CN114781508A - Clustering-based satellite measurement and control scheduling method and system - Google Patents

Clustering-based satellite measurement and control scheduling method and system Download PDF

Info

Publication number
CN114781508A
CN114781508A CN202210407679.0A CN202210407679A CN114781508A CN 114781508 A CN114781508 A CN 114781508A CN 202210407679 A CN202210407679 A CN 202210407679A CN 114781508 A CN114781508 A CN 114781508A
Authority
CN
China
Prior art keywords
task
time
time window
population
antenna
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210407679.0A
Other languages
Chinese (zh)
Inventor
宋彦杰
杜永浩
何磊
闫俊刚
陈英武
吕济民
陈宇宁
刘晓路
陈盈果
沈大勇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
National University of Defense Technology
Original Assignee
National University of Defense Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by National University of Defense Technology filed Critical National University of Defense Technology
Publication of CN114781508A publication Critical patent/CN114781508A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06316—Sequencing of tasks or work

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Biomedical Technology (AREA)
  • Physiology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Computational Linguistics (AREA)
  • Genetics & Genomics (AREA)
  • Development Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Educational Administration (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Game Theory and Decision Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a satellite measurement and control scheduling method and system based on clustering, wherein tasks are divided into K categories according to task characteristics, the close relation between the tasks is evaluated, similar tasks are as close as possible, but different tasks are as far as possible, and a genetic algorithm based on clustering is used for carrying out cross variation on two individuals in different categories when a population is crossed and varied, so that the balance of global search and local search is realized, a solution with higher quality can be obtained, and the problem of sequence dependence is well solved. Experiments show that the same optimization effect can be achieved by using the method of the invention only by using a knowledge-based genetic algorithm in about one third of time.

