - the pre-forecast start time extension chain time.

7. The method of claim 1 or 6, wherein: The specific process of step S4 includes: S41, based on the new scheduling solution generated in step S3, establish a task scheduling matrix, each row of the task scheduling matrix represents a successfully scheduled task, and records the arc segment corresponding to the task; S42, write the task scheduling matrix into an xml file and format the output by using the tinyxml2 library function to obtain the task scheduling result; S43, analyze the task completion rate and calculation time consumption of the task scheduling result.

Images