WO2019127948A1

WO2019127948A1 - Hierarchical distributed autonomous collaborative task planning system for intelligent remote sensing satellite

Info

Publication number: WO2019127948A1
Application number: PCT/CN2018/080422
Authority: WO
Inventors: 邢立宁; 何敏藩; 白国庆; 吕欣; 王炯琦; 伍国华; 陈剑; 黄勇; 熊彦; 文翰
Original assignee: 佛山科学技术学院; 佛山市有义家科技有限公司
Priority date: 2017-12-26
Filing date: 2018-03-26
Publication date: 2019-07-04
Also published as: CN108335012A

Abstract

A hierarchical distributed autonomous collaborative task planning system for an intelligent remote sensing satellite. The system comprises a multi-satellite task coordinator and an on-satellite scheduler, wherein the multi-satellite task coordinator distributes tasks in a task set, which is to be distributed, to a plurality of administered intelligent satellites and processes the tasks into meta-tasks which can be directly identified by the on-satellite scheduler; each intelligent satellite uses the on-satellite scheduler thereof to carry out uniform scheduling on a new distributed task and an existing task; and before the multi-satellite task coordinator carries out task distribution, a scheduling result of the relevant on-satellite scheduler is estimated in advance and is taken as a basis for task distribution. Key scientific problems and technical problems regarding imaging task schedulability prediction, multi-task multi-resource dynamic scheduling under a high-dimensional space, refined planning and scheduling algorithm design are solved, multi-satellite autonomous collaborative task planning technology can be better applied to the field of national defence construction, and the rationalization of a distribution solution and the efficiency of resource utilization are promoted.

Description

Hierarchical distributed autonomous collaborative mission planning system for intelligent remote sensing satellites

Technical field

The invention relates to the field of remote sensing satellite technology, in particular to a hierarchical remote distributed collaborative mission planning system for intelligent remote sensing satellites.

Background technique

Remote sensing satellites are artificial satellites used as remote sensing platforms for outer space. Typically, remote sensing satellites can operate in orbit for several years. Satellite orbits can be determined as needed. Remote sensing satellites can cover the entire Earth or any designated area within a specified time. When operating along geosynchronous orbit, it can continuously remotely sense a designated area on the Earth's surface. All remote sensing satellites require remote sensing satellite ground stations. Satellite data obtained from remote sensing market platforms can be used to monitor agriculture, forestry, oceans, land, environmental protection, meteorology, etc. Remote sensing satellites mainly include meteorological satellites, terrestrial satellites and marine satellites. Three types.

Remote sensing satellite technology has made great progress. The working modes and usage constraints of different remote sensing satellites are very complex. Generally, there are relatively independent mission planning systems. The current remote sensing satellite technology still has the following problems:

(1) The complexity of the task planning problem. Intelligent satellite mission planning has certain specialities in tasks, resources, constraints and optimization objectives. Common resource scheduling models and optimization methods are difficult to solve.

(2) The complexity and uncertainty of the scheduling algorithm. The randomness of the scheduling algorithm makes the scheduling result also uncertain, and also increases the difficulty of schedulability prediction.

(3) The complexity of task sample selection. Different satellites will accumulate a large amount of historical mission data during the orbital operation. How to select typical representative samples to improve the execution efficiency of the prediction algorithm is difficult.

(4) The complexity of sample feature extraction. Imaging tasks generally have both static and dynamic attribute characteristics: static attributes are mainly related to tasks that do not change with the change of the task set, such as the data type, resolution, priority, and demand observation duration of the imaging task. Meteorological conditions and imaging modes; dynamic attributes change with the set of tasks, such as describing resource competition between tasks and conflicts of observation opportunities. How to choose the characteristics that have a decisive influence on the prediction process among the various attributes is also very complicated.

(5) The number of imaging observations submitted by many users in different industries will exceed tens of thousands. How to effectively allocate a large number of imaging tasks submitted by multiple users to different remote sensing satellites lacks effective theoretical and technical support.

(6) Remote sensing satellite mission planning has complex constraints, unpredictable state information and complicated demand categories, making satellite mission planning problems a difficult point in the field of system engineering.

(7) With the rapid increase of the number of remote sensing satellites and imaging tasks, how to transmit the high-resolution remote sensing images obtained by the new satellites to the ground users quickly (in time, multi-task multi-ground station scheduling), there is still no effective technical support.

Summary of the invention

In order to solve the problems in the prior art, the object of the present invention is to provide a hierarchical remote distributed cooperative task planning system for intelligent remote sensing satellites, which successfully solves the schedulability prediction of imaging tasks and multi-tasks in high-dimensional space. Key scientific problems such as resource dynamic scheduling, refined planning and scheduling algorithm design and corresponding technical problems enable multi-star independent collaborative mission planning technology to be better applied to the actual defense construction field, which promotes the rationalization of distribution schemes and promotes Resource usage is more efficient.

In order to achieve the above object, the technical solution adopted by the present invention is:

An intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system, comprising a multi-star task coordinator and an on-board scheduler, the multi-role task coordinator assigning tasks in a task set to be assigned to a plurality of subordinates The intelligent satellite processes the task into a meta-task directly recognized by the on-board scheduler, and each intelligent satellite uses its on-board scheduler to uniformly schedule the assigned new task and the existing task, wherein the multi-star task coordinator is Before the task assignment, the scheduling result of the relevant on-board scheduler is estimated in advance, and this is used as the basis for task assignment, and the hysteresis of the feedback of the late scheduling result is avoided.

Preferably, the multi-star task coordinator allocates a task set in the rolling window to a plurality of intelligent satellites under the jurisdiction, and each intelligent satellite uses its on-board scheduler to schedule the assigned new task and the existing task, At the beginning of the current scrolling window, the multi-star task coordinator updates the task information, deletes the tasks that have been completed in the previous scrolling window, and the tasks that are being executed at the starting time, and does not allocate the last scrolling window. Tasks, and new tasks arriving within the previous scrolling window are combined into a set of tasks within the current scrolling window, and the multi-role task coordinator allocates the set of tasks to the plurality of smart satellites, wherein, based on the blending The trigger mode is used to determine the starting time of the scrolling window. On the one hand, the scrolling assignment is triggered every other time period, the time period is constant or dynamically changes according to a preset rule; on the other hand, the system state changes when appearing The event triggers a rolling assignment when it is subject to human intervention.

Preferably, in one aspect, the time period is set according to a measurement and control cycle; on the other hand, the event that changes a state of the system includes: receiving an emergency observation task, and the accumulated unallocated emergency observation task reaches five pieces or is 5% of the number of intelligent satellites under the multi-star mission coordinator.

Preferably, the multi-satellite task coordinator comprises a ground station and a geostationary orbit communication satellite, wherein the ground station performs task assignment within a measurement and control period; and the geostationary orbit communication satellite performs task assignment outside the measurement and control period And the emergency observation task is generated by the smart satellite, wherein the geostationary orbit communication satellite performs task assignment only to an intelligent satellite with a communication loop when performing the allocation, and at this time, the multi-star mission The number of intelligent satellites under the coordinator refers to the number of intelligent satellites having communication loops with the geostationary orbit communication satellite.

In the present invention, task assignment and actual delivery do not have to be performed in real time. That is to say, the multi-star task coordinator can perform task assignment for all satellites under the jurisdiction at a certain point in time, but only transmits related tasks to the smart satellite currently having the communication loop in real time, as for the temporary absence of The intelligent satellite of the communication loop transmits the assigned task to the on-board scheduler within the next communication time window. Similarly, the scheduling results of the assigned tasks on the star are not necessarily fed back to the multi-star coordinator in real time. Even the on-board scheduler does not schedule after receiving a newly assigned task. This is good for arranging on-board resources and on-board work plans. Improve the predictability of the overall plan.

Preferably, the task scheduling strategy of the onboard scheduler of each intelligent satellite is as follows:

(1) At the T-driven scheduling time point, the full rescheduling strategy in the progressive method is used to generate a new task plan in the next cycle time interval, and the T-driven scheduling time point is based on the given time interval T To determine a specific scheduling time point lT, 0 ≤ l ≤ L, LT ≤ H < (L + 1) T, each time a scheduling time point lT is reached, the calculation of the latter scheduling interval [lT, (l + 1) T Mission plan, where l is a positive integer, T is the given time interval, L is the maximum number of T-drive scheduling, and H is the total scheduling interval.

(2) At the C ^* -driven rescheduling time point, the scheduling plan repair strategy in the revised method is adopted. When the satellite is operating in a given scheduling interval, if it is at a certain time t (0 < t < H) emergency satellite observation missions on the accumulated amount of C _t exceeds a given threshold value C ^*, calculation is performed rescheduling, wherein C ^* is the critical threshold cumulative emergency observation tasks,

Except for the above two scheduling moments, scheduling is not performed at any other time.

The scheduling algorithm at the T-driven scheduling moment is as follows:

Enter:

– a set of emergency observation tasks that have arrived and are not scheduled before the T-drive scheduling time point;

– a set of conventional observation tasks that have been received and are not scheduled before the T-Drive scheduling time point;

Output:

- a scheduling plan within the next time period T;

Specific steps are as follows:

Step 11 separately from

with

Select whether the time window falls into the conventional observation task and the emergency observation task in the next time period T, and generate a conventional observation task set to be solved.

And emergency observation task set

Step 12 will

with

Integrated into a collection of observation tasks;

Step 13 Sort the tasks in the integrated observation task set according to the set heuristic rules;

Step 14 According to the sorting, the tasks in the integrated observation task set are scheduled one by one to determine whether to join the tasks.

In the above, until the integrated observation task set has no more tasks to join

in,

Step 15 Output the schedule in the next time period T

The scheduling algorithm at the C ^* -driven rescheduling point in time is as follows:

Enter:

- a scheduling plan within this time period T and later than the C ^* -drive scheduling time point t;

- a set of emergency observation missions that have arrived and are not scheduled before the scheduling time point t;

Output:

- the revised schedule at time t,

Specific steps are as follows:

Step 21: According to the condition that the observation time window is in the time interval from the time t to the next T-drive scheduling time point, the task collection

Select emergency observation tasks to generate new task sets

Step 22 According to the set heuristic rules,

Sort the emergency observation tasks in the middle;

Step 23 Select one by one according to the new task order.

Emergency observation mission

Revise until

No more emergency observation tasks can be added

in,

Step 24 Output the revised schedule

Preferably, the intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system comprises a structured neural network module, wherein the structured neural network module uses a structured neural network to predict an image task schedulability, wherein The structured neural network module is constructed by the causality theory, and all the connection relationships between the nodes are constructed based on the causal relationship of the actual system.

Preferably, the intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system comprises a structured neural network module, wherein the structured neural network module uses a structured neural network to predict imaging task schedulability, wherein The structured neural network module is constructed by the causality theory, and all the connection relationships between the nodes are constructed based on the causal relationship of the actual system.

The invention also provides a hierarchical remote distributed collaborative task planning system for intelligent remote sensing satellites, which comprises a structured neural network module, a learning intelligent optimization module and a constraint inference module; the structured neural network module adopts a structured neural network pair The imaging task can be predicted by scheduling; the learning intelligent optimization module uses a learning intelligent optimization method to dynamically schedule multi-task and multi-resources in a high-dimensional space, which includes a learning genetic algorithm module and a learning ant colony algorithm module. The learning type genetic algorithm module performs rolling allocation of multi-task multi-resources by a learning type genetic algorithm module, and the learning type ant colony algorithm module performs rolling scheduling on a multi-task multi-station station by a learning ant colony algorithm; the constraint inference module The single-star autonomous task is planned through intelligent optimization and constraint reasoning; the autonomous collaborative task planning system distributes the tasks and distributes them through a hierarchical distributed autonomous collaborative task system.

In a preferred solution, the structured neural network module is constructed by a causal relationship theory, and all connection relationships between the nodes are constructed based on the causal relationship of the actual system.

In a further preferred solution, the learning intelligent optimization module adopts a combination of an intelligent optimization model and a knowledge model to perform integrated modeling; the intelligent optimization model searches for a feasible space of the optimization problem according to the “neighbor search” strategy, The knowledge model is to extract useful knowledge from the previous optimization process, and use the obtained knowledge to guide the subsequent optimization process of the intelligent optimization method.

In a still further preferred aspect, the constraint inference module includes logical constraint inference, temporal inference, and resource constrained inference.

In a further preferred solution, the logical constraint inference adopts a conditional triggering manner, and generates a new activity according to the condition and inserts; the time inference adopts a path consistency check and a constraint propagation technique of the time constrained network, so that the time value domain is reduced and time Constraint satisfaction; the resource constraint reasoning is based on the time network, describes the resource time network, calculates the distribution of resource consumption levels, finds defects according to the distribution, and adjusts the constraints between activities based on the defect management mechanism.

The autonomous collaborative task system includes a multi-star task coordinator and a plurality of single-star task schedulers that are independently distributed and connected to the multi-star task coordinator.

The task coordination method of the autonomous collaborative task system is: the multi-star task coordinator accepts the new task, and performs task constraint analysis on the new task; then assigns the task to each single-star task scheduler through the allocation algorithm, and the task is The meta-task directly processed by the single-star task scheduler is processed; finally, the single-star task scheduler executes the scheduling algorithm to generate an observation scheme of the respective observation resources to the respective satellites.

The multi-star task coordinator and the single-star task scheduler are two-way information connection, the single-star scheduling result of each single-star task scheduler is fed back to the multi-star task coordinator, and the unfinished task is composed of multi-star tasks. The coordinator is reassigned according to the status of other satellites.

By adopting the above technical solution, the intelligent remote sensing satellite hierarchical distributed independent autonomous task planning system of the present invention has the beneficial effects compared with the prior art:

1. The present invention adopts a hierarchical distributed autonomous collaborative task planning mode, and a feedback redistribution mechanism between the master control level coordinator and a plurality of single star task schedulers facilitates the rationalization of the distribution plan, thereby promoting The use of resources is more efficient.

2. Hierarchical task planning pre-allocation and distributed processing of complex problems, greatly reducing the complexity of solving problems, and using task schedulability prediction method to reasonably establish the connection relationship between two levels of decision variables. Task schedulability prediction can pre-estimate the scheduling results of the lower platforms in the top-level task pre-planning stage, which serves as the basis for task allocation, avoiding the blindness of the previous task assignment caused by the lag of the feedback of the later scheduling results.

3. The invention successfully tackles key scientific problems such as schedulability prediction of imaging tasks, multi-task multi-resource dynamic scheduling in high-dimensional space, refined planning and scheduling algorithm design, and corresponding technical problems, so that multi-star independent collaborative mission planning Technology can be better applied to the actual field of national defense construction.

4. The present invention uses a structured neural network to predict the schedulability of an imaging task. All the connection relationships between nodes of the structured neural network model are constructed based on the causal relationship of the actual actual system, and have strong model parameters. Interpretation ability effectively solves various defects in the traditional feedforward neural network model, such as unstructured model, slow convergence, difficult to determine the number of neurons, and local minimum.

5. Using the multi-task multi-resource dynamic rolling allocation mechanism, the complex dynamic scheduling problem is decomposed into multiple simple static scheduling sub-problems, and then the sub-problems are combined to replace the optimal solution of the original problem. Greatly reduce the difficulty of solving the original problem.

DRAWINGS

1 is a framework diagram of an autonomous collaborative mission planning system for an intelligent remote sensing satellite network according to the present invention;

2 is a schematic diagram of an overall scheme of an autonomous collaborative mission planning system for an intelligent remote sensing satellite network according to the present invention;

3 is a schematic diagram of a learning type intelligent optimization method according to the present invention;

4 is a schematic diagram of a hierarchical distributed collaborative task planning framework of the present invention;

FIG. 5 is a structural diagram of a hierarchical distributed autonomous collaborative task system according to the present invention.

Detailed ways

In order to make the objects, technical solutions and advantages of the present invention more comprehensible, the present invention will be further described in detail below with reference to specific examples. It is to be understood that the description is not intended to limit the scope of the invention. In addition, descriptions of well-known structures and techniques are omitted in the following description in order to avoid unnecessarily obscuring the inventive concept.

The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system of the invention comprises a multi-star task coordinator and an on-board scheduler, and the multi-star task coordinator assigns tasks in the task set to be assigned to multiple intelligences under the jurisdiction The satellite processes the task into a meta-task directly recognized by the on-board scheduler, and each intelligent satellite uses its on-board scheduler to uniformly schedule the assigned new task and the existing task, wherein the multi-star task coordinator is performing Before the task assignment, the scheduling result of the relevant on-board scheduler is estimated in advance, and this is used as the basis for task assignment to avoid the lag of the feedback of the later scheduling results.

Pre-estimate that any suitable method or strategy can be employed. An alternative method is to estimate based on task saturation or the amount of on-board resource idleness. For example, if there are 20% of the remaining resources on the star, and the newly added task is expected to occupy 15% of the resources, it is estimated that the scheduling can be successful. Another alternative is to make a decision based on the task priority level. For example, the remaining 20% of the resources on the star, and the newly added task is expected to occupy 25% of the resources, but the priority of the pending tasks on the star is lower than the newly assigned task, it is also estimated that the scheduling can be successful. For tasks that may be replaced, return to the coordinator for reassignment. Advantageously, a neural network is employed to perform the aforementioned pre-estimation or prediction. Thereby, it is possible to have higher prediction accuracy and prediction fineness.

The multi-star task coordinator allocates a set of tasks in the rolling window to a plurality of intelligent satellites under the jurisdiction, and each intelligent satellite uses its on-board scheduler to schedule the assigned new tasks and existing tasks, in the current scrolling window. The start time, the multi-star task coordinator updates the task information, deletes the tasks that have been completed in the previous scroll window, and the tasks that are being executed at the start time, and the unallocated tasks in the previous scroll window, And the new task arriving in the last scrolling window is combined into a task set in the current scrolling window, and the multi-star task coordinator allocates the task set to the plurality of smart satellites, wherein the hybrid triggering mode is based on Determining the starting moment of the scrolling window, on the one hand, triggering the scrolling assignment every other time period, the time period is constant or dynamically changing according to a preset rule; on the other hand, an event causing the system state to change or Triggers a rolling assignment when subjected to human intervention.

In an optional embodiment, the multi-star mission coordinator comprises a ground station and a geostationary orbit communication satellite, wherein the ground station performs task assignment within a measurement and control period; and the geostationary orbit is outside the measurement and control period The communication satellite performs task assignment, and the emergency observation task is generated by the smart satellite, wherein the geostationary orbit communication satellite only performs task assignment to the intelligent satellite with which the communication loop is provided when performing the allocation, at this time, The number of intelligent satellites under the multi-star mission coordinator refers to the number of intelligent satellites having communication loops with the geostationary orbit communication satellite.

In addition to the above two scheduling moments, the on-board scheduler does not schedule at any other point in time.

Advantageously, in the case where the rolling window of the multi-star task coordinator is a constant period of time, the period of the T-driven scheduling moment is equal to the constant period of time. In the case where the rolling window of the multi-star task coordinator is a regularly varying period of time, the period of the T-driven scheduling moment is equal to or less than the minimum length of the period.

In another alternative embodiment, each time the task assigned by the multi-star task coordinator is received, the on-board scheduler schedules the task once and feeds back the scheduling result.

Specifically, the scheduling algorithm at the scheduling moment of the T-drive is as follows:

Enter:

Output:

- a scheduling plan within the next time period T;

Specific steps are as follows:

Step 11 separately from

with

And emergency observation task set

Step 12 will

with

Integrated into a collection of observation tasks;

in,

Step 15 Output the schedule in the next time period T

Enter:

Output:

- the revised schedule at time t,

Specific steps are as follows:

Select emergency observation tasks to generate new task sets

Step 22 According to the set heuristic rules,

Sort the emergency observation tasks in the middle;

Step 23 Select one by one according to the new task order.

Emergency observation mission

Revise until

No more emergency observation tasks can be added

in,

Step 24 Output the revised schedule

The invention is shown in FIG. 1 , which surrounds a hierarchical distributed autonomous collaborative task planning architecture, a structuring prediction based on structured neural network, a multi-task multi-resource rolling allocation based on a learning genetic algorithm, based on intelligent optimization and Five main research contents of single-star autonomous task planning with constrained reasoning and multi-task multi-ground station rolling scheduling based on learning ant colony algorithm, focusing on schedulability prediction of imaging tasks and multi-task multi-resource dynamic scheduling in high-dimensional space Key scientific issues such as refined planning and scheduling algorithm design and corresponding technical methods enable multi-star independent collaborative mission planning technology to be better applied to the actual defense construction field.

The invention mainly uses a research method of structured neural network, learning intelligent optimization method and constraint reasoning, and the overall scheme is shown in FIG. 2 .

(1) Structured neural networks. In view of the limitations of standard neural networks with a "black box model", the present invention employs a structured neural network to predict the schedulability of imaging tasks. The structured neural network is constructed based on the causality theory. All the connection relationships between the nodes are constructed based on the causal relationship of the actual actual system. It has strong ability to interpret the model parameters (each parameter has practical interpretability). meaning). The structured neural network model effectively solves various defects in the traditional feedforward neural network model, such as unstructured model, slow convergence rate, difficult to determine the number of neurons, and local minimum.

By constructing and extracting the feature values of the task planning result set, a structured neural network model for schedulability prediction of imaging tasks is designed. The paper focuses on how to dynamically adjust the structure based on the accumulated operational data of the satellite mission planning system. Structural models and related parameters of neural networks. The structured neural network model can establish a nonlinear mapping relationship between task eigenvalues and satellite capabilities during the learning process, so as to be able to predict the schedulability of imaging tasks.

(2) Learning-type intelligent optimization method. In the doctoral research stage, the inventor built a learning-based intelligent optimization method for complex optimization problems based on evolution and learning mechanism: an integrated modeling approach combining intelligent optimization model and knowledge model, intelligent optimization model according to “neighbor search” The strategy searches for the feasible space of the optimization problem; the knowledge model mines some useful knowledge from the previous optimization process, and then uses the acquired knowledge to guide the subsequent optimization process of the intelligent optimization method. The invention adopts a learning genetic algorithm and a learning ant colony algorithm to solve complex optimization problems such as multi-task multi-resource rolling allocation and multi-task multi-ground station rolling scheduling.

The invention aims at the large-scale characteristics of the intelligent satellite network autonomous collaborative task planning problem, comprehensively considers the task arrival time and the measurement and control time window to dynamically divide the scheduling period, and uses the heuristic rules and the forward-looking mechanism to adjust the original planning plan in a shorter period. Task-driven dynamic rolling planning and scheduling technology to achieve rapid response to dynamic environments and mission requirements. The genetic optimization model and the knowledge model are integrated to form a learning genetic algorithm, which can dynamically allocate tens of thousands of imaging tasks to hundreds of remote sensing satellites, which provides a useful reference for solving other similar high-dimensional assignment problems. The ant colony optimization model and the knowledge model are integrated to form a learning ant colony algorithm, which can dynamically schedule tens of thousands of data transmission tasks on hundreds of ground stations, providing effective technical support for solving other similar high-dimensional scheduling problems. In the learning intelligent optimization method for solving large-scale complex optimization problems, the innovative algorithm proposed by the present invention and its core operation are shown in FIG. 3.

Multi-task and multi-resource rolling allocation technology based on learning genetic algorithm. Based on the principle of rolling time domain control, the rolling mechanism of multi-task multi-resource dynamic allocation is constructed. The complex dynamic allocation problem is transformed into the static allocation problem of rolling update. The definition can reflect Several kinds of knowledge of the essential characteristics of multi-task multi-resource allocation problem, construct a knowledge model that can effectively manage these knowledge; design a genetic optimization model based on genetic algorithm to design a multi-task multi-resource allocation problem; focus on genetic optimization model and knowledge model The integration and interaction mechanism ultimately form a learning genetic algorithm that efficiently integrates the genetic optimization model and the knowledge model.

A multi-task multi-station rolling scheduling technology based on learning ant colony algorithm, which is based on the rolling time domain control principle to construct a rolling mechanism of multi-task multi-station dynamic scheduling, and transforms the complex dynamic scheduling problem into a static scheduling problem of rolling update; Based on the expert knowledge, user preferences and experience information in the field of resource scheduling in the ground station, the knowledge model for assisting ground station resource scheduling is constructed. The feasible scheme of multi-task multi-station scheduling problem is designed based on ant colony algorithm. The ant colony optimization model focuses on the integration and interaction mechanism between the ant colony optimization model and the knowledge model, and finally forms a learning ant colony algorithm that integrates the ant colony optimization model and the knowledge model efficiently.

(3) Constraint reasoning technology. Constraint reasoning mainly includes three parts: logical constraint reasoning, time reasoning and resource constraint reasoning. Logical reasoning mainly uses conditional triggering to generate new activities based on conditions and insert them. Time reasoning mainly uses the path consistency check and constrained propagation techniques of time-constrained networks to achieve time-domain reduction and time-constrained satisfaction. Resource reasoning is based on the time network. The resource time network describes the problem. Because the activity changes the resource state in a relative way, it is necessary to calculate the distribution of resource consumption level, find the defect according to the distribution, and adjust the activity and activity based on the defect management mechanism. constraint.

Single-star autonomous mission planning technology based on intelligent optimization and constraint reasoning, which modularizes domain model, time and resource constraint reasoning and problem model, and integrates aerospace domain models such as attitude control model, battery model, solid model and antenna model. The integrated planning and scheduling framework with extensibility and versatility is constructed. The single-star autonomous mission planning technology based on intelligent optimization and constraint reasoning is constructed. The intelligent optimization module selects the combined variables as the combined variables for task and activity opportunities. The constraint inference module Process and conflict resolution of logical relationships, time and resource constraints in the task decomposition activity diagram. Considering the use of heuristic information and user preferences related to the satellite domain to guide constraint reasoning and plan generation, resulting in better mission planning results with less computational effort.

The technical route and experimental means for each research content of the present invention are sequentially given below.

1. Hierarchical distributed autonomous collaborative task planning architecture

The bi-level programming theory and model have unique adaptability to deal with decision-making optimization problems with multi-level characteristics, and are also very suitable for multi-satellite collaborative task planning under distributed cooperative mechanism. Distributed collaboration emphasizes the information interaction between sub-problems through the top-level coordination unit. The multi-satellite collaborative task planning problem under distributed cooperative mechanism is suitable for describing the mathematical model of bi-level programming problem. The related modeling and solving technology can be used for reference: the multi-satellite independent collaborative planning process is divided into the top-level multi-platform multi-task collaborative allocation and the underlying single The platform's autonomous planning is a two-in-one, tightly connected decision-making process (Figure 4). The upper layer in Figure 4 is a multi-platform multi-task dynamic allocation. Through this allocation process, tasks are assigned to individual observation resources according to task characteristics and resource characteristics.

Hierarchical distributed autonomous collaborative task planning mode, based on centralized collaborative task planning, adds a multi-star task coordinator with total control level, and cancels the multi-star joint scheduler, which is still used for the scheduling of each satellite. Dedicated single-star task scheduling, as shown in Figure 5, the multi-star task coordinator has a high level of control, and the multi-star task coordinator performs task constraint analysis on the task, according to the task requirements and the state of the observing resources under the jurisdiction, through a specific allocation algorithm. The task is assigned to each observation resource, and the task is processed into a meta-task directly recognized by the single-star scheduler, and then the single-star scheduler executes the scheduling algorithm to generate an observation plan of the respective observation resources. Each single-star scheduler can feed back the single-star scheduling result to the multi-star task coordinator. For unfinished tasks, the multi-star task coordinator can be re-allocated according to the state of other satellites, and can be promoted through several feedback redistribution mechanisms. Rationalize the distribution plan to promote more efficient use of resources. When the new task arrives less, the multi-star task coordinator can assign the task to one or several satellites according to the task characteristics and the existing single-star task execution plan, thereby triggering the scheduling process of the satellites. A single star that has not newly assigned a task continues to implement the existing solution, thereby achieving multi-star asynchronous control and enhancing the flexibility of observation resource management. This model avoids the complexity of unified modeling of multi-star scheduling to a certain extent, and hierarchically processes the scheduling problem, which enhances the system's reusability and enhances system scalability. If the satellite is temporarily added or temporarily reduced. , only need to be modified at the multi-star task coordinator.

Hierarchical task planning pre-allocation and distributed processing of complex problems greatly reduces the complexity of solving problems. However, whether the connection between two levels of decision variables can be reasonably established is the basis for determining the effectiveness of hierarchical task planning. The essential. The present invention solves this convergence problem by studying a reasonable task schedulability prediction method. The role of task schedulability prediction is to pre-estimate the scheduling results of the lower platforms in the top-level task pre-planning stage, which serves as the basis for task allocation, avoiding the blindness of the previous task assignment due to the lag of feedback of the late scheduling results. . To this end, the eigenvalues of the sample set are constructed and extracted by the actual historical data accumulated in the management process of the previous high-scoring system, and the agent model of resource scheduling is established based on the integrated BP neural network, so that the scheduling result is given by the model. prediction. The model can be updated when the actual scheduling results are fed back online.

The specific embodiments described above are merely exemplary in order to enable those skilled in the art to understand the present invention and are not to be construed as limiting the scope of the invention; any equivalent changes made in accordance with the spirit of the disclosure. Or modifications, are included in the scope of this patent.

Claims

An intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system, comprising: a multi-star task coordinator and an on-board scheduler, wherein the multi-star task coordinator assigns tasks in the task set to be assigned to the subordinate Multiple intelligent satellites and process the tasks into meta-tasks directly recognized by the on-board scheduler, each intelligent satellite uses its on-board scheduler to uniformly schedule the assigned new tasks and existing tasks, wherein the multi-star tasks The coordinator pre-estimates the scheduling result of the relevant on-board scheduler before performing task assignment, and uses this as the basis for task assignment.
The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system according to claim 1, wherein the multi-star task coordinator allocates a task set in the rolling window to a plurality of intelligent satellites under the jurisdiction, each intelligent The satellite uses its on-board scheduler to schedule the assigned new tasks and existing tasks. At the beginning of the current scrolling window, the multi-star task coordinator updates the task information and deletes the tasks that have been completed in the previous scrolling window. And a task being executed at the start time, and combining unallocated tasks in the previous scroll window and new tasks arriving in the previous scroll window into a task set in the current scroll window, and the multi-star The task coordinator allocates the task set to the plurality of smart satellites, wherein the start time of the scroll window is determined based on the hybrid trigger mode, and on the other hand, the scroll allocation is triggered every other time period, the time period is constant Or dynamically change according to pre-set rules; on the other hand, in the event of a change in the state of the system or in the person Triggering a rolling assignment when an intervention occurs.
The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system according to claim 2, wherein, on one hand, the time period is set according to a measurement and control cycle; on the other hand, the event that changes a system state includes : Received emergency observation missions, and the accumulated unallocated emergency observation missions reached five or 5% of the number of intelligent satellites under the multi-star mission coordinator.
The multi-task multi-resource rolling allocation method based on learning genetic algorithm according to claim 3, wherein the multi-star task coordinator comprises a ground station and a geostationary orbit communication satellite, within a measurement and control period, The ground station performs task assignment; the geostationary orbit communication satellite performs task assignment outside the measurement and control period, and the emergency observation task is generated by the smart satellite, wherein the geostationary orbit communication satellite is only in the Assigning a smart satellite with a communication loop to perform task assignment. At this time, the number of intelligent satellites under the multi-star mission coordinator refers to an intelligent satellite having a communication loop with the geostationary orbit communication satellite. Quantity.
The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system according to any one of claims 1 to 4, characterized in that the task scheduling strategy of the onboard scheduler of each intelligent satellite is as follows:

(1) At the T-driven scheduling time point, the full rescheduling strategy in the progressive method is used to generate a new task plan in the next cycle time interval, and the T-driven scheduling time point is based on the given time interval T To determine a specific scheduling time point lT, 0 ≤ l ≤ L, LT ≤ H < (L + 1) T, each time a scheduling time point lT is reached, the calculation of the latter scheduling interval [lT, (l + 1) T Mission plan, where l is a positive integer, T is the given time interval, L is the maximum number of T-drive scheduling, and H is the total scheduling interval.

(2) At the C * -driven rescheduling time point, the scheduling plan repair strategy in the revised method is adopted. When the satellite is operating in a given scheduling interval, if it is at a certain time t (0 < t < H) emergency satellite observation missions on the accumulated amount of C t exceeds a given threshold value C *, calculation is performed rescheduling, wherein C * is the critical threshold cumulative emergency observation tasks,

Except for the above two scheduling moments, scheduling is not performed at any other time.

The scheduling algorithm at the T-driven scheduling moment is as follows:

Enter:

An emergency observation task set that has arrived and is not scheduled before the T-drive scheduling time point;

a set of conventional observation tasks that have been received and are not scheduled before the T-drive scheduling time point;

Output:

a scheduling plan within the next time period T;

Specific steps are as follows:

Step 11 respectively
with
Select whether the time window falls into the conventional observation task and the emergency observation task in the next time period T, and generate a conventional observation task set to be solved.
And emergency observation task set

Step 12 will
with
Integrated into a collection of observation tasks;

Step 13 sorts the tasks in the integrated observation task set according to the set heuristic rules;

Step 14 sorts the tasks in the integrated observation task set one by one according to the ordering to determine whether to join the task
In the above, until the integrated observation task set has no more tasks to join
in,

Step 15 outputs the scheduling plan in the next time period T

The scheduling algorithm at the C * -driven rescheduling point in time is as follows:

Enter:

a scheduling plan within this time period T and later than the C * -drive scheduling time point t;

a set of emergency observation tasks that have arrived and are not scheduled before the scheduling time point t;

Output:

The revised schedule at time t,

Specific steps are as follows:

Step 21: according to the condition that the observation time window is in the time interval from the time t to the next T-drive scheduling time point, from the task set
Select emergency observation tasks to generate new task sets

Step 22 according to the set heuristic rules,
Sort the emergency observation tasks in the middle;

Step 23 selects one by one according to the new task order.
Emergency observation mission
Revise until
No more emergency observation tasks can be added
in,

Step 24 Output the revised schedule
The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system according to any one of claims 1 to 4, characterized by comprising a structured neural network module, wherein the structured neural network module adopts a structured neural network The network predicts the schedulability of the imaging task, wherein the structured neural network module is constructed by causality theory, and all the connection relationships between the nodes are constructed based on the causal relationship of the actual system.
The intelligent remote sensing satellite hierarchical distributed autonomous collaborative task planning system according to claim 6, comprising a structured neural network module, wherein the structured neural network module uses a structured neural network to schedule an imaging task The prediction is made, wherein the structured neural network module is constructed by causality theory, and all the connection relationships between the nodes are constructed based on the causal relationship of the actual system.