Description

Clustering-based satellite measurement and control scheduling method and system
Technical Field
The invention belongs to the field of satellite task planning, and particularly relates to a satellite measurement and control scheduling method and system based on clustering.
Background
In recent years, with the rapid development of space technology, satellites increasingly play an indispensable role as a space platform which can have application value in many fields. Satellites are used to perform many types of tasks, such as target observation, navigation positioning, communication transmission, etc., depending on the user task requirements. Whatever tasks the satellite needs to perform, it is necessary for the tasks to be transmitted by the satellite earth station in the form of instructions by the uplink, either directly or indirectly (prior transmission to the relay satellite). The process of the ground station uploading task and action instructions to the satellite and acquiring the running state of the satellite is called satellite measurement and control. The satellite measurement and control scheduling system realizes effective management of satellite and ground station resources by using an efficient scheduling algorithm. With the rapidly increasing number of on-orbit operating satellites and the need for satellite management control, efficient task scheduling for satellite measurement and control scheduling problems becomes more challenging.
The satellite measurement and control scheduling describes a process of reasonably arranging communication between a satellite and a ground station to realize satellite state monitoring and instruction uploading, and can also be simply described as follows: and finding a proper combination of measurement and control tasks for a series of satellites and a series of satellite ground stations within a mutually visible time window. The main reasons that the satellite measurement and control scheduling problem is difficult to solve are sequence dependence and oversubscription characteristics. Sequence dependency refers to the fact that the scheduling result of one task affects the subsequent tasks. This makes finding a feasible solution difficult. Oversubscription means that the ground station time window, if used at all, will have a portion of the task that was not successfully performed. The simultaneous existence of these two features makes the measurement and control scheduling problem especially difficult. The satellite measurement and control scheduling problem has proven to be NP-complete in complexity.
Disclosure of Invention
The invention aims to solve the technical problem of how to quickly provide a large-scale satellite measurement and control scheme, and provides a satellite measurement and control scheduling method and system based on clustering.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a satellite measurement and control scheduling method based on clustering comprises the following steps:
step 1: acquiring schedulable satellite resources, ground station resources, a task set and a visible time window set;
step 2: constructing a mixed integer programming model according to the content acquired in the step 1;
and step 3: solving the mixed integer programming model;
and 4, step 4: and outputting the solved measurement and control scheme.
Further, the mixed integer programming model is:
the objective function is:
Figure BDA0003602687750000021
wherein p istThe benefit of the task t is represented by,
Figure BDA0003602687750000022
a decision variable of 0-1 indicating whether a task T is scheduled on the ith visible time window on antenna a, 1 is scheduled, 0 is no, T indicates a task set, a indicates an antenna set,
Figure BDA0003602687750000023
indicating that the task t is scheduled on the antenna a in the ith visible time window, and TWta indicating the set of time windows in which the task t is scheduled on the antenna a;
the constraint conditions are:
Figure BDA0003602687750000024
equation 2 indicates that the task needs to start executing the task after the allowed start time;
Figure BDA0003602687750000025
representing the actual measurement and control time length of the task t arranged on the ith visible time window on the antenna a, dtRepresenting the measurement and control time length required by the task t;
Figure BDA0003602687750000026
equation 3 indicates that the task needs to be completed before the allowed end time;
Figure BDA0003602687750000027
representing the actual start time of task t scheduled on antenna a on the ith visible time window,
Figure BDA0003602687750000031
represents the end time of the task t arranged on the ith visible time window on the antenna a;
Figure BDA0003602687750000032
equation 4 indicates that a task can be performed at most once;
Figure BDA0003602687750000033
represents the earliest allowed start time of the task t arranged on the ith visible time window on the antenna a;
Figure BDA0003602687750000034
equation 5 indicates that the actual execution time of the task should be the same as the required time;
Figure BDA0003602687750000035
represents the latest allowable ending time of the task t scheduled on the ith visible time window on the antenna a;
Figure BDA0003602687750000036
equation 6 indicates that the task execution process needs to be within a ground station time range;
Figure BDA0003602687750000037
represents the start time of the ith visible time window of the task t scheduled on the antenna a;
Figure BDA0003602687750000038
equation 7 indicates that each task can only be served by one antenna;
Figure BDA0003602687750000039
equation 8 indicates that each task can only be executed within one visible time window;
Figure BDA00036026877500000310
indicating the end time of the ith visible time window of the task t scheduled on the antenna a;
Figure BDA00036026877500000311
equation 9 indicates that each task is performed on all antennas at most once;
Figure BDA00036026877500000312
equation 10 indicates that each task is executed at most once in the entire time window;
Figure BDA00036026877500000313
equation 11 indicates that the task is executed at most once during the planning period;
Figure BDA00036026877500000314
equation 12 indicates that the two tasks are to meet the interval requirement of task switching time; γ represents the transition time between tasks.
Further, the method for solving the mixed integer programming model in step 3 is a clustering-based genetic algorithm.
Further, the cluster-based genetic algorithm is:
step 3.1: initializing genetic algorithm parameters and clustering K-means method parameters;
step 3.2: generating an initialization population, wherein each individual in the population is a coded sequence obtained by coding all tasks in a task set;
step 3.3: when the iteration algebra does not reach the maximum iteration algebra, executing a step 3.3.1; otherwise, executing step 3.4;
step 3.3.1: generating an initial solution of a measurement and control scheme for each individual in the population and calculating a target function value;
step 3.3.2: if the maximum objective function value in the present generation population is greater than the optimal objective function value, updating the optimal objective function value and counting the count1＝count1+1；
If the maximum objective function value in the current generation population is larger than the maximum objective function value in the previous generation population, the count is updated2＝count2+1；
If the maximum objective function value in the current generation population is less than the maximum function value in the previous generation population multiplied by the proportion per, the count is updated3＝count3+1；
Step 3.3.3: selecting individuals to generate a new population by using a roulette wheel according to the objective function value;
step 3.3.4: updating the new population by using a clustering-based crossover and mutation method;
step 3.3.5: if count1Is equal to the threshold value Thre1If so, updating the parameter K value of the K-means method, clustering the tasks in the task set based on the task attributes again, and resetting the count parameter count1；
If count2Is equal to Thre2Replacing the new individual with the minimum objective function value in the population with the optimal individual corresponding to the optimal objective function value, and resetting the count parameter count2；
If count3Is equal to Thre3Then, locally optimizing the individual with the maximum objective function value in the new population, randomly generating a new individual, deleting the individual with the minimum objective function value in the new population, and resetting the count parameter count3；
Step 3.3.6: recording the maximum objective function value in the new population as the maximum objective function value of the previous generation population;
step 3.4: and outputting the individual corresponding to the optimal objective function value as a measurement and control scheme.
Further, in step 3.3.1: the method for generating the initial solution of the measurement and control scheme for each individual in the population is a task arrangement algorithm, and specifically comprises the following steps:
trying to arrange all tasks in the task set to each task according to the coding sequence and the visible time window sequence in turn, if all time windows are tried and arranged successfully, outputting the result of successful arrangement as an initial solution, and if not, continuing to try to arrange; the specific arrangement method for each task comprises the following steps:
1): calculating the earliest actual available time eat of task ttAnd the latest actual available time latt；
2): if the length of the ith time window of the task t on the antenna a is larger than the measurement and control time length required by the task t, namely
Figure BDA0003602687750000051
And the actual available time length of the task t is longer than the measurement and control time length of the task t, namely (lat)t-eatt)≥dtThen go to step 3); otherwise, continuing to schedule the next task;
3): earliest task t to actual available time eattAs the task start time, if the task t is the earliest the actual available time eattEqual to the start time of the ith visible time window of the task t scheduled on the antenna a
Figure BDA0003602687750000052
Go to step 4); otherwise, go to step 5);
4): scheduling task t on antenna a for the ith visible time window
Figure BDA0003602687750000053
Updating the remaining time window of the window to a new time window
Figure BDA0003602687750000054
5): scheduling task t on antenna a for the ith visible time window
Figure BDA0003602687750000055
The remaining time window of the window is updated into two new time windows
Figure BDA0003602687750000056
And
Figure BDA0003602687750000057
further, the method for generating the initialization population in step 3.2 is to generate the initialization population by using a plurality of heuristic initialization methods, wherein individuals generated by the various heuristic initialization methods exist in the initialization population.
Further, using a clustering-based intersection method in step 3.3.4 means:
dividing all tasks into K-type tasks by using a K-means clustering method according to task attributes;
when in crossing, the gene segments with the same length are respectively selected from two groups of tasks with different categories to be crossed, and the categories of the tasks are selected by roulette according to the objective function values.
Further, the clustering-based mutation method used in step 3.3.4 includes two types, one is mutation in the same category, and the other is mutation in different categories, and one of the two types of mutation is randomly selected to perform mutation operation in each mutation operation;
a variant in the same category refers to a swap of the location of two tasks within an individual that are within the same category.
Variation in different categories refers to swapping the location of two tasks within an individual that are in different categories.
Further, the clustering-based method in step 3.1 is to perform clustering based on the attributes of the tasks, where the attributes of the tasks include the earliest allowed start time, the latest allowed end time, the profit, and the measurement and control time length required by the tasks, and when the K value is updated and clustering is performed again, it is increased whether the tasks are successfully scheduled. .
The invention also provides a satellite measurement and control scheduling system based on clustering, which comprises the following modules:
an input module: the system comprises a time window set, a scheduling module and a scheduling module, wherein the time window set is used for acquiring schedulable satellite resources, ground station resources, a task set and a time window set for a satellite to see a ground station;
a model construction module: the method is used for constructing a mixed integer programming model according to the content acquired in the step 1;
a solution module: for solving the mixed integer programming model;
a scheme output module: and outputting the solved measurement and control scheme.
By adopting the technical scheme, the invention has the following beneficial effects:
the invention relates to a satellite measurement and control scheduling method and a satellite measurement and control scheduling system based on clustering, which solve the problem of satellite measurement and control scheduling SRSP (satellite radio range scheduling protocol) by using a Cluster-based genetic algorithm C-BGA (Cluster-based genetic algorithm). The tasks are divided into K classes according to the task characteristics, the close relation between the tasks is evaluated, the similar tasks are as close as possible, but the different tasks are as far as possible, when the population is crossed and varied, individuals in two different classes, namely two dissimilar task classes, are selected to carry out cross variation, the balance of global search and local search is realized, a solution with higher quality can be obtained, and the problem of sequence dependence is well solved. And generating an initial solution of a measurement and control scheme for each individual in the population through a task arrangement algorithm, judging the feasibility of task execution, and performing effective arrangement. When initializing the population, the initialization population is generated by using various heuristic initialization methods, and is not limited to one strategy for generation, so that the solution is diversified. Experiments show that the same optimization effect can be achieved by using the C-BGA method only by using about one third of the time of the KBGA.
Drawings
FIG. 1 is a flow chart of the system of the present invention;
FIG. 2 is a variation curve of K-value in the clustering method;
FIG. 3 is a box plot of an example of L _ L-5 and L-H-5;
FIG. 4 is a medium scale test set result;
FIG. 5 is a large scale test set of results.
Detailed Description
The technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
A satellite measurement and control scheduling method based on clustering is disclosed, as shown in FIG. 1, and comprises the following steps:
step 1: acquiring schedulable satellite resources, ground station resources, a task set and a time window set for a satellite to see a ground station;
and 2, step: and (3) constructing a mixed integer programming model according to the content acquired in the step (1).
In this embodiment, the mixed integer programming model is:
the objective function is:
Figure BDA0003602687750000081
wherein p istThe benefit of the task t is represented by,
Figure BDA0003602687750000082
a decision variable of 0-1 indicating whether a task T is scheduled on the ith visible time window on antenna a, 1 is scheduled, 0 is no, T indicates a task set, a indicates an antenna set,
Figure BDA0003602687750000083
indicating that task t is scheduled on antenna a for the ith visible time window, TWtaA set of time windows representing the scheduling of task t on antenna a;
the constraint conditions are:
Figure BDA0003602687750000084
equation 2 indicates that the task needs to start executing the task after the allowed start time;
Figure BDA0003602687750000085
