CN117271146B - Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene - Google Patents
Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene Download PDFInfo
- Publication number
- CN117271146B CN117271146B CN202311567217.6A CN202311567217A CN117271146B CN 117271146 B CN117271146 B CN 117271146B CN 202311567217 A CN202311567217 A CN 202311567217A CN 117271146 B CN117271146 B CN 117271146B
- Authority
- CN
- China
- Prior art keywords
- task
- imaging
- knowledge
- target
- satellite
- 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.)
- Active
Links
- 238000003384 imaging method Methods 0.000 title claims abstract description 229
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012546 transfer Methods 0.000 title claims abstract description 27
- 238000013210 evaluation model Methods 0.000 claims abstract description 17
- 238000005457 optimization Methods 0.000 claims description 17
- 230000007613 environmental effect Effects 0.000 claims description 11
- 238000004891 communication Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 241000820057 Ithone Species 0.000 claims 1
- 230000000295 complement effect Effects 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 14
- 238000013468 resource allocation Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention relates to the technical field of aviation, and particularly discloses a multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes, which comprises the following steps: s10, constructing a task evaluation model according to the imaging requirements of the user and task planning scene information; s20, evaluating the imaging requirements of the user according to a task evaluation model, and generating a multi-star imaging task carrying execution priority information; s30, constructing a multi-star imaging task planning model under a complex heterogeneous scene; s40, determining the corresponding relation between the imaging satellites and the target tasks according to the number of the imaging satellites and the number of the target tasks to be observed currently; s50, calculating a task benefit value of the imaging satellite for executing each target task according to the task benefit objective function; s60, storing the task income value into a task income element knowledge storage library of the imaging satellite and the target task; s70, outputting a multi-star imaging mission planning scheme under a complex heterogeneous environment according to the multi-target evolutionary algorithm and the multi-star imaging mission planning model.
Description
Technical Field
The invention relates to the technical field of aviation, in particular to a multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes.
Background
With the rapid development of aerospace technology, the number of satellites in orbit is rapidly increased, and higher requirements on satellite observation quality and response time are put forward on earth observation tasks. Compared with a single satellite, a plurality of satellites can realize long-time effect and multidirectional continuous monitoring on an observation area through mutual cooperation. In addition, with the practice and development of multi-satellite mission planning systems of 'Beidou satellite', 'high-resolution satellite' and other series of satellites, the multi-satellite imaging mission planning technology plays an increasingly important role in the fields of geographic mapping, land resource investigation, disaster monitoring and the like.
The multi-satellite imaging task planning is to allocate one or more target tasks for a plurality of imaging satellites according to known environmental information and the earth observation requirements of users on the basis of optimal allocation and aiming at the constraints and cooperative constraints of the imaging satellites, and generate a multi-satellite imaging task planning scheme with optimal target functions, so that the maximum observation benefits of the multi-satellite imaging tasks are realized while the requirements of the users are met, the repeated allocation and resource conflict of the multi-satellite imaging tasks are avoided, and the satellite resource waste is reduced.
Heterogeneity refers to study variability caused by certain features during the course of a study, which requires identification and formulation of corresponding strategies. The multi-satellite imaging task planning has a plurality of task demand relations among multi-user demands, multi-satellites and multi-tasks in complex heterogeneous scenes, and the conventional multi-satellite task planning method only solves the problem that the multi-satellite imaging task planning demands in complex heterogeneous scenes cannot be met in a mode of solving a specific scene. How to improve the expandability of a multi-satellite imaging task planning model under a complex heterogeneous scene and efficiently process various task demand relations of imaging satellites and target tasks under the complex heterogeneous scene so as to fully exert limited satellite resources is a serious challenge facing the satellite task planning field at present.
In addition, in the multi-satellite imaging task planning process under a complex heterogeneous scene, the method has the characteristics of poor self-maneuvering performance of the satellite, limited imaging times and the like, and the multi-imaging satellite has complex constraint conditions in a task environment, and has the problems of low execution efficiency, multiple execution schemes, long planning time and the like due to explosive increase of information quantity along with increase of task planning time. How to construct an algorithm for quickly and accurately solving a multi-star imaging task scheme is another challenge for solving a multi-star imaging task planning problem in a complex heterogeneous scene.
Disclosure of Invention
Aiming at the problems, the invention aims to provide a multi-satellite imaging task planning method based on knowledge transfer under a complex heterogeneous scene, which solves the problems of inconsistent multi-user requirements, multi-satellite and multi-task multi-satellite imaging task planning models, long task planning time and conflict in satellite resource allocation under the complex heterogeneous scene, can realize different task requirements in a short time by limited satellite resources, and improves the operation efficiency of acquiring remote sensing images by imaging satellites.
The invention provides a multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes, which comprises the following steps:
step S10, constructing a task evaluation model according to the imaging requirements of the user and task planning scene information;
step S20, evaluating the imaging requirements of the user according to the task evaluation model, and generating a multi-star imaging task carrying execution priority information;
step S30, constructing a multi-star imaging task planning model under a complex heterogeneous scene according to the execution decision variables, the task benefit objective function, the response time objective function, the global optimization objective function, the imaging satellite self-performance constraint and the objective task allocation constraint of the multi-star imaging task;
step S40, determining the corresponding relation between the imaging satellites and the target tasks according to the number of the imaging satellites and the number of the target tasks to be observed currently;
step S50, calculating a task gain value of the imaging satellite for executing each target task according to a task gain objective function;
step S60, storing the task income value into a task income element knowledge storage library of the imaging satellite and the target task according to the corresponding relation between the imaging satellite and the target task;
and step S70, constructing a multi-target evolutionary algorithm based on knowledge transfer according to the satellite, the target task, the task profit value and the response time, and outputting a multi-star imaging task planning scheme under a complex heterogeneous environment according to the multi-target evolutionary algorithm and the multi-star imaging task planning model.
In one possible implementation, the user imaging requirements include: the method comprises the following steps of target task type, target task state, target task geometric position, image resolution, target task level and target task emergency degree; the task planning scene information includes: a multi-satellite imaging task planning period, an observation target task scale and an execution task imaging satellite scale; the step S10 includes:
constructing a task assessment model according to the following formula：
；
Wherein,for the imaging needs of the user,the type is required for the imaging of the user,for the number of types of imaging requirements of the user,the type of task that is to be targeted is indicated,the status of the target task is indicated,the geometric position of the target task is represented,representing the user's demand for image resolution,representing the emergency degree of the target task;
the judgment matrix of the task evaluation model has the following formula:
；
the task weight coefficient calculation function of the task evaluation model has the following formula:
；
wherein,representing the first of the task priority impact factorsA plurality of;
calculating the priority of the multi-star imaging task according to the following formula：
；
Wherein,for the relative importance between index i and index j,
。
in one possible implementation manner, the step S30 includes:
the decision variables of the multi-star imaging task are as follows:
；
where S represents the number of imaging satellites, T represents the number of observation target tasks,representing satellitesExecuting observation target tasksDecision variables of (2);
the task income objective functionThe following formula is given:
；
wherein,the task revenue function is represented as a function of the task,representing the decision variables for the execution of the multi-star imaging task,representing task priority;
the response time objective functionThe following formula is given:
；
wherein,representing execution of observation target tasksIs a response time of (2);
the global optimization objective functionThe following formula is given:
。
in one possible implementation manner, the step S30 further includes:
the imaging satellite self-performance constraint is as follows:
；
wherein,representing a track，Representing imaging satellitesOn the trackObservation targetIs used for the end time of (c),representing imaging satellites respectivelyOn the trackObservation targetIs used for the start time of (1),representing imaging satellitesIs used for the data transfer rate of (a),representing the maximum storage capacity of the imaged satellite payload,representing imaging satellitesIs indicative of the communication end time and start time of the imaging satelliteIndicating the end time of the communication of the imaging satellite,respectively representing communication start times of imaging satellites;
the target task allocation constraint is as follows:
；
wherein,representing imaging satellitesOn the trackObservation targetIs used for the observation time window of (a),represents the task yaw angle of the observation target,representing the maximum roll angle of the imaging satellite,representing imaging satellitesIs used for the resolution of the remote sensor,indicating that the target task requires a minimum resolution,representing a single field angle of view of the satellite remote sensor.
In one possible implementation manner, the step S40 includes:
when the number of the imaging satellites is consistent with the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is one-to-one correspondence;
when the number of the imaging satellites is smaller than the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is one-to-many;
when the number of the imaging satellites is larger than the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is many-to-one.
In one possible implementation, the task revenue metadata repository includes a first sub-repository and a second sub-repository; the row index of the task income value is a sub-library number; the step S60 includes:
the row index of the first sub-library is a target task number, and the column index is a satellite number;
the first sub-library has meta-knowledge ofFor imaging satellitesExecuting target tasksThe formula is as follows:
；
wherein,representing imaging satellitesExecuting target tasksTask benefit values of (2);
the row index of the second sub-library is a target task number executed for the first time, and the column index is a target task number executed again;
meta knowledge of the second sub-libraryFor imaging satellitesExecuting target tasksCompleting by target taskPerforming the target task for the originThe formula is as follows:
；
wherein,representing imaging satellites performing target tasksCompleting by target taskPerforming the target task for the originS represents the number of imaging satellites and T represents the number of observation target tasks.
In one possible implementation manner, the step S70 includes:
step S71, randomly selecting a first amount of knowledge from the task income element knowledge storage library as an initialization environment; the first number is less than the total knowledge number in the task revenue metadata repository; the satellite, the target mission, the mission benefits, the response time are uniformly described as knowledge according to the following formula:
；
wherein,knowledge storage for task incomeV th generation of (2)Knowledge, in particular satelliteExecuting target tasksThe task profit value isResponse time is as follows；
Step S72, calculating the crowding distance between the knowledge; the knowledge is satellite and/or target mission and/or mission gain and/or response time;
step S73, dividing the knowledge into useful knowledge and sub-useful knowledge according to the crowding distance between the knowledge, and storing the useful knowledge and the sub-useful knowledge in a useful knowledge niche and a sub-useful knowledge niche respectively, wherein the specific formulas are as follows:
；
step S74, calculating the environmental variability of the knowledge according to the global optimization objective function value corresponding to the knowledge in the environment;
and step S75, a knowledge transfer strategy is adaptively selected according to knowledge difference, and a multi-star imaging task planning scheme under a complex heterogeneous environment is output.
In one possible implementation manner, the step S72 includes:
the crowding distance between the knowledge is calculated according to the following formula:
；
wherein,representing the minimum euclidean distance between the knowledge,representing the maximum euclidean distance between the knowledge,representing in a historical environmentThe average of the optimal solution in each dimension.
In one possible implementation, the step S74 includes:
the knowledge difference between the environments is calculated according to the following formula:
；
wherein,represent the firstThe environmental variability of the individual knowledge is that,representing the first in a historical environmentKnowledge ofGlobal optimization objective function values corresponding to the respective positions,representing the first in the current environmentKnowledge ofGlobal optimization objective function values corresponding to the respective positions.
In one possible implementation manner, the step S75 includes:
if it isAnd calculating the crowding distance between the knowledge again, and adaptively selecting to directly transfer the knowledge corresponding to the historical environment to the current environment according to the actual user requirements:
；
wherein the historical environmental knowledge is used as source dataKnowledge of the current environmentAs the data to be used for the purpose,representing randomly selected knowledge in the environment;
if it isAdopting a subspace distribution alignment method to mine association rules between the historical environment and the current environment, and transferring knowledge from the historical environment to the current environment;
the subspace distribution alignment method is as follows:
wherein,is thatIs used for the feature vector of (a),is thatIs used for the feature vector of (a),is thatIs a set of orthogonal complements of (a),、respectively areAndis a matrix of orthogonality of the (c),、、is thatAnd (3) withIs a diagonal matrix of (a);
；
wherein,represents the v th generationKnowledge of the individual;
and outputting a multi-star imaging task planning scheme under the complex heterogeneous environment until the termination condition is reached.
The multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes has the following beneficial effects:
1) Aiming at the actual requirements of multi-star multi-task in a complex heterogeneous scene, a task evaluation model is built, target tasks which arrive in batches are converted into a series of task sets carrying priority information, loss caused by unreasonable satellite resource allocation is reduced, and a more reasonable decision basis is provided for the target task observation sequence of multi-star imaging task planning in the complex heterogeneous scene.
2) Aiming at the conditions of multiple users, multiple satellites and multiple tasks in complex heterogeneous scenes. On the basis of the optimization theory, a multi-star imaging task planning model under a complex heterogeneous scene is constructed, and the problems that model expansibility in task allocation is poor, algorithm pertinence is weak and the like under the actual demands of different users in multi-star imaging task planning under the complex heterogeneous scene are solved.
3) The task income element knowledge storage library of the satellite and the target task is constructed, the task income is defined as an element knowledge, and the element knowledge is used as a unified medium of multiple satellites and multiple tasks of different types, so that the task income value can be directly called from the task income element knowledge storage library in the process of optimizing and searching, the repeated calculation of the task income value is reduced, and the task planning efficiency is improved.
4) Aiming at the problems of complex constraint conditions, low execution efficiency, incomplete execution scheme and the like of a multi-imaging satellite in a complex heterogeneous scene, a multi-objective evolutionary algorithm based on knowledge transfer is constructed, the crowded distance between the knowledge is calculated to divide the knowledge, the knowledge difference between the current environment and the historical environment is evaluated, the knowledge transfer strategy is adaptively selected, the historical knowledge is adapted to the current search space, the convergence speed is increased, and a more reasonable multi-star imaging task planning scheme in the complex heterogeneous environment is output for a decision maker to select.
Drawings
FIG. 1 is a flow chart of a multi-star imaging task planning method according to an embodiment of the present invention;
FIG. 2 (a) is a schematic diagram of one-to-one correspondence between imaging satellites and target tasks according to an embodiment of the present invention;
FIG. 2 (b) is a schematic diagram of one-to-many imaging satellites and target tasks according to an embodiment of the present invention;
FIG. 2 (c) is a schematic diagram of an imaging satellite and a target mission in many-to-one manner according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a task revenue metadata repository provided by an embodiment of the present invention;
fig. 4 (a) is a schematic diagram of a target task matching policy when s=t provided in an embodiment of the present invention;
fig. 4 (b) is a schematic diagram of a target task matching policy when S < T provided in an embodiment of the present invention;
fig. 4 (c) is a schematic diagram of a target task matching policy when S > T is provided in an embodiment of the present invention.
Detailed Description
Embodiments of the present invention are described in further detail below with reference to the accompanying drawings and examples. The following detailed description of the embodiments and the accompanying drawings are provided to illustrate the principles of the invention and are not intended to limit the scope of the invention, i.e. the invention is not limited to the preferred embodiments described, which is defined by the claims.
In the description of the present invention, it is to be noted that, unless otherwise indicated, the meaning of "plurality" means two or more; the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance; the specific meaning of the above terms in the present invention can be understood as appropriate by those of ordinary skill in the art.
Fig. 1 is a flow chart of a multi-star imaging task planning method according to an embodiment of the present invention, as shown in fig. 1, the multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenarios provided by the present invention includes:
step S10, constructing a task evaluation model according to the imaging requirements of the user and task planning scene information;
in one possible implementation manner, the complex heterogeneous scene has multi-user, multi-satellite and multi-task conditions, and in order to complete the information transmission of multi-user multi-satellite, a ground control center sends the imaging requirements of the user and task planning scene information to each imaging satellite.
Wherein, user imaging requirements include: the method comprises the following steps of target task type, target task state, target task geometric position, image resolution, target task level and target task emergency degree; the task planning scene information includes: a multi-satellite imaging mission planning period, an observation target mission scale and an execution mission imaging satellite scale.
And the ground control center sends the imaging requirements of the user and task planning scene information to each imaging satellite to finish the information transmission of multi-user multi-satellite under the complex heterogeneous scene. According to different imaging requirements of users and different task planning scene information, a task evaluation model is constructed according to the following formula：
；
Wherein,for the imaging needs of the user,for the type of imaging requirements of the user,for the type number of the imaging demands of the user, the above formula performs unified mathematical description on the task priority influence factorsThe type of task that is to be targeted is indicated,the status of the target task is indicated,the geometric position of the target task is represented,representing the user's demand for image resolution,indicating the urgency of the target task.
The judgment matrix of the task evaluation model has the following formula:
；
the task weight coefficient calculation function of the task evaluation model has the following formula:
；
wherein,representing the first of the task priority impact factorsA plurality of;
prioritizing multiple rows of imaging tasks according to the following formula：
；
Wherein,for the relative importance between index i and index j,
。
step S20, evaluating the imaging requirement of the user according to a task evaluation model, and generating a multi-star imaging task carrying execution priority information;
in one possible implementation manner, a task evaluation model is constructed to evaluate the imaging requirements of the user according to the difference of the target task type, the target task state, the geometric position, the user requirement image resolution and the target task emergency degree, and a task set carrying the execution priority information is generated.
Step S30, constructing a multi-star imaging task planning model under a complex heterogeneous scene according to an execution decision variable, a task benefit objective function, a response time objective function, a global optimization objective function, an imaging satellite self-performance constraint and a target task allocation constraint of the multi-star imaging task;
in one possible implementation, the decision variables for performing the multi-star imaging task are as follows:
；
wherein S represents an imaging deviceThe number of stars, T, represents the number of observation target tasks,representing satellitesExecuting observation target tasksDecision variables of (2);
task benefit objective functionThe following formula is given:
；
wherein,the task revenue function is represented as a function of the task,representing the decision variables for the execution of the multi-star imaging task,representing task priority;
response time objective functionThe following formula is given:
；
wherein,representing execution of observation target tasksIs a response time of (2);
globally optimizing objective functionsThe following formula is given:
。
the performance constraint of the imaging satellite mainly considers the load capacity of the imaging satellite, the single imaging observation time of the imaging satellite and the single imaging communication time of the imaging satellite. The imaging satellite itself performance constraints are as follows:
；
wherein,representing a track，Representing imaging satellitesOn the trackObservation targetIs used for the end time of (c),representing imaging satellites respectivelyOn the trackObservation targetIs used for the start time of (1),representing imaging satellitesIs used for the data transfer rate of (a),representing the maximum storage capacity of the imaged satellite payload,representing imaging satellitesIs indicative of the communication end time and start time of the imaging satelliteIndicating the end time of the communication of the imaging satellite,respectively representing communication start times of imaging satellites;
the target task allocation constraint mainly considers the roll angle constraint, the resolution constraint and the view angle constraint. The target task allocation constraint is as follows:
；
wherein,representing imaging satellitesOn the trackObservation targetIs used for the observation time window of (a),represents the task yaw angle of the observation target,representing the maximum roll angle of the imaging satellite,representing imaging satellitesIs used for the resolution of the remote sensor,indicating that the target task requires a minimum resolution,representing a single field angle of view of the satellite remote sensor.
Step S40, determining the corresponding relation between the imaging satellites and the target tasks according to the number of the imaging satellites and the number of the target tasks to be observed currently;
in one possible implementation, whenThat is, when the number of imaging satellites is consistent with the number of target tasks to be observed currently, the corresponding relationship between the imaging satellites and the target tasks is one-to-one, as shown in fig. 2 (a), and fig. 2 (a) is a schematic diagram of one-to-one correspondence between the imaging satellites and the target tasks provided in the embodiment of the present invention. The imaging satellite performs only a single target task, which is observed only once.
When (when)That is, the number of imaging satellites is smaller than the number of target tasks to be observed currently, and the corresponding relationship between the imaging satellites and the target tasks is one-to-many, as shown in fig. 2 (b), and fig. 2 (b) is a schematic diagram of one-to-many imaging satellites and target tasks provided in the embodiment of the present invention. The limited satellite resources can not meet all requirements of users, a plurality of satellites are reasonably distributed to execute target tasks, and the same imaging satellite is plannedDifferent target tasks can be executed for many times in the inter-period;
when (when)That is, the number of imaging satellites is greater than the number of target tasks to be observed currently, and the corresponding relationship between the imaging satellites and the target tasks is many-to-one, as shown in fig. 2 (c), and fig. 2 (c) is a schematic diagram of many-to-one of the imaging satellites and the target tasks provided in the embodiment of the present invention. And observing a complex target task, wherein a single target task is observed by a plurality of imaging satellites, and the captured images are compared to obtain three-dimensional information of the observation task.
Constructing an adaptive target task matching strategy, which specifically comprises the following steps ofThe target task is matched to the policy and,the target task is matched to the policy and,target task matching policy.
Fig. 4 (a) is a schematic diagram of a target task matching policy when s=t provided in the embodiment of the present invention, as shown in fig. 4 (a),target task matching strategy: the imaging satellites and the target tasks are in one-to-one correspondence distribution relation, a single satellite in knowledge is required to be matched with a single target task, a task income element knowledge storage library is indexed according to satellite numbers, a target task number with the largest task income value is matched, the satellite, the target task and the task income are stored as a piece of knowledge, the knowledge is stored in a multi-satellite imaging task planning knowledge library, and the rows and columns where the task income values corresponding to the satellite and the target task are located are deleted in the task income element knowledge storage library.
FIG. 4 (b) shows an embodiment of the invention<A schematic diagram of the target task matching strategy at T, as shown in figure 4 (b),target task matching strategy: the imaging satellite and the target task are distributed in a one-to-many mode, a plurality of target tasks are matched for a single satellite in knowledge, and target matching is needed to be carried out for a plurality of times. The sub-library 1 is indexed, a target task with the largest task profit value is matched for a single satellite to serve as a first execution target task, the satellite, the target task and task profit are stored as a piece of knowledge, and rows and columns where the upper half part corresponds to the global optimization target value are deleted. And then, indexing the sub-library 2 by the satellite number, re-executing the target task with the maximum matching task profit value, storing the target task as a new knowledge in a multi-star imaging task planning knowledge base, and deleting the row and column where the corresponding global optimization target value is located.
FIG. 4 (c) shows an embodiment of the invention>A schematic diagram of the target task matching strategy at T, as shown in figure 4 (c),target task matching strategy: the distribution relation between the imaging satellites and the target tasks is many-to-one, a plurality of satellites can execute the same target task in knowledge, and a single target task is executed by at least 1 imaging satellite, so that target matching is required to be carried out for a plurality of times. And indexing the sub-library 1 according to the target task number, performing the matching of the satellite with the biggest task profit value for a single target task, storing the target task and the imaging satellite as a new knowledge into a multi-satellite imaging task planning knowledge base, and deleting the row where the matching satellite is located. And (5) circularly matching until all satellites obtain matching target tasks.
Step S50, calculating a task gain value of the imaging satellite for executing each target task according to the task gain objective function;
step S60, storing the task income value into a task income element knowledge storage library of the imaging satellite and the target task according to the corresponding relation between the imaging satellite and the target task;
the task income element knowledge storage library comprises a first sub-library and a second sub-library; the row index of the task benefit value is the sub-library number.
Fig. 3 is a schematic diagram of a task revenue metadata repository provided in an embodiment of the present invention, where, as shown in fig. 3, a row index of a first sub-repository is a target task number, and a column index is a satellite number.
Meta-knowledge of first sub-libraryFor imaging satellitesExecuting target tasksThe formula is as follows:
；
wherein,representing imaging satellitesExecuting target tasksTask benefit values of (2);
the row index of the second sub-library is the target task number executed for the first time, and the column index is the target task number executed again;
meta-knowledge of the second sub-libraryFor imaging satellitesExecuting target tasksCompleting by target taskPerforming the target task for the originThe formula is as follows:
；
wherein,representing imaging satellites performing target tasksCompleting by target taskPerforming the target task for the originS represents the number of imaging satellites and T represents the number of observation target tasks.
According to the corresponding relation between the imaging satellite and the target task, the task benefit value can be directly called from the task benefit meta-knowledge storage in the process of optimizing search, so that repeated calculation of task benefit is reduced.
And step S70, constructing a multi-target evolutionary algorithm based on knowledge transfer according to the satellite, the target task, the task profit value and the response time, and outputting a multi-star imaging task planning scheme under a complex heterogeneous environment according to the multi-target evolutionary algorithm and the multi-star imaging task planning model.
Step S71, randomly selecting a first amount of knowledge from a task income element knowledge storage library as an initialization environment;
wherein the first quantity is less than the total knowledge quantity in the task revenue metadata repository.
In one example, the total knowledge quantity in the task revenue metadata repository is N, the first quantity is 0.1N, and N is a positive integer.
The satellite, target mission, mission benefits, and response time are uniformly described as knowledge according to the following formula:
；
wherein,knowledge storage for task incomeV th generation of (2)Knowledge, in particular satelliteExecuting target tasksThe task profit value isResponse time is as follows；
Step S72, calculating the crowding distance between the knowledge; the knowledge is satellite and/or target mission and/or mission gain and/or response time;
the crowding distance between the knowledge is calculated according to the following formula:
；
wherein,representing the minimum euclidean distance between the knowledge,representing the maximum euclidean distance between the knowledge,representing in a historical environmentThe average of the optimal solution in each dimension.
Step S73, dividing the knowledge into useful knowledge and sub-useful knowledge according to the crowding distance between the knowledge, and storing the useful knowledge and sub-useful knowledge in the useful knowledge niche and the sub-useful knowledge niche respectively, wherein the specific formulas are as follows:
；
step S74, calculating the environmental variability of the knowledge according to the global optimization objective function value corresponding to the knowledge in the environment;
the knowledge difference between the environments is calculated according to the following formula:
；
wherein,represent the firstThe environmental variability of the individual knowledge is that,representing the first in a historical environmentKnowledge ofGlobal optimization objective function values corresponding to the respective positions,representing the first in the current environmentKnowledge ofGlobal optimization objective function values corresponding to the respective positions.
And step S75, a knowledge transfer strategy is adaptively selected according to knowledge difference, and a multi-star imaging task planning scheme under a complex heterogeneous environment is output.
If it isThe method shows that the current environment is similar to the historical environment knowledge, a direct knowledge transfer strategy is adopted, namely, the crowding distance between the knowledge is calculated again, and the corresponding knowledge of the historical environment is directly transferred to the current environment according to the self-adaptive selection of the actual user demands:
；
wherein the historical environmental knowledge is used as source dataKnowledge of the current environmentAs the data to be used for the purpose,representing randomly selected knowledge within the environment.
Aiming at periodical conventional large-scale imaging task demands, maximizing task income, adaptively selecting a secondary useful knowledge niche in a historical environment to transfer the secondary useful knowledge to a current environment so as to promote the richness of a multi-star imaging task planning scheme, and helping a decision maker to search an execution scheme for maximizing task income; aiming at emergency guarantee task requirements of emergency situations in small-scale areas such as earthquake, forest fire and the like, the average response time of the task is required to be shortest, and useful knowledge niches in the history environment are adaptively selected to transfer useful knowledge to the current environment so as to guide the knowledge to quickly approachSo that reasonable multi-star imaging mission planning candidates are searched for in a shorter response time.
If it isIndicating that the current environment has larger difference from the historical environment, the knowledge in the historical environment is closer to the real environment than the knowledge in the current environmentAnd selecting a cross-environment knowledge transfer strategy, namely adopting a subspace distribution alignment method to mine association rules between the historical environment and the current environment, and transferring the knowledge from the historical environment to the current environment.
The subspace distribution alignment method is as follows:
；
wherein,is thatIs used for the feature vector of (a),is thatIs described.Is thatIs a set of orthogonal complements of (a),、respectively areAndis a matrix of orthogonality of the (c),、、is thatAnd (3) withIs a diagonal matrix of (a);
at this time, alignment method is performed by subspace distributionKnowledge in the historical environment can be transferred to the current environment to quickly adapt to the current environment.
；
Wherein,represents the v th generationKnowledge of the individual;
and (3) cycling the steps until reaching a termination condition, and outputting a multi-star imaging task planning scheme under a complex heterogeneous environment.
The multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes can convert target tasks arriving in batches into a series of task sets carrying priority information, and reduces loss caused by unreasonable satellite resource allocation; the method solves the problems that model expansibility in task allocation is poor, algorithm pertinence is weak and the like in multi-star imaging task planning under different actual demands of different users in complex heterogeneous scenes, can effectively reduce repeated calculation of task income values, improves task planning efficiency, accelerates convergence speed, and outputs a more reasonable multi-star imaging task planning scheme under complex heterogeneous environments.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.
Claims (9)
1. A multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scenes is characterized by comprising the following steps:
step S10, constructing a task evaluation model according to the imaging requirements of the user and task planning scene information;
step S20, evaluating the imaging requirements of the user according to the task evaluation model, and generating a multi-star imaging task carrying execution priority information;
step S30, constructing a multi-star imaging task planning model under a complex heterogeneous scene according to the execution decision variables, the task benefit objective function, the response time objective function, the global optimization objective function, the imaging satellite self-performance constraint and the objective task allocation constraint of the multi-star imaging task;
step S40, determining the corresponding relation between the imaging satellites and the target tasks according to the number of the imaging satellites and the number of the target tasks to be observed currently;
step S50, calculating a task gain value of the imaging satellite for executing each target task according to a task gain objective function;
step S60, storing the task income value into a task income element knowledge storage library of the imaging satellite and the target task according to the corresponding relation between the imaging satellite and the target task;
step S70, constructing a multi-target evolutionary algorithm based on knowledge transfer according to satellites, target tasks, task profit values and response time, and outputting a multi-star imaging task planning scheme under a complex heterogeneous environment according to the multi-target evolutionary algorithm and the multi-star imaging task planning model; the user imaging requirements include: the method comprises the following steps of target task type, target task state, target task geometric position, image resolution, target task level and target task emergency degree; the method comprises the steps of carrying out a first treatment on the surface of the The task planning scene information includes: a multi-satellite imaging task planning period, an observation target task scale and an execution task imaging satellite scale; the step S10 includes:
the task assessment model w is constructed according to the following formula:
wherein X is i For the user imaging requirement, i is the type of the user imaging requirement, n is the type number of the user imaging requirement, T type Representing the type of target task, T state Representing the state of the target task, T pos Representing the geometric position of the target task, I res Representing the resolution of the image required by the user, T urg Representing the emergency degree of the target task;
the judgment matrix of the task evaluation model has the following formula:
wherein i epsilon [1, n ], j epsilon [1, n ];
the task weight coefficient calculation function of the task evaluation model has the following formula:
wherein X is i Representing an ith one of the task priority impact factors;
the Priority of the multi-star imaging task is calculated according to the following formula:
wherein B is ij For the relative importance between index i and index j,
2. the multi-star imaging mission planning method of claim 1, wherein step S30 includes:
the decision variables of the multi-star imaging task are as follows:
wherein S represents the number of imaging satellites, T represents the number of observation target tasks, and x ji A decision variable representing the execution of the observation target task i by satellite j;
the task benefit objective function maxf 1 (x ji ) The following formula is given:
wherein, benefit represents a task profit function, x ji Execution decision variables and priority representing multi-star imaging tasks task Representing task priority;
the response time objective function maxf 2 (x ji ) The following formula is given:
wherein,the response time of executing the observation target task i is represented, and T represents the number of the observation target tasks;
the global optimization objective function maxF (x ji ) The following formula is given:
maxF(x ji )＝{f 1 (x ji ),f 2 (x ji )}。
3. the multi-star imaging mission planning method of claim 1, wherein step S30 further comprises:
the imaging satellite self-performance constraint is as follows:
wherein, the orbit k Representing the track k, et k (x ji ) Representing the end time, st, of the imaging satellite j observing the object i in orbit k k (x ji ) Respectively the start time of the imaging satellite j observing the object i in orbit k,representing the data transfer rate of imaging satellite j, maxM a Maximum storage capacity representing imaging satellite payload, < >>The communication rate of the imaging satellite j is represented, con_et in the communication end time and the start time of the imaging satellite is represented, and Con_st respectively represents the communication on of the imaging satelliteStart time;
the target task allocation constraint is as follows:
wherein w is ijk Representing the observation time window for an imaging satellite j to observe a target i in orbit k,represents the yaw angle of the observation target task, A j Representing the maximum roll angle of the imaging satellite, +.>Remote sensor resolution, r, representing imaging satellite j min Representing minimum resolution, θ, required by the target task j Representing a single field angle of view of the satellite remote sensor.
4. The multi-star imaging mission planning method of claim 1, wherein step S40 includes:
when the number of the imaging satellites is consistent with the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is one-to-one correspondence;
when the number of the imaging satellites is smaller than the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is one-to-many;
when the number of the imaging satellites is larger than the number of the target tasks to be observed currently, the corresponding relation between the imaging satellites and the target tasks is many-to-one.
5. The multi-star imaging mission planning method of claim 1, wherein the mission benefit metadata repository comprises a first sub-repository and a second sub-repository; the row index of the task income value is a sub-library number; the step S60 includes:
the row index of the first sub-library is a target task number, and the column index is a satellite number;
the first sub-library is the sub-pool of the meta-knowledge 1 The mission benefit value for performing the target mission i for imaging satellite j is formulated as follows:
wherein, bene (x ji ) A mission benefit value representing the performance of the target mission i by the imaging satellite j;
the row index of the second sub-library is a target task number executed for the first time, and the column index is a target task number executed again;
the meta-knowledge sub pool of the second sub pool 2 After the imaging satellite j finishes executing the target task i, the task profit value of the target task u is executed by taking the target task i as a starting point, and the formula is as follows:
wherein, bene (x (i,u) ) And (3) after the imaging satellite finishes executing the target task i, taking the target task i as a starting point and then executing the task profit value of the target task u, wherein S represents the number of the imaging satellites, and T represents the number of the observed target tasks.
6. The multi-star imaging mission planning method of claim 1, wherein step S70 includes:
step S71, randomly selecting a first amount of knowledge from the task income element knowledge storage library as an initialization environment; the first number is less than the total knowledge number in the task revenue metadata repository; the satellite, the target mission, the mission benefits, the response time are uniformly described as knowledge according to the following formula:
knowledge v,g ∈MSIMP knowledge ，
knowledge v,g ＝(sat j ,(t i ,...t u ),(bene(x ji )∪Sbene(x ji ,x ju )),(Time(x ji )∪Time(x ji ,x ju ) A) wherein knowledges v,g Storing a library MSIMP for task revenue metadata knowledge G knowledge of the v generation of (a), in particular satellite sat j Execution target task (t) i ,...t u ) The task profit value is (ene (x ji )∪Sbene(x ji ,x ju ) The response Time is (Time (x) ji )∪Time(x ji ,x ju ))；
Step S72, calculating the crowding distance between the knowledge; the knowledge is satellite and/or target mission and/or mission gain and/or response time;
step S73, dividing the knowledge into useful knowledge and sub-useful knowledge according to the crowding distance between the knowledge, and storing the useful knowledge and the sub-useful knowledge in a useful knowledge niche and a sub-useful knowledge niche respectively, wherein the specific formulas are as follows:
step S74, calculating the environmental variability of the knowledge according to the global optimization objective function value corresponding to the knowledge in the environment;
and step S75, a knowledge transfer strategy is adaptively selected according to knowledge difference, and a multi-star imaging task planning scheme under a complex heterogeneous environment is output.
7. The multi-star imaging mission planning method of claim 6, wherein step S72 includes:
the crowding distance between the knowledge is calculated according to the following formula:
wherein minEuc (M i ,M j ) Represents the minimum Euclidean distance between knowledge, maxEuc (M i ,M j ) Representing knowledgeMaximum euclidean distance between them.
8. The multi-star imaging mission planning method of claim 6, wherein step S74 includes:
the knowledge difference between the environments is calculated according to the following formula:
wherein, the ethernet t Representing the environmental variability of knowledge of t, F (x i,t ) A global optimization objective function value corresponding to the ith location of the nth knowledge in the historical environment is represented,and the global optimization objective function value corresponding to the ith position of the ith knowledge in the current environment is represented.
9. The multi-star imaging mission planning method of claim 6, wherein step S75 includes:
if it is an ethernet t And (3) calculating the crowding distance between the knowledge again, and adaptively selecting to directly transfer the knowledge corresponding to the historical environment to the current environment according to the actual user requirements:
C knowledge (knowledge)＝knowledge rand ∪H knowledge (knowledge)
wherein the historical environmental knowledge is taken as source data H knowledge Current environmental knowledge C knowledge As target data, knowledges rand Representing randomly selected knowledge in the environment;
if it is an ethernet t >le-5, adopting a subspace distribution alignment method to mine association rules between the historical environment and the current environment, and migrating knowledge from the historical environment to the current environment;
the subspace distribution alignment method is as follows:
wherein H is s Is H knowledge Feature vector of C T Is C knowledge Feature vector of (2), O S Is H s Orthogonal complement of A 1 、A 2 Respectively areAnd->Lambda of the orthogonal matrix 1 、λ 2 、λ 3 Is H s And C T Is a diagonal matrix of (a);
C knowledge (knowledge v,g )＝H knowledge (knowledge v,g )·SDM S
wherein, knowledges v,g Representing knowledge of generation g of v, SDM S A subspace distribution alignment method;
and outputting a multi-star imaging task planning scheme under the complex heterogeneous environment until the termination condition is reached.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311567217.6A CN117271146B (en) | 2023-11-23 | 2023-11-23 | Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311567217.6A CN117271146B (en) | 2023-11-23 | 2023-11-23 | Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117271146A CN117271146A (en) | 2023-12-22 |
| CN117271146B true CN117271146B (en) | 2024-02-23 |
Family
ID=89208438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311567217.6A Active CN117271146B (en) | 2023-11-23 | 2023-11-23 | Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117271146B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114580209A (en) * | 2022-01-13 | 2022-06-03 | 中国人民解放军国防科技大学 | Multi-satellite cooperative task allocation method and system for non-time-sensitive moving target |
| CN116307535A (en) * | 2023-02-13 | 2023-06-23 | 中国人民解放军战略支援部队航天工程大学 | Multi-star collaborative imaging task planning method based on improved differential evolution algorithm |
| CN116562466A (en) * | 2023-07-07 | 2023-08-08 | 中国人民解放军战略支援部队航天工程大学 | Planning method for high orbit satellite observation task |
| CN116629511A (en) * | 2023-04-12 | 2023-08-22 | 中国人民解放军战略支援部队航天工程大学 | Multi-star dynamic task planning method and device based on two-stage hybrid scheduling in uncertain environment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IT201700056428A1 (en) * | 2017-05-24 | 2018-11-24 | Telespazio Spa | INNOVATIVE SATELLITE SCHEDULING METHOD BASED ON GENETIC ALGORITHMS AND SIMULATED ANNEALING AND RELATIVE MISSION PLANNER |
-
2023
- 2023-11-23 CN CN202311567217.6A patent/CN117271146B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114580209A (en) * | 2022-01-13 | 2022-06-03 | 中国人民解放军国防科技大学 | Multi-satellite cooperative task allocation method and system for non-time-sensitive moving target |
| CN116307535A (en) * | 2023-02-13 | 2023-06-23 | 中国人民解放军战略支援部队航天工程大学 | Multi-star collaborative imaging task planning method based on improved differential evolution algorithm |
| CN116629511A (en) * | 2023-04-12 | 2023-08-22 | 中国人民解放军战略支援部队航天工程大学 | Multi-star dynamic task planning method and device based on two-stage hybrid scheduling in uncertain environment |
| CN116562466A (en) * | 2023-07-07 | 2023-08-08 | 中国人民解放军战略支援部队航天工程大学 | Planning method for high orbit satellite observation task |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117271146A (en) | 2023-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109116349B (en) | Multi-sensor cooperative tracking joint optimization decision method | |
| Liu et al. | Energy-efficient UAV crowdsensing with multiple charging stations by deep learning | |
| CN107886201B (en) | Multi-objective optimization method and device for multi-unmanned aerial vehicle task allocation | |
| CN107977743B (en) | Multi-unmanned aerial vehicle cooperative task allocation method and device | |
| CN106656308B (en) | The planning method and device of task in Information Network | |
| CN106384169B (en) | A kind of satellite task planing method based on meta-heuristic algorithms | |
| CN116307535B (en) | Multi-star collaborative imaging task planning method based on improved differential evolution algorithm | |
| Liu et al. | Multi-task-oriented vehicular crowdsensing: A deep learning approach | |
| CN101894367A (en) | Target clustering method of imaging satellite observation scheduling | |
| CN109919484B (en) | On-satellite autonomous task planning method | |
| CN111612384B (en) | Multi-star relay task planning method with time resolution constraint | |
| CN111313957B (en) | Hybrid satellite communication system resource allocation method based on classification multi-objective optimization | |
| CN111897640A (en) | Satellite imaging task planning method and system for regional mapping | |
| CN111178419B (en) | Agile remote sensing satellite multi-target task planning method based on task clustering | |
| Li et al. | An improved ant colony optimization for path planning with multiple UAVs | |
| CN112734239A (en) | Task planning method, device and medium based on task and resource capacity attributes | |
| CN110874413B (en) | Association rule mining-based method for establishing efficacy evaluation index system of air defense multi-weapon system | |
| Petković et al. | Machine learning for predicting thermal power consumption of the mars express spacecraft | |
| CN114035913A (en) | Satellite multi-target repeated observation task planning method and terminal | |
| Fu et al. | Mobile robot object recognition in the internet of things based on fog computing | |
| CN117271146B (en) | Multi-star imaging task planning method based on knowledge transfer under complex heterogeneous scene | |
| CN111399533B (en) | Heterogeneous multi-unmanned aerial vehicle cooperative task allocation and path optimization method | |
| Yang et al. | Learning graph-enhanced commander-executor for multi-agent navigation | |
| CN109859063B (en) | Community discovery method and device, storage medium and terminal equipment | |
| CN114580208A (en) | On-satellite autonomous task generation method and system for ocean moving target |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |