CN109409773B - Dynamic planning method for earth observation resources based on contract network mechanism - Google Patents
Dynamic planning method for earth observation resources based on contract network mechanism Download PDFInfo
- Publication number
- CN109409773B CN109409773B CN201811351029.9A CN201811351029A CN109409773B CN 109409773 B CN109409773 B CN 109409773B CN 201811351029 A CN201811351029 A CN 201811351029A CN 109409773 B CN109409773 B CN 109409773B
- Authority
- CN
- China
- Prior art keywords
- planning
- task
- resources
- tasks
- resource
- 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
- 238000000034 method Methods 0.000 title claims abstract description 177
- 238000013439 planning Methods 0.000 title claims abstract description 168
- 230000007246 mechanism Effects 0.000 title claims abstract description 27
- 230000008569 process Effects 0.000 claims abstract description 45
- 238000009826 distribution Methods 0.000 claims abstract description 21
- 238000004891 communication Methods 0.000 claims description 14
- 238000007667 floating Methods 0.000 claims description 12
- 238000010845 search algorithm Methods 0.000 claims description 8
- 238000007726 management method Methods 0.000 claims description 4
- 230000008901 benefit Effects 0.000 abstract description 13
- 230000001360 synchronised effect Effects 0.000 abstract description 4
- 230000011218 segmentation Effects 0.000 abstract description 2
- 238000004422 calculation algorithm Methods 0.000 description 39
- 230000008859 change Effects 0.000 description 16
- 238000005457 optimization Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 6
- 101100005071 Drosophila melanogaster CanB gene Proteins 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000005295 random walk Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000008672 reprogramming Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012733 comparative method Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 239000003999 initiator Substances 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
- 238000003892 spreading Methods 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
Classifications
-
- 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
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/08—Auctions
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Theoretical Computer Science (AREA)
- Educational Administration (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Computer Security & Cryptography (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the field of satellite remote sensing, and discloses a dynamic planning method for earth observation resources based on a contract network mechanism, which adopts a bottom-up distributed collaborative planning architecture and a collaborative planning process based on a contract network; and dynamically distributing the large-scale concurrent tasks by a multi-round incomplete combination distribution method. The invention provides a bottom-up distributed contract network collaborative planning framework and a planning flow from bottom to top by starting from a bottom layer framework on the basis of analyzing the existing planning system and resource operation mode, breaking through the thinking fixed form of a top-down inherent planning mode and combining the distributed computing advantages of the contract network to the problem of dynamic planning of space-sky-ground heterogeneous resources, so as to give full play to the computing advantages of the distributed resources and further improve the task allocation efficiency. On the basis, a multi-class incomplete combination distribution method facing large-scale tasks is provided by adopting three strategies of combined task segmentation, multi-task set synchronous distribution and multi-level matching, and a large number of concurrent tasks can be quickly distributed.
Description
Technical Field
The invention relates to a dynamic planning method for earth observation resources based on a contract network mechanism.
Background
The air-space-ground integrated earth observation including observation resources such as satellites, unmanned aerial vehicles and ground sensors plays a key role in many aspects such as environmental monitoring, disaster assessment, urban analysis and national security. Different application fields have different requirements on observation resources in terms of spatial resolution, time window, spectral band, and the like. With the improvement of industrial production and observation means, earth observation is more and more refined, so as to meet different earth observation requirements in various fields. However, due to the limitation of the inherent observation mode and observation capability of a single resource, a single resource is difficult to meet the requirements of numerous heterogeneous observations, so that the collaborative planning of heterogeneous resources is a necessary development trend. In addition, in the process of executing tasks, the external environment has high dynamics and uncertainty, the tasks may change at any time, and the observation resources are at any time in danger of being damaged and broken with a communication link. Therefore, the original observation scheme needs to be dynamically adjusted at any time.
The collaborative planning research on the earth observation resources mostly focuses on the scheduling of single-class resources, and the collaborative planning method for the multi-class heterogeneous earth observation resources is still in a starting stage. Planning and scheduling technologies for single resources such as satellites, unmanned planes and ground equipment tend to mature gradually, and various heuristic models and intelligent optimization algorithms are widely applied. Due to inherent limitations of single-class resource observation modes and capabilities, observation benefits of isolated planning reach a bottleneck, and a general planning method for multiple classes of observation resources gradually receives attention in the face of accurate and diversified observation requirements.
In the aspect of collaborative planning of multiple types of earth observation resources, the existing method generally proposes a top-down hierarchical heterogeneous resource collaborative planning architecture, and integrates different observation resources to form a loosely-coupled earth observation system. The architecture comprises 4 different levels: the system comprises an input layer, a coordination layer, a task planning layer and an observation resource layer. The input layer is responsible for receiving and managing tasks submitted by users, and the cooperation layer has a global view of a planning system and comprises a planning center and manages all secondary sub-planning centers in a coordinated manner. The task planning layer comprises a plurality of sub-planning centers, and each sub-planning center is responsible for managing single-class observation resources under the jurisdiction of the sub-planning center. The observation resource layer includes numerous heterogeneous resources and is responsible for performing tasks. On the basis, the scheme takes the collaborative planning process of various resources as a task allocation process, 3 heuristic criteria of observation opportunities, conflict degrees and resource consumption degrees are constructed by analyzing conditions such as observation income, observation opportunities, resource capacities and the like of different resources, and then a heuristic task allocation model facing central planning is provided, so that tasks are distributed to a plurality of secondary sub-planning centers from a planning center, and the sub-planning centers further assign observation tasks by combining the existing single-class resource planning mode.
The conventional multi-class earth observation resource collaborative planning method is suitable for resource planning in a static environment. However, in the actual task execution process, the external environment and the subjective intention of the user have high dynamics and uncertainty, which requires that the collaborative planning method has the capability of fast planning and re-planning of large-batch tasks.
The defects of the prior art mainly comprise: 1) the adopted top-down resource planning framework comprises a planning center, tasks are uniformly distributed downwards by the center, the mode similar to centralized planning can cause the system to have weaker robustness and low distribution efficiency, and the planning center is required to have strong computing capacity and good and stable communication environment; 2) the existing collaborative planning technology ignores the high dynamic property and uncertainty of a task execution environment, is difficult to dynamically and locally adjust the original planning scheme, and is difficult to rapidly re-plan a task distribution result; 3) the existing technical framework does not fully exert the independent distributed computing capability of the observation resources, and is difficult to plan large-scale burst tasks.
The earth observation tasks in various fields have diversity, complexity and dynamics, and the concurrency characteristics of typical large-scale tasks, which puts higher requirements on the capabilities of accuracy, persistence, timeliness, rapid strain and the like of an air-space-earth cooperative observation system. In the existing observation system, most of various observation platforms operate independently, and the isolated task planning working mode can not meet the observation requirements under the condition of multi-task concurrence, so that the bottleneck for restricting the rapid processing of emergency disaster tasks is formed. In addition, research on collaborative planning of a small number of heterogeneous observation resources is still limited to a static external environment, and dynamic actual mission planning requirements are difficult to fit. Therefore, the space-air-ground integrated earth observation resources are effectively organized to form an efficient and dynamic cooperative observation network so as to support various concurrent disaster emergency tasks, meet the disaster emergency requirements in an all-round manner, and become a major challenge for space-air-ground integrated earth observation application. The nature of the heterogeneous resource dynamic planning problem is based on the dynamic and fast matching of tasks and resources of the original observation scheme.
The heterogeneous resource dynamic collaborative planning process mainly has the following difficulties and requirements: firstly, the space-air-ground observation resources are not only distributed discretely in space, but also have great differences in use constraints, maneuverability, load performance, observation capability, maneuvering forms and the like. Each planning center is provided with a relatively independent planning system, and it is difficult to construct a distributed system framework to integrally schedule all observation resources. Secondly, the observation environment has high uncertainty, observation resources must be dynamically accessed to the system in the task planning process, and randomly occurring observation tasks can be dynamically and efficiently processed under the condition that the resource capacity is constantly changed. Finally, the emergency observation tasks have random concurrency, and the planning process must dynamically allocate large-scale concurrent tasks on the basis of maximizing the total observation income.
At present, in the field of earth observation, each planning center has relatively independent observation resources, and the planning system is relatively independent. As shown in fig. 1, as the uploading and data receiving centers of the instructions, communication can be realized between the planning centers, while some internal resources belonging to the same planning center have communication conditions, and communication is difficult for resources not belonging to the same planning center. When an emergency disaster event occurs, the monitoring task is usually submitted to a nearby planning center by a user according to the distribution position of the observed resource, and the planning center distributes tasks to the resource nodes in a centralized manner according to a set heuristic criterion. The traditional top-down type distribution framework usually adopts the hierarchical structure, and tasks are successively distributed to the local resources of the secondary hierarchy by a high-level global planning center. However, it is difficult for a dynamically uncertain planning environment to ensure its required high computing power and a good stable communication environment.
Disclosure of Invention
The invention aims to provide a dynamic planning method for earth observation resources based on a contract network mechanism, which mainly solves the problems of distributed integration of heterogeneous earth observation resources and rapid planning under a dynamic uncertain environment.
In order to achieve the aim, the invention provides a dynamic planning method for earth observation resources based on a contract network mechanism, which adopts a bottom-up distributed collaborative planning architecture and a collaborative planning process based on a contract network; and dynamically distributing the large-scale concurrent tasks by a multi-round incomplete combination distribution method.
Further, the bottom-up distributed collaborative planning architecture includes the following four levels:
the task management layer is responsible for receiving tasks submitted by users, dividing the large-scale tasks into different task sets according to the resource observation capacity and further distributing the task sets to corresponding resources of the next level;
the resource coordination layer comprises a plurality of heterogeneous resources, the observation resources receive the task set to be completed and are distributed to a neighborhood resource set capable of communicating with the observation resources through negotiation with the task to be re-planned; on the basis, the resource node submits the unallocated tasks to the next level;
the planning center internal negotiation layer assigns the unallocated tasks to the internal resources governed by the planning center on the basis of internal negotiation according to the resource observation capacity;
and the planning center cooperates with the layer, and the planning center continuously submits the remaining uncompleted tasks to other planning centers for bidding.
Further, the collaborative planning process based on the contract network is used for task allocation in an active bidding mode of earth observation resources.
Further, the active bidding mode includes the following five phases:
phase 1: resource riReceiving task sequence T ═ T (T) to be inserted1,t2,…tn) When j belongs to the n instruction, actively judging whether T can be completed according to the cache task to be completed and T;
phase 2: if r isiCan not complete task T, then riChanges the role of the target to a sender and sends the target to a resource group RS capable of communicating with the targeti-LReleasing bidding information; according to RSi-LJudging whether the task T can be completed or not by the returned scalar value and the completion information;
phase 3: if all resources cannot complete T, then riFeeding back task information to a sub-planning center P at the upper levelk；PkServing as a resource to be issued to the inside of the planning center by the issuer, riThe role of (2) is changed from a publisher to a bidder; pkJudging whether resources can complete T or not according to the returned scalar values and the completed task subsets;
phase 4: if no resources complete T, PkAgain as a publisher to set PS to a communicable neighborhood planning centerk-LReleasing the bidding document; each planning center [ P1,P2,…Pk-1,Pk+1…Ppn]Respectively appointing task subsets which can be completed according to internal resources and feeding back the task subsets; pkEvaluating a final task allocation scheme according to the feedback result;
phase 5: if P iskAccording to the neighborhood set PS when the neighborhood planning center can not complete Tk-LContinues to expand the scope of the bid until the task T can be completed or there are no remaining planning centers.
Further, the multi-round incomplete combination allocation method allocates task sets in three levels from bottom to top in sequence respectively:
at the first level, according to a contract network method, each resource is used as a bidding direction to release a bidding document to a communication neighborhood resource set of the resource, and a combined task allocation scheme is determined through resource negotiation;
at the second level, each planning center receives uncompleted tasks and appoints resources inside the planning center to complete the tasks;
at the third level, all planning centers negotiate together to determine the allocation scheme for the incomplete tasks.
Further, a local search algorithm based on a floating bid selection mechanism is adopted to solve the problem of determining the winning bid of the contract network in the heterogeneous resource collaborative planning process.
Through the technical scheme, the following beneficial technical effects can be realized:
the invention provides a bottom-up distributed cooperative planning architecture and a contract network driven large-scale dynamic task allocation method based on cooperative planning of heterogeneous earth observation resources, based on deep analysis of task matching problems in the existing air-space-earth planning system and dynamic environment, combined with the inherent advantages of a contract network mechanism in the field of distributed planning, aiming at two core problems of distributed cooperation and dynamic high-efficiency planning of heterogeneous resources, the method realizes loose coupling between heterogeneous observation resources facing dynamic task allocation, furthest exerts potential computing power of all distributed observation resources, improves observation benefits of air-space-ground integrated earth observation resources, breaks through technical bottleneck of large-scale task rapid planning under dynamic uncertain environment, finally, accurately, rapidly, reliably and flexibly directly serves disaster emergency departments to guide disaster rescue decisions.
The invention breaks through the thinking fixed mode of a top-down inherent planning mode on the basis of analyzing the existing planning system and resource operation mode, combines the distributed computing advantages of the contract network to face the problem of dynamic planning of space-sky-ground heterogeneous resources, innovatively provides a bottom-up distributed contract network collaborative planning framework and provides a planning flow, so that the computing advantages of the distributed resources are fully exerted, and further the task allocation efficiency is improved. On the basis, a multi-class incomplete combination distribution method facing large-scale tasks is provided by adopting three strategies of combined task segmentation, multi-task set synchronous distribution and multi-level matching, and a large number of concurrent tasks can be quickly distributed.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the embodiments of the invention without limiting the embodiments of the invention. In the drawings:
FIG. 1 is a schematic diagram of an aerospace-ground resource-to-ground observation system according to the present invention;
FIG. 2 is a schematic diagram of a bottom-up distributed contract network collaborative planning architecture of the present invention;
FIG. 3 is a bottom-up collaborative framework task planning flow diagram of the present invention;
FIG. 4 is a schematic diagram of the contract net bidding process of the earth observation system of the present invention;
FIG. 5 is a schematic diagram of the task allocation process of the present invention, in which (a) large-scale tasks are divided according to the Voronoi diagram of the resources, (b) partial tasks are allocated to the communicable neighborhood resources based on the contract network, and (c) the remaining tasks are sequentially allocated to the internal resources of the planning center and the communicable planning center;
FIG. 6 is a graph illustrating the comparison of task completion rate results for different methods of the present invention;
FIG. 7 is a graph showing the comparison of solution time results according to various methods of the present invention;
FIG. 8 is an exploded schematic diagram of solution time of the method of the present invention in three levels, where (a) is resource collaborative layer algorithm runtime, (b) is planning center internal collaborative layer algorithm runtime, and (c) is planning center collaborative layer algorithm runtime;
FIG. 9 is a schematic diagram of the comparison of planning scheme quality for different methods of the present invention;
FIG. 10 is a graph illustrating a comparison of task completion results for dynamic re-planning by different methods of the present invention;
FIG. 11 is a comparison of solution time results for different methods of dynamic re-planning of the present invention;
FIG. 12 is a schematic diagram showing the comparison of the rate of change of the solution dynamically re-planned by the different methods of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating embodiments of the invention, are given by way of illustration and explanation only, not limitation.
The dynamic planning method for the earth observation resources based on the contract network mechanism is specifically set forth as follows:
1) bottom-up distributed contract network collaborative planning architecture and process
A. Bottom-up distributed collaborative planning architecture
The collaborative planning framework for the space-sky-ground heterogeneous observation resources must satisfy three requirements. First, it must conform to the existing management structure and communication mechanism of various observation resources. Second, the framework must have good scalability to ensure dynamic access and release of resources and to enable real-time insertion and demand modification of large-scale observation tasks. Third, in order to achieve the goal of dynamic planning, the framework must have a high planning efficiency. The invention provides a bottom-up distributed contract network cooperative framework by fully considering the principle. Compared with a top-down framework, the framework emphasizes the dynamic response of the individual resources to uncertain environments, focuses more on the self-organization coordination strategy of the individual resources under the task stress reaction, and is an allocation strategy for promoting global optimization from individual local interaction response.
As shown in fig. 2, the architecture is divided into four levels from bottom to top. And the task management layer is responsible for receiving tasks submitted by users and dividing the large-scale tasks into different task sets according to the resource observation capacity. And then the task set is distributed to the corresponding resource of the next layer. The resource coordination layer comprises a plurality of heterogeneous resources, and the observation resources receive a task set to be completed and are distributed to a neighborhood resource set capable of communicating with the observation resources together with a task needing to be re-planned through negotiation. On this basis, the resource node submits the unassigned tasks to the next level. And the planning center internal cooperation layer assigns the unallocated tasks to the internal resources governed by the planning center on the basis of internal negotiation according to the resource observation capability. And the uncompleted tasks are continuously submitted to the next planning center collaborative layer. Finally, the planning center distributes the legacy tasks to the communicable neighborhood planning centers according to the observation capability of the adjacent planning centers. This framework does not change the operating mode of existing resources. In order to achieve the goal of rapid planning, tasks are allocated from the resources to the planning center from bottom to top, the tasks are preferentially allocated to the neighborhood observation resources, and then the tasks are gradually diffused to other planning centers through the bottom to top multi-turn task allocation process until the tasks can be completed. In addition, in order to realize the dynamic and extensible performance of the collaborative planning system, the framework integrates a Contract network Protocol mechanism with the task planning process of space-ground observation resources, and a CNP (Contract network Protocol) can provide a planning scheme for the task allocation process of each round.
B. Collaborative planning process based on contract network
As a high-level communication interaction protocol, the CNP supports cooperation and competition of large-scale resources in a dynamic scenario. The CNP borrows from the "bid" mechanism in economic behavior that can change the centralized allocation of tasks to active bidding of resources. The distributed computing power of each observation resource can be utilized to greatly improve the task allocation efficiency, and meanwhile, the active bidding mechanism is beneficial to the dynamic expansion of the resources of the task planning system.
The collaborative planning process can be regarded as a process of spreading tasks among resources. In a bottom-up distributed contract network cooperative architecture, a planning center serves as a high-level manager to govern a plurality of low-level resources. Each resource can be viewed as a bidder with varying capabilities, with their role transitioning between bidders and publishers as the task allocation process progresses. Only planning centers or resources having communication paths can communicate with each other. The process is shown in fig. 3, when a certain observation resource monitors that there is a task that needs to be dynamically inserted, the establishment of the task planning scheme is gradually completed in 5 stages based on the contract network:
phase 1: resource riReceiving task sequence T ═ T (T) to be inserted1,t2,…tn) When j belongs to the n instruction, actively judging whether T can be completed according to the cache task to be completed and T;
phase 2: if r isiCan not complete task T, then riChanges the role of the target to a sender and sends the target to a resource group RS capable of communicating with the targeti-LPost bid information (fig. 3 a); according to RSi-LJudging whether the task T can be completed or not by the returned scalar value and the completion information;
phase 3: if all resources cannot complete T, then riFeeds back task information toFirst-level sub-planning center Pk。PkServing as a resource to be issued to the inside of the planning center by the issuer, riThe role of (2) is changed from the publisher to the bidder (fig. 3 b); pkJudging whether resources can complete T or not according to the returned scalar values and the completed task subsets;
phase 4: if no resources complete T, PkAgain as a publisher to set PS to a communicable neighborhood planning centerk-LReleasing the bidding document; each planning center [ P1,P2,…Pk-1,Pk+1…Ppn]Respectively appointing task subsets which can be completed according to internal resources and feeding back the task subsets; pkEvaluating a final task allocation scheme according to the feedback result;
phase 5: if P iskAccording to the neighborhood set PS when the neighborhood planning center can not complete Tk-LContinues to expand the scope of the bid until the task T can be completed or there are no remaining planning centers.
And determining a resource allocation scheme by adopting a contract network mechanism at each stage in the planning process. The air-space-ground observation contract network mechanism has three types of participation roles: a bidder, and a winning bidder. As shown in fig. 4, the operation of the contract network can be divided into the following four steps: bidding document issuing, resource bidding, winning bid optimization, contract signing and executing.
In the bidding document issuing process, the task initiator issues bidding announcement as a potential neighborhood resource which can be communicated with the bidding direction. In the air-space-ground observation contract network, the triggering of task release mainly comes from two aspects: the resources have no capability to continuously execute the tasks in the reserved task buffer pool due to external interference or insufficient capability of the resources and the like; the current resource itself monitors new tasks that need to be monitored but that itself are not capable of being completed. The format of the message for the invitation announcement is as follows: the contract identifier is a unique identifier, wherein the contract identifier is a unique identifier, and the contract identifier is a unique identifier. The TaskInfo is a detailed description of the observation task, including the spatial location of the task, the execution time, the task name, etc. The taskRequirements are additional requirements and constraints of the task, including spatial resolution, spectral band, etc. The tasskgrade indicates the level of importance of the task. Expiretime is the bid cutoff time. QuoteRequirement is a range of bid values.
In the bidding stage, the air-space-ground resource as a candidate bidder can evaluate the execution income, cost and influence on the existing observation plan of the task according to contract requirements and constraint conditions after receiving the bidding announcement. And then determining a bidding task and a bid price and feeding back the tender to the tenderer within a specified time. The resource must satisfy the constraint conditions such as the current position of the resource, the task sequence to be executed, the residual cruising ability and the like in the bidding process. Therefore, the bidding process for resources can be viewed as a Constraint Satisfaction Problem (CSP). The bid file format is: BidDoucument [ < ContractID, Bid, executive Scheme, BidPrice, TaskSequences, Indicators status >. Where Bid is a boolean variable indicating whether to Bid. The executionScheme represents the execution scheme of the bidder for completing the task, and comprises information such as completion time, resolution, band and the like. BidPrice represents the bid value of the feedback. The TaskSequences are the set of all tasks and their execution sequences that the current bidder will execute. The indica torssstatus is a status description of whether all task indicators and constraints can be completed.
In the bid selection stage, after successfully receiving a plurality of bidding documents, the tenderer selects an optimal execution scheme from all bidding documents through a contract optimization algorithm and awards the contract to the selected bidder. The selection of the optimal solution can be seen as a Winner Determination Problem (WDP).
In the task execution phase, the tenderer completes contract signing with the successful bidder and notifies all bidders of successful signing information. And the successfully signed resource inserts the signed task into the task buffer pool, and dynamically adjusts the execution sequence of the tasks in the buffer pool according to the task execution strategy. Other bidders not receiving the offer continue to perform their existing sequence of tasks.
2) Multi-round incomplete combination distribution method for large-scale tasks
In the process of executing tasks by the air-space-ground resources, once an emergency event occurs, a large number of observation tasks are triggered at the same time. The efficiency of the distribution of these tasks is critical to the success or failure of task execution. The single-task continuous allocation scheme improved based on the traditional contract network model is difficult to meet the requirement of high timeliness in dynamic planning. Therefore, the invention provides a multi-round incomplete combination task allocation Method (MICA) on the basis of a bottom-up framework, and the method can rapidly allocate a large number of tasks in batches through multi-round matching and support multi-task, multi-winner and multi-return bidding. Wherein, the determination of the bidding plan in each round is the core of the quick task allocation. For this reason, the present invention regards it as a winner determination problem and proposes a local search algorithm (FLS) based on floating targeting to solve the target selection problem for each round. The FLS algorithm combines the characteristics of task allocation and resource bidding, and improves the optimal solution convergence speed by using taboo and priority strategies. In addition, probability parameters and a floating label selecting mechanism are adopted to avoid the algorithm from falling into local optimization.
A. Multi-wheel incomplete combination distribution method
According to a bottom-up framework, the present invention proposes a multi-round incomplete combinatorial task assignment (MICA) approach to quickly match large-scale concurrent tasks. The MICA method employs three strategies to improve the efficiency of task allocation: task combination distribution, multi-task set synchronous distribution and multi-level resource matching. As shown in fig. 5, in order to solve the problems of difficulty in solving and low efficiency caused by large-scale tasks, the method first constructs a resource Voronoi diagram according to the current position of the resource and the observation capability thereof. The tasks are then divided according to this Voronoi diagram into a plurality of small task sets (fig. 5a) and synchronously allocated to the different resources. Since the resources have distributed computing power, tasks in multiple task sets may be allocated in parallel. Secondly, according to the collaborative planning framework, the MICA method allocates task sets in three levels from bottom to top in sequence. At the first level, a bidding document is issued by each resource as a bidding party to its communication neighborhood resource set according to the contract network method (fig. 5 b). Determining a combined task allocation scheme through resource negotiation. At the second level, each planning center receives the unfinished tasks separately and designates the planning center internal resources to finish the tasks (fig. 5 c). At the third level, all planning centers negotiate together to determine the allocation scheme of the incomplete tasks (fig. 5 c). The detailed algorithm is described below and in tables 1-2.
A Tenderer is set to tender a Tenderrer and a bid resource set Bidder; planning center set P ═ P1,P2,…,Ppn],k∈pn；PkManaging resource sets Rk＝(r1,r2,…rm),i∈m；PSk-LIs PkA communicable neighborhood planning center set; resource riReceiving a task set T ═ T (T) to be planned1,t2,…tn),j∈n；RSi-LIs a resource riA communicable neighborhood resource set; b ═<VR-T,GT>Set of bids, V, representing the return of Bidder to TendererR-TRepresenting a Bidder pair task set GTOf the bid offer set, GTA set of bid tasks is represented that is,
presentation scheme
The task set T contained in (1).
At the first level, if resource riCommunicable neighborhood resource set RSi-LNot equal to phi, then the resource riPrioritizing RS for task set Ti-LIssue a bid-for announcement, i.e. Tenderer ri,Bidder＝RSi-L. Each resource in the Bidder receives the bid notification, and a task set G capable of being completed is determined according to the task to be completed and the resource load condition of the resourceTAnd a quote value VR-TAnd returns the bid set B ═<VR-T,GT>. Then, resource riSelecting an optimal bidding scheme set according to a bid selection Algorithm (Algorithm 3)
Due to the limitation of resource capacity, the optimal solution often cannot complete all tasks of the initial set of tasks, i.e. the optimal solution is not able to complete all tasks of the initial set of tasks
At this point, the second level is entered, currently planning center PkAll the resources R which can be managed and controlled are appointedkCompleting remaining set of tasks
If all tasks in the task set T can not be completed, expanding the bidding scope to a neighborhood planning center set PSk-L. This is the third level. After each planning center arranges a task observation scheme, the bidding result is fed back according to the algorithm 2. If the tasks are still not completed, the communicable neighborhood planning center is used as a terminator to continue expanding the bidding scope until the tasks are completed or all resources are utilized. Finally according to the optimal scheme CbestAnd completing the signing of the respective resources and tasks.
TABLE 1 MICA Algorithm framework
TABLE 2 CNP-based planning center task allocation algorithm framework
B. Local search algorithm based on floating bid-selecting mechanism
The air-space-ground resource winner decision problem is a combination optimization problem. During each round of the negotiation assignment process of the contract network, all bidders will return a set of bids to the tenderer. Setting task set T to be allocated as (T)1,t2,…tn) J is an element n; bid term B for all resources (B ═ B1,B2,…Bi…Bm) Forming a candidate set CanB; v ═ V (V)1,V2,…Vm),i∈m,ViRepresenting bid term BiThe set of bid quotations of; g ═ G (G)1,G2,…Gm),i∈m,GiRepresenting bid term BiThe set of bidding tasks. If a bid term BiIf it is selected as the winning winner, BiBecomes one of the composition items of the solution C. The Winner Decision Problem (WDP) is to select a subset from the candidate set CanB as the feasible solution C to maximize the sum of the bid quotations in the feasible solution. The solution may be represented by a Boolean set x, xi1 denotes the bid term BiAnd (6) selecting. Let S be a binary matrix of m x n if task tj∈GiThen S isij1, otherwise S ij0. The objective function can thus be expressed as:
the constraint conditions are as follows:
xi∈{0,1} (2)
constraint (2) indicates that the bid term is in both the selected and unselected states; the constraint (3) indicates that each task can be selected only once, namely the selected tasks do not conflict with each other.
Defining a conflicting bid: for two bidders bdiAnd bdkBid task set G ofiAnd Gk，
If at least one task exists in both sets, i.e. { Gi∩GkNot equal to phi }, then called BiAnd BkMutually conflicting bids, GiAnd GkMutually conflicting bidding tasks. Otherwise, the two bids are said to be consistent. According to the two-by-two conflict relationship, a conflict bid matrix M can be constructedcon(symmetric array).
Aiming at the WDP problem in the air-space-ground resource contract network mechanism, the invention provides a local search algorithm (FLS) based on a floating bid selection mechanism, and the algorithm frame is shown in a table 3. The algorithm utilizes the probability parameters and the floating label selecting machine to control random walk, thereby enhancing the diversity of the solution and greatly improving the accuracy of the solution. Repeated retrieval of the candidate solution set space is prevented by using a tabu and priority strategy, and the optimal solution convergence speed is improved.
The FLS algorithm is a process of multiple iterations and stepwise optimization. The number of iterations y may be determined manually or until an optimal solution is found. In the searching process, if the algorithm searches all candidate bidding solution sets in each iteration, the convergence rate is necessarily slowed down. In practice, the prior search weights of the candidate solution sets are different. Therefore, in order to accelerate the search speed, the algorithm designs a preferential search bid set QBAnd search-prohibited bid set HB。QBIs a bid set compatible with the current optimal solution set, HBIs the set of bids that conflicts with the candidate optimal solution set C. To improve the solution efficiency, at the beginning of each iteration, the algorithm searches the key bid set Q preferentiallyBExclusion of contraindicated bid set HB. At the end of each iteration, the algorithm bases on the collision bid matrix MconUpdating Q in sequenceBAnd HB。
The continuous optimization process tends to trap the algorithm into local optimization. In order to avoid the problem and reduce the complexity of solving, the algorithm designs a probability parameter rho and a profit floating interval sigma. The algorithm performs a greedy search with probability p and a random walk with probability 1-p. According to the optimal solution C and the taboo list H as shown in the formula (4)BA candidate solution set CanB may be obtained. Algorithm or probability rho for determining maximum bid and bid value V from candidate task set CanBmaxBid B ofcanAdding the optimal solution C or randomly selecting a scalar value B from the B-C by the probability 1-rhocanAdding the optimal solution C. However, only the bid with the largest bid is selected too greedy in the algorithm. This is still not sufficient to enhance the diversity of the solution. Therefore, as shown in equation (5), theThe method selects the maximum report value V from the candidate set CanBmaxBidding set F with phase difference within floating interval sigmaB. And from the floating set FBIn randomly selecting a bid BcanAdding the optimal solution C. In this way, the algorithm can jump out of the locally optimal solution as much as possible. Finally, the algorithm outputs a global optimal solution C by updating the optimal solution C for each iterationbest。
CanB＝B-C-HB (4)
FB＝{Bi||Vmax-Vi|≤σ},Bi∈CanB (5)
Table 3 local search algorithm framework based on floating bid selection mechanism
Through deep analysis of the air-space-ground resource planning problem in the dynamic environment, a traditional top-down planning framework is abandoned, a bottom-up distributed collaborative planning framework is innovatively provided, and a whole set of solution for the air-space-ground resource dynamic planning problem from a bottom framework to an upper algorithm is designed. The bottom-up architecture is more consistent with dynamic and uncertain characteristics in the planning process. In addition, the invention deeply fuses the task allocation process and the contract network agreement, and fully exerts the distributed computing power of all the observation resource nodes.
Aiming at the problem of rapid re-planning of large-scale concurrent tasks, the invention provides a dynamic environment-oriented multi-round incomplete combination task allocation method. Three strategies are adopted to improve the task allocation efficiency. Firstly, a large-scale task is divided into a plurality of task sets so as to achieve the purpose of synchronous distribution of the plurality of task sets. Second, the combined allocation of multiple tasks in a set is properly bundled according to resource capabilities, thereby being more efficient than continuous single task allocation. And finally, according to a bottom-up collaboration framework, tasks are allocated to a system from resources to a planning center layer by layer, which is also a means for decomposing large-scale tasks to improve allocation efficiency.
The invention provides a local search algorithm to solve the problem of determining contract network winning marks in a heterogeneous resource collaborative planning process. The algorithm utilizes the probability parameters and the floating label selecting machine to control random walk, thereby enhancing the diversity of the solution and greatly improving the accuracy of the solution. Repeated retrieval of the candidate solution set space is prevented by using a tabu and priority strategy, and the optimal solution convergence speed is improved.
In order to test the collaborative planning performance of the bottom-up distributed contract network collaborative framework and the MICA method in large-scale task emergency and dynamic environment, the invention is provided with two groups of comparison tests. After an emergency event such as an earthquake, a landslide and the like occurs, in order to acquire the situation of a disaster area, an observation scheme of earth observation resources needs to be rapidly formulated according to an emergency task. According to the characteristics of emergency events, the continuous concurrency of large-scale tasks and the dynamic planning of uncertain environments are two main problems faced by the rapid planning scheme. Therefore, the first set of experiments of the present invention was used to verify whether the MICA method could effectively and quickly formulate an observation scheme for resources in the case of concurrent large-scale emergency tasks. The second set of experiments was used to verify whether the MICA method can quickly make reasonable observations about the original planning scenario in a dynamic uncertain environment.
1) Comparison and verification of different planning methods under large-scale task concurrency condition
In the comparison experiment, a simulation scene containing three different observation resources, namely a satellite, an unmanned aerial vehicle and an airship, is set. The three types of observation resources are respectively managed by four planning centers in a coordinated manner: 1 satellite planning center manages 2 earth observation satellites; the 2 unmanned plane planning centers respectively manage 25 unmanned planes and 28 unmanned planes; 1 airship planning center manages 9 airships. To verify that the MICA method of the present invention can cope with the pressure of large-scale tasks, we set 600 randomly distributed tasks in space (to eliminate experimental occasional errors, we randomly generated 5 sets of the same amount of data in the same space). The invention verifies the scheme making effect of the collaborative planning method under all the conditions that the task number is from 1 to 600, and compares the scheme making effect with four common collaborative planning methods. The first method is a contract network-based single continuous auction method (SSA), in which a virtual planning center serves as a publisher to manage all resources that feed back bid bids and observation plans as bidders. The virtual planning center auctions the single tasks in turn and selects the resource with the highest bid as the winner, and allocates the task to the resource with the winning bid. The virtual planning center takes turns auctioning all tasks until all tasks are allocated or cannot be observed. The second method is to assign tasks (AUS) in order of airship, drone and satellite, and there is an article indicating that this sort can achieve the highest task completion rate (Wu,2016) in a single task continuous assignment method. In the same category of resources, the method preferentially allocates the tasks to the resources with larger observation benefits (the observation benefits are in direct proportion to the weight of the tasks and in inverse proportion to the distance from the resources to the tasks). Both the SSA and AUS methods are iterative assignment methods for a single task, and neither of them considers cooperation and cooperation between resources. While the method three (MCP) and the method four (BCP) are centralized collaborative planning methods, they consider how the observation resources perform tasks collaboratively from a global perspective. In the third method, a Mosek optimization tool is selected as a solver to solve the task allocation model, and a proper optimization algorithm can be selected to obtain a more accurate planning scheme. The method solves the problems by using a four-choice Branch-and-bound algorithm (Branch and bound), and the BnB algorithm has higher optimization efficiency, so that the distribution result can be obtained relatively quickly. The integer programming model shown in the formula (1) is an abstract modeling of a collaborative programming problem, both the MCP method and the BCP method are used as task allocation models, and the optimization goal is the maximization of global benefits. In order to verify the performance of the method under the pressure of a large-scale task, three important indexes of five methods are compared respectively: task completion rate, method operation efficiency, and planning solution rationality. As shown in the formula (6), the average energy consumption of the observation task is used as the basis for evaluating the rationality of the scheme, and the smaller the average energy consumption is, the more rational the scheme is.
In the formula, Si, k represents the sum of the distances that the resources ri in the assigned mission plan center Pk need to move to complete. Ni, k represents the number of tasks that the resources ri in the planning center Pk can complete.
As shown in fig. 6 and 7, the MCP and BCP methods are centralized collaborative planning methods. Due to the cooperation and cooperation among the resources, the method can complete more tasks than a single task continuous distribution method. Due to the advantages of the Mosek optimizer in the aspect of solving accuracy, the MCP method based on global cooperation achieves the highest task completion rate, and the completion rate of 100% in the available results is kept (all task quantities cannot be tested due to computer memory overflow errors). However, the centralized computing method causes the computing efficiency of the MCP and BCP methods to be extremely low, so that it is difficult to deal with the emergency of large-scale tasks. The solution time of the MCP method is increased explosively with the continuous increase of tasks, and when the task amount reaches about 400, the solution time even needs dozens of hours, so that a computer cannot bear the memory pressure. This also proves that the Mosek solver can obtain a relatively better solution but has a problem of low computational efficiency. According to the experimental result, the solving efficiency is considered to be more unacceptable in case of emergency rescue compared with the high completion rate of the MCP method. In addition, due to the characteristic of fast search of the branch-and-bound method, the BCP method has higher solving efficiency (fig. 7) than the MCP method, but as can be seen from the curve fluctuation of the task completion rate in fig. 6, the BCP method has certain randomness in the aspect of solving accuracy and is difficult to obtain an optimal solution.
The SSA method and the AUS method are iterative distribution processes for a single task, so that the SSA method and the AUS method have strong similarity in the aspects of task completion rate, calculation efficiency and the like. The two methods have only one iteration process for all tasks, and only one task is allocated to each iteration, so the computational efficiency is obviously higher than that of the MCP and BCP methods (figure 7)The calculation consumption time is strongly and positively correlated with the number of tasks, i.e. Tmethod∝NtaskWherein T ismethodTime taken to solve the expression method, NtaskIndicating the number of tasks that need to be planned. However, due to the lack of inter-resource coordination mechanisms, the task completion rate indicators of the SSA and AUS methods are clearly at a disadvantage (fig. 6). The SSA method selects an optimal solution according to the highest bid price of all resources, and the AUS method only considers the income of single-class resources, so the task completion rate of the SSA method is slightly higher than that of the AUS method, but the operation efficiency of the AUS method is better.
By integrating all planning methods, the completion rate and the solution efficiency of the planning scheme are often not compatible, and the method achieves balance between the completion rate and the solution efficiency. The MICA method has a task completion rate very close to that of the MCP method, and even under the pressure of 500 large-scale tasks, the MICA method still achieves about 95% of task completion rate, and is remarkably superior to other three planning methods (figure 6). In the MICA method, when the number of tasks reaches about 318, the resource capacity gradually becomes saturated, the task completion rate starts to slowly decrease (fig. 6), and the algorithm running time is also changed from a steady increase to a rapid rise state (fig. 7). To explore the internal mechanism of the rapid rise of the MICA method runtime, we decomposed the MICA method runtime according to three levels from bottom to top in the distributed contract network collaboration framework (FIG. 8). In the resource coordination layer, tasks are allocated to communicable neighborhood resources, and the layer has high allocation efficiency by adopting three optimization strategies and a label selection algorithm. And along with the increase of the task quantity, the resource gradually enlarges the bidding scope, thereby causing the communication times scale in the process of bidding of the contract network to be enlarged, and further causing the rapid reduction of the operation efficiency. Once the observation power of the neighborhood resources reaches saturation, the incomplete tasks are passed into the project center internal protocol layer (fig. 8 (b)). At this level, the planning center bids on internal resources, as shown in fig. 8(b), when the number of tasks is between 250 and 350, the sparseness of the data points in the graph indicates that only a small number of tasks need to be allocated at the second level. And as the number of tasks increases, the points in the graph become more dense, so that more tasks need to be distributed in the second layer or even the third layer. Resulting in an increase in solution time, but still below the first level. In the planning center cooperation layer, the planning center tenders the remaining uncompleted tasks to other planning centers, and the solution efficiency of the third layer is the highest because most tasks are distributed in the first two layers. In general, the MICA method is significantly superior to other comparative methods in terms of algorithm solution efficiency due to the adoption of distributed computation and local search algorithms (fig. 7).
The rationality index of the planning scheme can reflect the quality of the planning scheme, and the smaller the numerical value of the rationality index is, the smaller the energy consumed by the single task is averagely. The results of the rationality of the solution have a high correlation with the task completion rate, the higher the task completion rate the higher the quality of the planned solution is relatively (fig. 9). In addition, as the number of tasks increases, resource capacity is more reasonably and compactly utilized, and the average flight of executing tasks tends to become smaller.
2) Rapid re-planning comparison verification of different planning methods in dynamic environment
We designed a set of experiments to verify the effect of dynamic reprogramming of the MICA method. When the resource is executing the earth observation task, an emergency task needing to be observed suddenly occurs, which often means that the resource observation capability is insufficient, and the new task site and the original planning scheme need to be comprehensively considered to determine whether to abandon some original tasks and make a new planning scheme. In order to verify the dynamic planning effect of the MICA method under the condition of resource capacity saturation, the number of tasks larger than the resource observation capacity is set. The simulation scene comprises 1 satellite, 18 unmanned aerial vehicles and 3 airships and is managed by 4 planning centers respectively. They were initially asked to perform 40 tasks that were randomly distributed in space. 30-50 emergency observation tasks can be generated in 6 continuous rounds of random dynamic processes when the observation resources are executing the tasks, and new observation schemes need to be continuously formulated for the observation resources on the basis of the original schemes according to different planning algorithms. The MCP, BCP and AUS methods recycle the unexecuted tasks and distribute the unexecuted tasks and the newly added tasks to all resources. The SSA method and the MICA method take the new task as a candidate task to target all resources, and the resources feed back a profit value according to the current position and the residual observation capability and decide which new tasks to observe and which original tasks to abandon.
In the dynamic re-planning process, the task completion rate is not the only concerned index. The process of executing tasks by resources is in a continuous and dynamic changing environment, and the re-planning time and the observation scheme change rate (the ratio of the number of changed tasks in the original scheme to the number of all tasks in the original scheme) are key factors for determining the quality of the re-planning method. The shorter the reprogramming time is, the less energy is consumed by the resource waiting for intermission, and the remote sensing data of the monitoring area can be obtained earlier. In addition, on the basis of ensuring the observation yield, the smaller the change rate of the existing scheme is, the less the action adjustment of the observation resource is, and the better the planning scheme is. Therefore, the invention compares the task completion rate, re-planning time and scheme change rate indexes of different methods when continuously inserting new tasks.
According to the experimental results shown in table 4, in the process of dynamically inserting new tasks for multiple times, the number of tasks completed by different planning methods is continuously increased, but the task completion rate tends to be decreased. Similar to the results of experiment (1), the task completion rate of the MCP method is generally superior to that of other methods, and the methods based on continuous assignment of a single task (AUS and SSA methods) are still inferior in the task completion rate (fig. 10). However, the task completion rate is not the only indicator of interest to the dynamic planning process. In a dynamic uncertain environment, a planning scheme is required to be formulated as fast as possible, and an original planning scheme is required to be maintained as far as possible on the basis of ensuring observation income. As shown in fig. 11, the re-planning time of different methods increases with the increase of the re-planning times, and is in positive correlation with the total number of tasks, wherein the computation efficiency of the MCP method is much lower than that of other planning methods. And because a distributed computing mode is adopted and the label selection algorithm is optimized, the MICA method has obvious advantages in the aspect of solving efficiency, the algorithm consumption time of continuous 6-round dynamic planning is in a steady ascending trend, and the algorithm consumption time is always maintained at a lower level.
The scheme change rate can reflect the influence of the newly added tasks on the original planning scheme, and different planning methods have different influences on the original planning scheme. As shown in fig. 12, the rate of change of the scenario gradually decreases as the number of times of dynamic programming increases, which is actually caused by the fact that the cardinality of the tasks in the last scenario is too large and the number of new tasks is too small. Therefore, we indirectly reflect the effect of the task base on the rate of change of the scenario with the task ring ratio growth rate (the ratio of the number of newly added tasks to the total number of tasks in the previous scenario). If the marked point of the rate of change of the recipe is inside the histogram, this indicates that the method changed fewer than the number of new tasks. Instead, this method changes more tasks than the number of new tasks. For example, in the 3 rd round of dynamic planning, most of the plan change rate markers of the planning methods are located outside the histogram, which means that the new observation plans of the methods all change more than 33 tasks in the original plans. The rate of change of the MICA method of the present invention is always kept below the mission ring ratio growth rate and is superior to other planning methods. Therefore, whether the influence of the task base is ignored or not, the scheme change rate of the MICA method is better than that of other methods, and the dynamic re-planning requirement is better met.
TABLE 4 comparison of dynamic re-planning results for different methods
Note: NT represents the number of tasks that newly occur to be observed; the AT indicates the number of all tasks. TCR represents task completion rate index; RPT represents a re-planning time index; RSC represents the rate of change indicator of the original scheme.
Although the embodiments of the present invention have been described in detail with reference to the accompanying drawings, the embodiments of the present invention are not limited to the details of the above embodiments, and various simple modifications can be made to the technical solutions of the embodiments of the present invention within the technical idea of the embodiments of the present invention, and the simple modifications all belong to the protection scope of the embodiments of the present invention.
It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. In order to avoid unnecessary repetition, the embodiments of the present invention do not describe every possible combination.
In addition, any combination of various different implementation manners of the embodiments of the present invention is also possible, and the embodiments of the present invention should be considered as disclosed in the embodiments of the present invention as long as the combination does not depart from the spirit of the embodiments of the present invention.
Claims (4)
1. A dynamic planning method for earth observation resources based on a contract network mechanism is characterized in that a bottom-up distributed collaborative planning architecture and a contract network-based collaborative planning process are adopted, the contract network-based collaborative planning process is subjected to task allocation in an active bidding mode of the earth observation resources, and the active bidding mode comprises the following five stages:
phase 1: resource riReceiving task sequence T ═ T (T) to be inserted1,t2,…tn) When j belongs to the n instruction, actively judging whether T can be completed according to the cache task to be completed and T;
phase 2: if r isiCan not complete task T, then riChanges the role of the target to a sender and sends the target to a resource group RS capable of communicating with the targeti-LReleasing bidding information; according to RSi-LJudging whether the task T can be completed or not by the returned scalar value and the completion information;
phase 3: if all resources cannot complete T, then riFeeding back task information to a sub-planning center P at the upper levelk；PkServing as a resource to be issued to the inside of the planning center by the issuer, riThe role of (2) is changed from a publisher to a bidder; pkJudging whether resources can complete T or not according to the returned scalar values and the completed task subsets;
phase 4: if no resources complete T, PkAgain as a publisher to set PS to a communicable neighborhood planning centerk-LReleasing the bidding document; each planning center [ P1,P2,…Pk-1,Pk+1…Ppn]Respectively appointing task subsets which can be completed according to internal resources and feeding back the task subsets; pkEvaluating the final task according to the feedback resultA distribution scheme;
phase 5: if P iskAccording to the neighborhood set PS when the neighborhood planning center can not complete Tk-LThe communication neighborhood of (2) continues to expand the bid sending range until the task T can be completed or no planning center remains;
and dynamically distributing the large-scale concurrent tasks by a multi-round incomplete combination distribution method.
2. The method for dynamically planning earth observation resources based on contract network mechanism according to claim 1, wherein the bottom-up distributed collaborative planning architecture comprises the following four levels:
the task management layer is responsible for receiving tasks submitted by users, dividing the large-scale tasks into different task sets according to the resource observation capacity and further distributing the task sets to corresponding resources of the next level;
the resource coordination layer comprises a plurality of heterogeneous resources, the observation resources receive the task set to be completed and are distributed to a neighborhood resource set capable of communicating with the observation resources through negotiation with the task to be re-planned; on the basis, the resource node submits the unallocated tasks to the next level;
the planning center internal negotiation layer assigns the unallocated tasks to the internal resources governed by the planning center on the basis of internal negotiation according to the resource observation capacity;
and the planning center cooperates with the layer, and the planning center continuously submits the remaining uncompleted tasks to other planning centers for bidding.
3. The method for dynamically planning earth observation resources based on contract network mechanism according to claim 1, wherein the multi-round incomplete combination allocation method allocates task sets in three levels from bottom to top in sequence respectively:
at the first level, according to a contract network method, each resource is used as a bidding direction to release a bidding document to a communication neighborhood resource set of the resource, and a combined task allocation scheme is determined through resource negotiation;
at the second level, each planning center receives uncompleted tasks and appoints resources inside the planning center to complete the tasks;
at the third level, all planning centers negotiate together to determine the allocation scheme for the incomplete tasks.
4. The method for dynamically planning earth observation resources based on contract network mechanism according to any one of claims 1-3, characterized in that a local search algorithm based on floating bid selection mechanism is adopted to solve the problem of contract network winning bid determination in the heterogeneous resource collaborative planning process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811351029.9A CN109409773B (en) | 2018-11-14 | 2018-11-14 | Dynamic planning method for earth observation resources based on contract network mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811351029.9A CN109409773B (en) | 2018-11-14 | 2018-11-14 | Dynamic planning method for earth observation resources based on contract network mechanism |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109409773A CN109409773A (en) | 2019-03-01 |
| CN109409773B true CN109409773B (en) | 2022-04-08 |
Family
ID=65472973
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811351029.9A Active CN109409773B (en) | 2018-11-14 | 2018-11-14 | Dynamic planning method for earth observation resources based on contract network mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409773B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111124637B (en) * | 2019-12-09 | 2023-08-29 | 中国空间技术研究院 | Dynamic task planning method for distributed satellite system |
| CN112101747B (en) * | 2020-08-28 | 2023-11-03 | 上海交通大学 | Unmanned ship interception task allocation method based on tabu consensus auction algorithm |
| CN112181658B (en) * | 2020-09-30 | 2024-04-05 | 南京工程学院 | Calculation task allocation method for maximizing network benefits in heterogeneous network |
| CN112991745B (en) * | 2021-04-30 | 2021-08-03 | 中南大学 | Traffic flow dynamic cooperative allocation method under distributed framework |
| CN113240324A (en) * | 2021-06-02 | 2021-08-10 | 中国电子科技集团公司第五十四研究所 | Air and space resource collaborative planning method considering communication mechanism |
| CN113313414B (en) * | 2021-06-21 | 2024-05-14 | 哈尔滨工程大学 | Task collaborative planning method for multi-class heterogeneous remote sensing platform |
| CN113537782B (en) * | 2021-07-19 | 2023-08-18 | 福州大学 | Multi-satellite situation awareness system distributed task planning method based on contract network |
| CN113487220B (en) * | 2021-07-23 | 2022-07-15 | 中南大学 | Static target observation-oriented space-sky heterogeneous earth observation resource cooperative scheduling method |
| CN115616902B (en) * | 2022-11-07 | 2023-03-07 | 中国人民解放军国防科技大学 | Cluster spacecraft task allocation method and device based on robust auction algorithm |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107180309A (en) * | 2017-05-31 | 2017-09-19 | 中南大学 | The collaborative planning method of resource is observed in a kind of empty world |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7908304B2 (en) * | 2001-03-15 | 2011-03-15 | Versata Development Group, Inc. | Method and system for managing distributor information |
| CN102903028B (en) * | 2012-09-28 | 2016-01-20 | 中国人民解放军国防科学技术大学 | A kind of towards emergent many dirigibles earth observation task cooperation distribution method |
| CN108256671A (en) * | 2017-12-26 | 2018-07-06 | 佛山科学技术学院 | A kind of more resources of multitask based on learning-oriented genetic algorithm roll distribution method |
-
2018
- 2018-11-14 CN CN201811351029.9A patent/CN109409773B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107180309A (en) * | 2017-05-31 | 2017-09-19 | 中南大学 | The collaborative planning method of resource is observed in a kind of empty world |
Non-Patent Citations (2)
| Title |
|---|
| 基于勘探和开采策略控制的智能优化算法及其应用研究;伍国华;《中国博士学位论文全文数据库工程科技I辑》;20160115;第B021-31页 * |
| 多类异构对地观测平台协同任务规划方法;王慧林,伍国华等;《航空学报》;20151204;第3卷(第37期);第997-1014页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109409773A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109409773B (en) | Dynamic planning method for earth observation resources based on contract network mechanism | |
| CN112016812B (en) | Multi-unmanned aerial vehicle task scheduling method, system and storage medium | |
| CN112766813B (en) | Space-day collaborative observation complex task scheduling method | |
| US8255344B2 (en) | Systems and methods for parallel processing optimization for an evolutionary algorithm | |
| Abdel-Basset et al. | A comprehensive review of quadratic assignment problem: variants, hybrids and applications | |
| CN108880663A (en) | Incorporate network resource allocation method based on improved adaptive GA-IAGA | |
| CN103744733B (en) | Method for calling and configuring imaging satellite resources | |
| CN101237469B (en) | Method for optimizing multi-QoS grid workflow based on ant group algorithm | |
| CN103345504A (en) | Operator construction method of single-star scheduling | |
| CN113705866B (en) | Scheduling optimization method and system based on resource-constrained project scheduling problem model | |
| CN105929690B (en) | A kind of Flexible Workshop Robust Scheduling method based on decomposition multi-objective Evolutionary Algorithm | |
| CN101836190B (en) | Method and system for distributing a plurality of jobs to a plurality of computers | |
| CN104842564B (en) | A kind of 3 D-printing multitask Optimization Scheduling based on NSGA II | |
| CN112633654A (en) | Multi-unmanned aerial vehicle task allocation method based on improved cluster expansion consistency bundle algorithm | |
| CN104077634B (en) | active-reactive type dynamic project scheduling method based on multi-objective optimization | |
| CN109523178A (en) | A kind of O&M method and device towards power communication scene | |
| CN110456633A (en) | Airborne multi-platform distributed task scheduling distribution method | |
| CN111784211A (en) | Cluster-based group multitask allocation method and storage medium | |
| CN108199763B (en) | Repeater satellite method for scheduling task and system | |
| CN109409746A (en) | A kind of production scheduling method and device | |
| CN109886574A (en) | A kind of multi-robot Task Allocation based on improvement threshold method | |
| Xun et al. | Distributed tasks-platforms scheduling method to holonic-C2 organization | |
| CN113240324A (en) | Air and space resource collaborative planning method considering communication mechanism | |
| CN114565203A (en) | Multi-satellite multi-task planning method based on large neighborhood search algorithm and K-means | |
| CN110689320A (en) | Large-scale multi-target project scheduling method based on co-evolution algorithm |
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 |