representing the actual measurement and control time length of the task t arranged on the ith visible time window on the antenna a, dtRepresenting the measurement and control time length required by the task t;
Figure BDA0003602687750000086
equation 3 indicates that the task needs to be completed before the allowed end time;
Figure BDA0003602687750000087
representing the actual start time of task t scheduled on antenna a on the ith visible time window,
Figure BDA0003602687750000088
represents the end time of the task t arranged on the ith visible time window on the antenna a;
Figure BDA0003602687750000089
equation 4 indicates that a task can be performed at most once;
Figure BDA00036026877500000810
represents the earliest permitted start time of the task t scheduled on the ith visible time window on the antenna a;
Figure BDA00036026877500000811
equation 5 indicates that the actual execution time of the task should be the same as the required time;
Figure BDA00036026877500000812
represents the latest allowable ending time of the task t scheduled on the ith visible time window on the antenna a;
Figure BDA00036026877500000813
equation 6 indicates that the task execution process needs to be within a ground station time range;
Figure BDA00036026877500000814
represents the start time of the ith visible time window of the task t scheduled on the antenna a;
Figure BDA0003602687750000091
equation 7 indicates that each task can only be served by one antenna;
Figure BDA0003602687750000092
equation 8 indicates that each task can only be executed within one visible time window;
Figure BDA0003602687750000093
indicating the end time of the ith visible time window of the task t scheduled on the antenna a;
Figure BDA0003602687750000094
equation 9 indicates that each task is performed at most once on all antennas;
Figure BDA0003602687750000095
equation 10 indicates that each task is executed at most once in the entire time window;
Figure BDA0003602687750000096
equation 11 indicates that the task is executed at most once during the planning period;
Figure BDA0003602687750000097
equation 12 indicates that the two tasks are to meet the interval requirement of task switching time; γ represents the transition time between tasks.
And step 3: and solving the mixed integer programming model.
In this embodiment, the method for solving the mixed integer programming model is a clustering-based genetic algorithm.
The genetic algorithm based on clustering is:
step 3.1: initializing genetic algorithm parameters and clustering K-means method parameters;
step 3.2: generating an initialization population, wherein each individual in the population is a coded sequence after coding all tasks in a task set;
in this embodiment, the method for generating the initialization population is to generate the initialization population by using a plurality of heuristic initialization methods, where individuals generated by the various heuristic initialization methods exist in the initialization population.
In this embodiment, the plurality of heuristic initialization methods used include four heuristic population initialization methods and a random population initialization method.
The four types of heuristic population initialization are respectively called an initialization method based on earliest permitted start time sequencing, an initialization mode based on latest permitted end time sequencing, an initialization mode based on task profit sequencing and an initialization mode based on task duration sequencing.
Initialization based on earliest allowed start time ordering (EST): and sequentially sequencing the measurement and control tasks from the front to the back according to the earliest permitted start time of the measurement and control tasks, and generating individuals according to the sequenced sequence.
Initialization based on latest allowed end time ordering (LET): and sequencing the measurement and control tasks in sequence from the front to the back according to the latest allowable ending time of the measurement and control tasks, and generating individuals according to the sequenced sequence.
Initialization based on task revenue ranking (TP): and sequencing according to the descending order of the task benefits of the measurement and control tasks, and generating individuals according to the sequenced order.
Initialization (TD) based on task duration ordering: and sequencing according to the ascending sequence of the task duration time of the measurement and control task from small to large, and generating individuals according to the sequenced sequence.
After each heuristic rule is adopted to generate population individuals, the individuals are required to have more diversity through a mode of adjusting partial sequences of the individuals. Combining heuristic and locally random approaches makes it easier for each individual to find a good search starting location. The four heuristic methods and the random method respectively occupy 20% of all individuals in the population. Of course, the proportion of the four heuristic methods and the random method occupying all the individuals of the population can be adjusted according to needs, and in order to increase the diversity of the solution, the heuristic initialization method can further include other methods not limited to the four heuristic methods mentioned above.
Step 3.3: when the iteration algebra does not reach the maximum iteration algebra, executing a step 3.3.1; otherwise, executing step 3.4;
step 3.3.1: and generating an initial solution of a measurement and control scheme for each individual in the population and calculating an objective function value.
In this example, in step 3.3.1: the method for generating the initial solution of the measurement and control scheme for each individual in the population is a task arrangement algorithm, and specifically comprises the following steps:
all tasks in the task set are tried and arranged to each task according to the coding sequence and the visible time window sequence in turn, if all time windows are tried and arranged successfully, the result of successful arrangement is output as an initial solution, otherwise, the arrangement is tried continuously; the specific arrangement method for each task comprises the following steps:
1): calculating the earliest actual available time eat of task ttAnd the latest actual available time latt；
2): if the length of the ith time window of the task t on the antenna a is larger than the measurement and control time length required by the task t, namely
Figure BDA0003602687750000111
And the actual available time length of the task t is longer than the measurement and control time length of the task t, namely (lat)t-eatt)≥dtGo to step 3); otherwise, continuing to schedule the next task;
3): earliest task t to actual available time eattAs the task start time, if the task t is the earliest the actual available time eattEqual to the start time of the ith visible time window of the task t scheduled on the antenna a
Figure BDA0003602687750000112
Go to step 4); otherwise, turning to the step 5);
4): scheduling task t on antenna a for the ith visible time window
Figure BDA0003602687750000113
Updating the remaining time window of the window to a new time window
Figure BDA0003602687750000114
5): scheduling task t on antenna a for the ith visible time window
Figure BDA0003602687750000115
The remaining time window of the window is updated into two new time windows
Figure BDA0003602687750000116
And
Figure BDA0003602687750000117
in this embodiment, whether each time window can execute a task is sequentially determined for all tasks in the task set. And if the length of the time window exceeds the time required by the measurement and control task, attempting to schedule the task. The invention adopts a quick judgment method to determine whether the task has the possibility of being scheduled, and firstly, two new variables eat are required to be introducedtAnd lattRespectively, indicate the earliest practical availabilityTime and latest actual available time, the calculation method is shown in equation 13. If the actual available time exceeds the measurement and control task required time, the task can be successfully scheduled, and the task is scheduled to start executing at the earliest actual available time. And after the task is successfully scheduled, the residual available time window resources need to be updated, if the earliest practical available time of the task is equal to the earliest allowed starting time of the task, the residual time window of the window is cut into a new time window, and otherwise, the residual time window of the window is cut into two new time windows. And updating the time window set after finishing the time window clipping, and starting to arrange the next task.
Figure BDA0003602687750000118
Figure BDA0003602687750000119
Step 3.3.2: if the maximum objective function value in the current generation population is greater than the optimal objective function value, the optimal objective function value is updated and the count is counted1＝count1+ 1; the optimal objective function value in this embodiment is the optimal objective function value in the whole optimization process, and its initial value may be set to 0, and then in each iteration, the maximum objective function value in each generation of population is compared with the optimal value obtained by updating the maximum objective function value.
If the maximum objective function value in the current generation population is larger than the maximum objective function value in the previous generation population, the count is updated2＝count2+1；
If the maximum objective function value in the current generation population is less than the maximum function value in the previous generation population multiplied by the proportion per, the count is updated3＝count3+1；
Step 3.3.3: selecting individuals to generate a new population by using roulette according to the value of the objective function;
in this embodiment, a roulette method is used to select individuals to generate a new population, and then the new population is crossed and mutated.
The invention utilizes the roulette method to select individuals and task categories according to the objective function values of the individuals, so that the individuals with better performance and the categories are easier to select. The probability formula for selecting individuals using roulette is as follows:
Figure BDA0003602687750000121
wherein,
Figure BDA0003602687750000122
is the probability of the individual j, fjIs the objective function value for individual j, and P is the population.
Step 3.3.4: the new population is updated using a cluster-based crossover and mutation approach.
In this embodiment, the clustering-based intersection method is:
dividing all tasks into K-type tasks by using a K-means clustering method according to task attributes;
in the crossing process, gene segments with equal length are selected from two different groups of tasks to be crossed. Specifically, a category is selected based on roulette, a task is randomly selected in the category, and the randomly selected task is used as the start of a gene segment. Another category is then selected based on the roulette wheel, and upon selection, the currently selected task category is removed. The probability formula for selecting categories based on roulette is:
Figure BDA0003602687750000131
wherein,
Figure BDA0003602687750000132
is the probability of being of the class k,
Figure BDA0003602687750000133
is task t in category kkAn objective function value of (f)tAnd K is a category set.
The clustering-based mutation method comprises two types, wherein one type is mutation in the same category, the other type is mutation in different categories, and one of two mutation modes is randomly selected to perform mutation operation in each mutation operation;
a variant in the same category refers to a swap of locations of two tasks within an individual that are within the same category.
Variation in different categories refers to swapping the location of two tasks within an individual that are in different categories.
By adopting the two variation modes, the execution position exchange between tasks with similar data characteristics can be realized, and the exchange between the selected task and the task with larger characteristic difference can be completed. The two variation modes can increase the diversity of variation, and the algorithm has better opportunity to find a better task execution sequence. Individuals in the population are updated through clustering-based crossing and variation, and crossing variation is performed on individuals in two different task categories, namely two dissimilar task categories, so that the balance of global search and local search is realized, a solution with higher quality can be obtained, and the problem of sequence dependence is well solved.
Step 3.3.5: if count1Is equal to the threshold value Thre1Updating the parameter K value of the K-means method, clustering the tasks in the task set based on the task attributes again, and resetting the counting parameter count1；
If count2Is equal to Thre2Replacing the new individual with the minimum objective function value in the population with the optimal individual corresponding to the optimal objective function value, and resetting the count parameter count2；
If count3Is equal to Thre3Then, locally optimizing the individual with the maximum objective function value in the new population, randomly generating a new individual, deleting the individual with the minimum objective function value in the new population, and resetting the count parameter count3；
Step 3.3.6: recording the maximum objective function value in the new population as the maximum objective function value of the previous generation population;
the clustering method aims to divide a sample set into a plurality of subsets according to data characteristics. The method divides data into a plurality of groups of discrete or hierarchical structures according to the similarity of data characteristics. The final effect to be achieved by classification is that the samples within a group are as close as possible to each other, not that the samples of a group differ significantly. The clustering algorithm is selected because the satellite task data is required to be fully utilized and the data is utilized to drive the optimization process, and the satellite task data has classification standards which are difficult to determine in the aspect of planning and scheduling, so that the tasks can be classified according to characteristics better by adopting the clustering method. In order to prevent the calculated clustering result from deviation due to over-large dimension of part of the features, the feature data needs to be normalized (normalization). The data can be free from the influence of dimensions and units through normalization, and the content of the data characteristics can be more highlighted. The normalized calculation formula is as follows:
Figure BDA0003602687750000141
wherein y is a characteristic value, ymaxIs the maximum value of the feature y, yminIs the minimum value of the feature y.
The clustering method in the embodiment comprises the following steps:
inputting: task feature matrix (each column represents a feature), number of features, number of classes K
And (3) outputting: clustering results
1): randomly selecting K samples from the feature set as clustering centers (1,2, …, K);
2): calculating Euclidean distances from the data to the central points, and selecting the nearest central point as the category of the data;
3): recalculating each class to obtain new center point according to formula
Figure BDA0003602687750000142
|CjL is the number of data belonging to the j class, and x represents the data;
4): if the central point changes, go to step 2; otherwise, go to step 5;
5): and outputting a clustering result.
In this embodiment, the clustering-based method is to perform clustering based on task features, where the task features include earliest permitted start time, latest permitted end time, benefit, and task requirement measurement and control time length of a task, each column of a task feature matrix represents a task feature to form a matrix, and when re-calculation clustering occurs, a task execution result of an optimal scheme from the beginning of optimization to the present is added to the rightmost column of the feature matrix, that is, when clustering needs to be performed again after a certain condition is reached, whether a task is successfully arranged is increased, the arranged task is marked as 1, and the non-arranged task is marked as 0. The clustering method of the invention divides task data into K categories according to task characteristics, evaluates the close relationship between tasks, enables similar tasks to be as close as possible and different tasks to be as far as possible, and performs cross variation from individuals in two different categories, namely dissimilar two task categories when population is crossed and varied, thereby obtaining a solution with higher quality and improving the optimization process of genetic algorithm. The number of categories of the clustering method changes along with the optimization process, and the information provided by clustering for the genetic algorithm is proved to be more valuable through experiments below.
For the SRSP problem, the number K of clusters into which a cluster is divided should be changed along with the optimization process, and a fixed value is liable to generate invalid search due to inaccurate classification after the optimization process is advanced to a certain stage. In order to improve the search efficiency and the search effect, the K value is updated as count in this embodiment1Is equal to the threshold value Thre1Updating the parameter K value of the K-means method and clustering the tasks in the task set based on the task characteristics, wherein the updating curve of the K value can be preset, and in the embodiment, the classification number K is selected to be decreased by 1 in the way of triggering the change condition number to [ last, next]Interval range constantly changing, KThe process of value change is shown in FIG. 2, where the addition task is successfully executed if the clustering is resumed.
Step 3.4: and outputting the individual corresponding to the optimal objective function value as a measurement and control scheme.
The genetic algorithm based on clustering improves the conditions of better global search and poor local search performance of the genetic algorithm, and improves the local search performance of the genetic algorithm.
And 4, step 4: and outputting the solved measurement and control scheme.
Experimental verification
1) Planning results under different task scales
30 instances of different densities were used to verify the planning performance of the algorithm. As a result, as shown in table 1, the increase in the scale of the example is accompanied by an increase in the value of gain, and in the scene number expression in the leftmost column in table 1, the leftmost three symbols S indicate a small scale, M indicates a medium scale, and L indicates a large scale; the middle two symbols L represent low density, H high density; the rightmost digit represents a serial number. As the size of the task increases to a medium level, the magnitude of the increase in revenue begins to slow down. This is because limited satellite and ground station resources make it particularly difficult to find a better task performance solution among many tasks. The C-BGA algorithm achieved very good planning results in 30 scenarios. In small-scale and medium-scale scenes, the planning result sequentially comprises C-BGA, Firework algorithm FWA (Fireworks algorithm), self-adaptive large neighborhood search algorithm ALNS/TPF (Tabu-based adaptive large neighbor search) with a Tabu strategy and knowledge-based genetic algorithm KBGA (knowledge based genetic algorithm) from good to bad. In large-scale scenarios, the KBGA and ALNS/TPF algorithms are intermediate between the C-BGA and FWA algorithms. As can be seen from the overall trend, the C-BGA algorithm is more likely to achieve performance in larger scale scenarios over other comparative algorithms.
130 scenes income comparison table under different task scales
Figure BDA0003602687750000161
Figure BDA0003602687750000171
2) Algorithm stability analysis
Large-scale instances more readily reflect the stability performance of the algorithm, and L-L-5 and L-H-5 example box plots are shown in FIG. 3. The stability of the C-BGA algorithm is best, maintaining less volatility at an instance of 1000 tasks. ALNS/TPF performs the worst, and results with larger deviation from the mean are most easily obtained. The FWA algorithm and the KBGA algorithm lie between the C-BGA algorithm and the ALNS/TPF algorithm.
3) Optimizing temporal analysis
The time required for the cluster-based genetic algorithm C _ BGA algorithm to achieve the same average optimization result for KBGA is compared here, the results are shown in fig. 4 and 5, fig. 4 and 5 show the medium-scale and large-scale optimization results, respectively, but experimentally, the results are consistent, the 100% bars in the figure represent the time taken for the KBGA algorithm to achieve the average performance of each example, and the black bars represent the percentage of time taken for the cluster-based genetic algorithm C-BGA algorithm of the present invention to take up the reference algorithm. As can be seen from the results, the C-BGA algorithm only requires approximately one third of the time for the KBGA to achieve the same optimization. That is, a good quality solution can be found a short time after population initialization.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and these modifications or substitutions do not depart from the spirit of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A satellite measurement and control scheduling method based on clustering is characterized by comprising the following steps:
step 1: acquiring schedulable satellite resources, ground station resources, a task set and a visible time window set;
step 2: constructing a mixed integer programming model according to the content acquired in the step 1;
and step 3: solving the mixed integer programming model;
and 4, step 4: and outputting the solved measurement and control scheme.
2. The method of claim 1, wherein the mixed integer programming model is:
the objective function is:
Figure FDA0003602687740000011
wherein p istThe benefit of the task t is represented,
Figure FDA0003602687740000012
a decision variable of 0-1, indicating whether a task T is scheduled on the antenna a in the ith visible time window, 1 is scheduled, 0 is no, T indicates a task set, a indicates an antenna set,
Figure FDA0003602687740000013
indicating that task t is scheduled on antenna a for the ith visible time window, TWtaA set of time windows representing the scheduling of task t on antenna a;
the constraint conditions are:
Figure FDA0003602687740000014
equation 2 indicates that the task needs to start executing the task after the allowed start time;
Figure FDA0003602687740000015
representing the actual measurement and control time length of the task t arranged on the ith visible time window on the antenna a, dtRepresenting the measurement and control time length required by the task t;
Figure FDA0003602687740000016
equation 3 indicates that the task needs to be completed before the allowed end time;
Figure FDA0003602687740000017
representing the actual start time of task t scheduled on antenna a on the ith visible time window,
Figure FDA0003602687740000018
represents the end time of the task t arranged on the ith visible time window on the antenna a;
Figure FDA0003602687740000021
equation 4 indicates that a task can be performed at most once;
Figure FDA0003602687740000022
represents the earliest permitted start time of the task t scheduled on the ith visible time window on the antenna a;
Figure FDA0003602687740000023
equation 5 indicates that the actual execution time of the task should be the same as the required time;
Figure FDA0003602687740000024
the latest allowed end time of the task t on the ith visible time window of the antenna a;
Figure FDA0003602687740000025
equation 6 indicates that the task execution process needs to be within a ground station time range;
Figure FDA0003602687740000026
represents the start time of the ith visible time window of the task t scheduled on the antenna a;
Figure FDA0003602687740000027
equation 7 indicates that each task can only be served by one antenna;
Figure FDA0003602687740000028
equation 8 indicates that each task can only be executed within one visible time window;
Figure FDA0003602687740000029
indicating the end time of the ith visible time window of the task t arranged on the antenna a;
Figure FDA00036026877400000210
equation 9 indicates that each task is performed on all antennas at most once;
Figure FDA00036026877400000211
equation 10 indicates that each task is executed at most once in the entire time window;
Figure FDA00036026877400000212
equation 11 indicates that the task is executed at most once during the planning period;
Figure FDA00036026877400000213
equation 12 indicates that the two tasks are to meet the interval requirement of task switching time; γ represents the transition time between tasks.
3. The method of claim 2, wherein the method of solving the mixed integer programming model in step 3 is a clustering-based genetic algorithm.
4. The method of claim 3, wherein the cluster-based genetic algorithm is:
step 3.1: initializing genetic algorithm parameters and clustering K-means method parameters;
step 3.2: generating an initialization population, wherein each individual in the population is a coded sequence obtained by coding all tasks in a task set;
step 3.3: when the iteration algebra does not reach the maximum iteration algebra, executing a step 3.3.1; otherwise, executing step 3.4;
step 3.3.1: generating an initial solution of a measurement and control scheme for each individual in the population and calculating a target function value;
step 3.3.2: if the maximum objective function value in the present generation population is greater than the optimal objective function value, updating the optimal objective function value and counting the count1＝count1+1；
If the maximum objective function value in the current generation population is larger than the maximum objective function value in the previous generation population, the count is updated2＝count2+1；
If the maximum objective function value in the current generation population is less than the maximum function value in the previous generation population multiplied by the proportion per, the count is updated3＝count3+1；
Step 3.3.3: selecting individuals to generate a new population by using a roulette wheel according to the objective function value;
step 3.3.4: updating the new population by using a clustering-based intersection and variation method;
step 3.3.5: if count1Is equal to the threshold value Thre1Updating the parameter K value of the K-means method, clustering the tasks in the task set based on the task attributes again, and resetting the counting parameter count1；
If count2Is equal to Thre2Replacing the new individual with the minimum objective function value in the population with the optimal individual corresponding to the optimal objective function value, and resetting the count parameter count2；
If count3Is equal to Thre3Then, locally optimizing the individual with the maximum objective function value in the new population, randomly generating a new individual, deleting the individual with the minimum objective function value in the new population, and resetting the count parameter count3；
Step 3.3.6: recording the maximum objective function value in the new population as the maximum objective function value of the previous generation population;
step 3.4: and outputting the individual corresponding to the optimal objective function value as a measurement and control scheme.
5. The method according to claim 4, characterized in that in step 3.3.1: the method for generating the initial solution of the measurement and control scheme for each individual in the population is a task arrangement algorithm, and specifically comprises the following steps:
all tasks in the task set are tried and arranged to each task according to the coding sequence and the visible time window sequence in turn, if all time windows are tried and arranged successfully, the result of successful arrangement is output as an initial solution, otherwise, the arrangement is tried continuously; the specific arrangement method for each task comprises the following steps:
1): calculating the earliest actual available time eat of task ttAnd the latest actual available time latt；
2): if the length of the ith time window of the task t on the antenna a is larger than the requirement of the task tControlling the length of time, i.e.
Figure FDA0003602687740000041
And the actual available time length of the task t is longer than the measurement and control time length of the task t, namely (lat)t-eatt)≥dtThen go to step 3); otherwise, continuing to schedule the next task;
3): earliest task t to actual available time eattAs the task start time, if the task t is the earliest the actual available time eattEqual to the start time of the ith visible time window of the task t scheduled on the antenna a
Figure FDA0003602687740000042
Go to step 4); otherwise, go to step 5);
4): scheduling task t on antenna a for the ith visible time window
Figure FDA0003602687740000043
Updating the remaining time window of the window to a new time window
Figure FDA0003602687740000044
5): scheduling task t on antenna a for the ith visible time window
Figure FDA0003602687740000045
The remaining time window of the window is updated into two new time windows
Figure FDA0003602687740000046
And
Figure FDA0003602687740000047
6. the method of claim 4, wherein the method of generating the initialization population in step 3.2 is to generate the initialization population using a plurality of heuristic initialization methods, wherein individuals generated by the various heuristic initialization methods are present in the initialization population.
7. The method according to claim 4, wherein using a clustering based intersection method in step 3.3.4 is:
dividing all tasks into K-type tasks by using a K-means clustering method according to task attributes;
when in crossing, the gene segments with the same length are respectively selected from two groups of tasks with different categories to be crossed, and the categories of the tasks are selected according to roulette.
8. The method of claim 7, wherein the clustering-based mutation method used in step 3.3.4 comprises two types, one type is mutation in the same category, and the other type is mutation in different categories, and one of the two types of mutation is randomly selected to perform mutation operation in each mutation operation;
a variant in the same category refers to a swap of the location of two tasks within an individual that are within the same category.
Variation in different categories refers to swapping the location of two tasks within an individual that are in different categories.
9. The method according to claim 4, wherein the clustering based method in step 3.1 is based on clustering of task features including earliest allowed start time, latest allowed end time, profit, task requirement measure and control time length of task, and when K value update is clustered again, it is added whether the task is scheduled successfully.
10. A satellite measurement and control scheduling system based on clustering is characterized by comprising the following modules:
an input module: the system comprises a time window set, a resource acquisition module, a task acquisition module and a scheduling module, wherein the time window set is used for acquiring schedulable satellite resources, ground station resources, a task set and a time window set for a satellite to see a ground station;
a model construction module: the method is used for constructing a mixed integer programming model according to the content acquired in the step 1;
a solution module: for solving the mixed integer programming model;
a scheme output module: and outputting the solved measurement and control scheme.
CN202210407679.0A 2022-03-28 2022-04-19 Clustering-based satellite measurement and control scheduling method and system Pending CN114781508A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210308731 2022-03-28
CN2022103087317 2022-03-28

Publications (1)

Publication Number Publication Date
CN114781508A true CN114781508A (en) 2022-07-22

Family

ID=82431851

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210407679.0A Pending CN114781508A (en) 2022-03-28 2022-04-19 Clustering-based satellite measurement and control scheduling method and system

Country Status (1)

Country Link
CN (1) CN114781508A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115460304A (en) * 2022-11-10 2022-12-09 广州铭创通讯科技有限公司 Protocol layer data analysis method and system for intercepting wireless communication

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115460304A (en) * 2022-11-10 2022-12-09 广州铭创通讯科技有限公司 Protocol layer data analysis method and system for intercepting wireless communication
CN115460304B (en) * 2022-11-10 2023-01-31 广州铭创通讯科技有限公司 Protocol layer data analysis method and system for intercepting wireless communication

Similar Documents

Publication Publication Date Title
CN111913787B (en) Imaging satellite scheduling method and system based on genetic algorithm
CN111176807A (en) Multi-satellite cooperative task planning method
CN109544998B (en) Flight time slot allocation multi-objective optimization method based on distribution estimation algorithm
CN109039428B (en) Relay satellite single-address antenna scheduling random search method based on conflict resolution
CN112434435B (en) Imaging satellite intensive task scheduling method based on task synthesis
CN117077981B (en) Method and device for distributing stand by fusing neighborhood search variation and differential evolution
CN113204417A (en) Multi-satellite multi-point target observation task planning method based on improved genetic and firefly combined algorithm
CN114781508A (en) Clustering-based satellite measurement and control scheduling method and system
Fakhrzad et al. A new multi-objective job shop scheduling with setup times using a hybrid genetic algorithm
CN116938323B (en) Satellite transponder resource allocation method based on reinforcement learning
CN111831421B (en) Task allocation method and terminal equipment
Ying et al. Hybrid-directional planning: improving improvement heuristics for scheduling resource-constrained projects
D’Ambrosio et al. Optimizing cellular automata through a meta-model assisted memetic algorithm
CN113807646B (en) Intelligent optimization method and system for unmanned aerial vehicle and vehicle cooperative task allocation
CN115169754A (en) Energy scheduling method and device, electronic equipment and storage medium
CN115296717A (en) Scheduling method of ground station antenna array of low-earth-orbit satellite system
Azizi et al. Satellite broadcast scheduling based on a boosted binary differential evolution
CN113487031A (en) Multi-unmanned aerial vehicle task allocation method based on improved simulated annealing fusion genetic algorithm
Chang et al. Multiobjective permutation flow shop scheduling using MOEA/D with local search
CN113141272A (en) Network security situation analysis method based on iteration optimization RBF neural network
CN117579500B (en) Network traffic prediction method, device, equipment and medium
JP2003016421A (en) Optimization problem processor
CN111800310B (en) Scheduling method for task management module of Internet of things test cloud platform
CN116629586A (en) Airport guarantee vehicle scheduling method and system based on ALNS
CN109039920B (en) Method for constructing a plurality of minimum spanning trees with depth and degree constraint

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination