A multi-platform collaborative perception intelligent planning method and system

Abstract

The present invention proposes an intelligent planning method and system for multi-platform collaborative perception, which belongs to the technical field of remote sensing satellite planning. The method generates remote sensing requirements based on network open source information, historical mission information, system real-time status and multi-source guidance information, or accepts user remote sensing requirements, and the requirements arrive in order; constructs and maintains a dynamic task queue for existing remote sensing requirements, and triggers rolling intelligent planning when the event timeliness threshold or the measurement and control window threshold is reached, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned and formed. At the same time, the execution estimation of the planning scheme is carried out to provide commanders with auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability, and after the user or the on-board business agency adjusts the decision, a remote sensing plan is formed. The task queue is dynamically updated over time and driven by events.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing satellite planning, and in particular relates to an intelligent planning method and system for multi-platform collaborative perception. Background Art

[0002] At present, the space-based remote sensing operation process is a closed-loop link of "demand-planning-control-satellite-processing-user", which mainly includes demand acceptance and planning, mission control, satellite measurement and control, satellite remote sensing, data reception and transmission, information processing, calibration processing, processing and distribution applications, etc. The traditional passive mission control mode has redundant processes and many restrictions, and can no longer meet application requirements, which is mainly reflected in the following three aspects:

[0003] First, process redundancy does not meet the needs of mission assurance. The current management and control model for periodic planning has many links and a long chain. Each link is tightly coupled and serial. When changes are required, the requirements must start again from the previous link, and the process is not optimized enough. Moreover, each link is individually checked for constraints, with many levels and inconsistent granularity of constraints, which can easily lead to repeated processes, long processing response cycles, and insufficient rapid response capabilities for tasks.

[0004] Second, external resource guarantee factors affect the ability to respond quickly. Satellites fly in fixed orbits. Under the current control mode and technical capabilities, satellites can only upload mission instructions and download remote sensing data when they are within the measurement and control and data transmission range of ground stations, or upload instructions and download data through relay satellites, and then fly to the target area for observation. However, the number of ground and relay measurement and control, data transmission resources is small, there are many constraints, and the response is slow. The coordination and guarantee mechanism is still imperfect, which greatly restricts the effectiveness of space-based remote sensing systems, resulting in long rapid response time and poor emergency observation capabilities. At the same time, it is also limited by external factors such as meteorology and communications.

[0005] The third is that the passive acceptance of the mission mode limits the performance. The current response mode is that the application processing department accepts the remote sensing needs of all parties, conducts demand analysis and coordination, and then sends the remote sensing task to the control department for task conflict resolution and detailed payload constraint inspection. Since the application processing department does not have a detailed grasp of the satellite payload constraints and does not understand the real-time status and capabilities of the system, the demand and satellite resources cannot be optimally matched, making it difficult to maximize the system's performance, especially in the emergency response mode, which increases the task acceptance link and affects the system's response speed. Summary of the invention

[0006] In order to solve the above technical problems, the present invention proposes an intelligent planning solution with multi-platform collaborative perception.

[0007] The present invention proposes an intelligent planning method for multi-platform collaborative perception. The method comprises:

[0008] Step S1: Establishing a dynamic intelligent task planning model; specifically including:

[0009] Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data;

[0010] Step S2: Establishing an intelligent planning and scheduling kernel; specifically including:

[0011] The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan;

[0012] Step S3: Establishing a dynamic intelligent task planning process; specifically including:

[0013] Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection non-periodically, monitor task execution throughout the entire process and issue system fault warnings.

[0014] According to the method of the first aspect of the present invention, in step S1:

[0015] Conduct remote sensing demand mining based on multi-source guidance; specifically include: mining network open source information to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and mining observation opportunities, observation needs and observation tasks in combination with real-time remote sensing data feedback, thereby generating the remote sensing demand;

[0016] Dynamically maintain the remote sensing demand queue; specifically, including: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution;

[0017] The task planning based on the remote sensing demand is triggered by an event-driven method; specifically, the task planning is started after the event-driven trigger reaches a specified threshold, the demand set in the current demand queue is used as the task source, an optimization strategy is selected to perform multi-satellite joint intelligent task planning, and the nearest measurement and control window after the task planning is triggered is scheduled for instruction injection;

[0018] The optimization strategy of the task planning is adjusted according to the quantitative evaluation result of the task planning; specifically, each time the task planning is rolled out, the task planning scheme is quantitatively evaluated to determine the reliability and execution cost of the task planning scheme to update the optimization strategy.

[0019] Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically, the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing requirements are met;

[0020] Executing the observation task according to the task plan and feeding back the observation result data; specifically comprising: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task according to the observation data and the situation awareness information;

[0021] Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

[0022] According to the method of the first aspect of the present invention, in step S2, a measurement and control cycle is divided into multiple task planning time domains, a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan, a single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and no change is allowed after acceptance, after the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated, and after reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection, and the task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

[0023] According to the method of the first aspect of the present invention, in the step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

[0024] The second aspect of the present invention provides a multi-platform collaborative perception intelligent planning system. The system includes a processing unit configured to execute:

[0025] Step S1: Establishing a dynamic intelligent task planning model; specifically including:

[0026] Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data;

[0027] Step S2: Establishing an intelligent planning and scheduling kernel; specifically including:

[0028] The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan;

[0029] Step S3: Establishing a dynamic intelligent task planning process; specifically including:

[0030] Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection non-periodically, monitor task execution throughout the entire process and issue system fault warnings.

[0031] According to the system of the second aspect of the present invention, in step S1:

[0032] Conduct remote sensing demand mining based on multi-source guidance; specifically include: mining network open source information to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and mining observation opportunities, observation needs and observation tasks in combination with real-time remote sensing data feedback, thereby generating the remote sensing demand;

[0033] Dynamically maintain the remote sensing demand queue; specifically, including: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution;

[0034] The task planning based on the remote sensing demand is triggered by an event-driven method; specifically, the task planning is started after the event-driven trigger reaches a specified threshold, the demand set in the current demand queue is used as the task source, an optimization strategy is selected to perform multi-satellite joint intelligent task planning, and the nearest measurement and control window after the task planning is triggered is scheduled for instruction injection;

[0035] The optimization strategy of the task planning is adjusted according to the quantitative evaluation result of the task planning; specifically, each time the task planning is rolled out, the task planning scheme is quantitatively evaluated to determine the reliability and execution cost of the task planning scheme to update the optimization strategy.

[0036] Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically, the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing requirements are met;

[0037] Executing the observation task according to the task plan and feeding back the observation result data; specifically comprising: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task according to the observation data and the situation awareness information;

[0038] Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

[0039] According to the system of the second aspect of the present invention, in step S2, a measurement and control cycle is divided into multiple task planning time domains, a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan, a single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and is not allowed to change after acceptance, after the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated, and after reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection, and the task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

[0040] According to the system of the second aspect of the present invention, in the step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

[0041] The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in the multi-platform collaborative perception intelligent planning method described in the first aspect of the present disclosure are implemented.

[0042] The fourth aspect of the present invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the multi-platform collaborative perception intelligent planning method described in the first aspect of the present disclosure are implemented.

[0043] In summary, the architecture adopted by the technical solution of the present invention is relatively flexible and easy to expand; it can realize autonomous demand generation, rapid task response and intelligent task planning, and support heterogeneous platform resource expansion; compared with the traditional model, it has the advantages of flexible emergency adjustment and low resource loss, and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 Schematic diagram of an intelligent task planning mode according to an embodiment of the present invention.

[0046] Figure 2 Schematic diagram of an event-driven intelligent dynamic planning process according to an embodiment of the present invention.

[0047] Figure 3 A schematic diagram of a rolling task planning timing and scheduling kernel design according to an embodiment of the present invention.

[0048] Figure 4 The figure is a structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0050] The present invention proposes an intelligent planning method for multi-platform collaborative perception. The method comprises:

[0051] Step S1: Establishing a dynamic intelligent task planning model; specifically including:

[0052] Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data;

[0053] Step S2: Establishing an intelligent planning and scheduling kernel; specifically including:

[0054] The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan;

[0055] Step S3: Establishing a dynamic intelligent task planning process; specifically including:

[0056] Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection non-periodically, monitor task execution throughout the entire process and issue system fault warnings.

[0057] Specifically, the present invention studies and establishes a new model of intelligent task planning: multi-source guided demand mining, dynamically maintained demand queues, event-driven intelligent planning, and timely and effective decision support. One feature of this model is that it changes the traditional remote sensing demand analysis and task generation from passive to active. A large amount of user demand historical information data is analyzed using big data methods, the characteristics and attributes of the data are extracted, and a standardized data description method is established. A task history data learning model is designed based on a deep learning method, and the model is trained through open source and internal information. Then, based on the capability status of the aerospace system, the current political hot spots, major user behavior habits, historical remote sensing conditions, and hot target distribution information are integrated to extract the standardized input of the learning model, thereby deeply mining user needs and automatically generating standardized remote sensing tasks. The schematic diagram is as follows Figure 1 shown.

[0058] According to the method of the first aspect of the present invention, in step S1:

[0059] Remote sensing demand mining is carried out based on multi-source guidance; specifically, it includes: mining open source information on the Internet to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and combining real-time remote sensing data feedback to mine observation opportunities, observation needs and observation tasks, thereby generating the remote sensing demand.

[0060] Specifically, multi-source guided demand mining. Mining open source information such as the Internet to extract recent hot events; analyzing user preferences by mining information downloads and usage; mining and discovering new observation opportunities based on real-time remote sensing data feedback, generating new observation needs and tasks; and autonomously predicting and generating new remote sensing needs through learning and mining historical operation data.

[0061] Dynamically maintain the remote sensing demand queue; specifically include: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution.

[0062] Specifically, dynamically maintained demand queues. A demand queue is maintained in the system, and its demands mainly come from remote sensing demands proposed by various users; based on the latest remote sensing products, network status and user preferences, the next step of remote sensing demands is generated through event evolution prediction; new remote sensing demands are generated by learning and mining historical operation data. After the multi-source demands are aggregated, they are maintained through the demand queue. As time goes by, new demands are continuously added, and the demand optimization level is dynamically adjusted according to the changes in the situation. The demand disappears as the planning is triggered, and whether to generate new remote sensing demands is decided based on the remote sensing results.

[0063] Event-driven triggering of task planning based on the remote sensing demand; specifically comprising: starting the task planning after the event-driven trigger reaches a specified threshold, taking the demand set in the current demand queue as the task source, selecting an optimization strategy for multi-satellite joint intelligent task planning, and scheduling the nearest measurement and control window after the task planning is triggered for instruction injection.

[0064] Specifically, event-driven intelligent planning. When a key event is triggered or reaches a specified threshold, the mission planning is triggered and the demand set in the current demand queue is used as the mission source. The optimization strategy is selected for multi-satellite joint intelligent mission planning to maximize system efficiency and schedule the nearest measurement and control window after the planning is triggered for instruction injection. Mission planning has no fixed cycle and no fixed number of tasks. It is randomly initiated with event triggering. With the development of intelligent satellites and relay measurement and control, real-time planning of a single mission will eventually be achieved.

[0065] The optimization strategy of the task planning is adjusted according to the quantitative evaluation result of the task planning; specifically, each time the task planning is rolled out, the task planning scheme is quantitatively evaluated to determine the reliability and execution cost of the task planning scheme to update the optimization strategy.

[0066] Specifically, timely and effective decision support. When the system conducts each rolling planning, it generates a planning scheme based on the opinions of the demand processing department, and conducts a quantitative evaluation of the scheme, giving the reliability and cost of the scheme execution. The execution cost mainly includes task conflicts and losses. The quantitative evaluation results assist decision-making. If the scheme does not meet the requirements, the optimization strategy is changed and the planning scheme is readjusted and optimized until the quantitative evaluation results meet the requirements.

[0067] Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically including: the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing needs are met.

[0068] Specifically, quick application information acquisition. Users equipped with aerospace remote sensing information acquisition terminals can access at any time, submit information support applications in real time, and obtain aerospace remote sensing information. The platform resources have on-board processing and broadcast distribution capabilities. While transmitting data to the ground, the platform resources broadcast and distribute the processed information on the satellite to each terminal in real time. At the same time, the system has situational awareness capabilities and can perceive the aerospace remote sensing needs and satisfaction of each terminal in real time.

[0069] Execute the observation task according to the task plan and feed back the observation result data; specifically include: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task based on the observation data and the situation awareness information.

[0070] Specifically, real-time closed-loop data feedback. After the mission planning is completed and the instructions are generated, the remote sensing satellite will carry out the observation mission. After the data is transmitted to the ground for processing, the target remote sensing situation will be fed back to the mission control center in real time. The control center will make real-time decisions on whether to arrange the next round of missions based on the real-time target remote sensing situation and the latest situation awareness.

[0071] Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

[0072] Specifically, intelligent-driven autonomous control. With the development of technologies such as intelligent satellites, intersatellite links, and satellite-ground integrated networks, on-orbit real-time processing, on-board autonomous planning, and intersatellite guidance remote sensing are gradually being realized. The event-driven active control mode is designed around intelligence and autonomy, and can adapt to subsequent satellite-ground integrated autonomous control, realize joint task control between ground systems and satellites, autonomously generate requirements, autonomous task planning, dynamic situation perception, and rapid task response.

[0073] According to the method of the first aspect of the present invention, in step S2, a measurement and control cycle is divided into multiple task planning time domains, a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan, a single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and no change is allowed after acceptance, after the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated, and after reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection, and the task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

[0074] Specifically, an intelligent planning and scheduling kernel is designed.

[0075] The demand for aerospace remote sensing has shown explosive growth, and the timeliness, accuracy and requirements, large-scale rapid response dynamic task scheduling, require minimizing the disturbance of immediate remote sensing needs and complex and changeable battlefield environment to task planning schemes, and put forward higher requirements for the response speed and flexibility of the system. At present, for the problem of satellite remote sensing task scheduling, the "batch task periodic planning + emergency task adjustment" mode is adopted, satellite scheduling is abstracted as a mathematical optimization problem, and intelligent optimization algorithms or heuristic algorithms are used to solve these optimization problems. The batch processing method is used to schedule the tasks collected in a scheduling cycle (in a batch processing method, tasks are collected first and scheduled at the end of the scheduling cycle). It is difficult for tasks to get immediate response. For newly arrived important tasks, the emergency task adjustment is carried out by scheduling the nearest measurement and control resources and clearing the on-board instructions, which often loses a large number of already scheduled tasks, and the process is cumbersome, consuming more observation, measurement and control and data transmission resources, and it is difficult to adapt to tactical applications. The present invention adopts event-driven short-cycle rolling planning. Based on the short-cycle rolling planning period, it avoids the problem of long-cycle with multiple emergency adjustments and inflexible resource use. Through the design of important event-driven mechanism, it dispatches the nearest measurement and control resources in real time to respond to important remote sensing needs autonomously, effectively improving the response timeliness. Scheduling mode such as Figure 2 shown.

[0076] The design rules of the multi-satellite dynamic task scheduling mechanism are as follows:

[0077] (1) Accept the remote sensing needs that arrive in sequence. Remote sensing mainly comes from the remote sensing needs generated by the system's autonomous mining or the remote sensing needs proposed by external user units, and carry out multi-source demand planning through target access deduction, priority planning and demand analysis;

[0078] (2) Based on the historical operation status, the system formulates a short-term measurement and control plan for each satellite, responds immediately to the needs that arrive in sequence, and when the short-term mission planning arrives or an important event is triggered, multi-satellite joint mission planning and scheduling are carried out, taking into account the load conditions, mission execution conditions, expected benefits, storage conditions, energy conditions, etc. of each satellite. Based on the comprehensive deep neural network algorithm, the system realizes real-time scheduling of large-scale remote sensing tasks, accurately assigns remote sensing tasks to specific satellite task queues, and achieves short-term rapid response.

[0079] (3) Each satellite maintains a task planning queue according to the measurement and control cycle. A measurement and control cycle can be divided into multiple task planning time domains. A round of multi-satellite task planning is triggered according to a short cycle. The single satellite accepts the task planning scheme corresponding to this round of task planning cycle assigned by the scheduler, and is not allowed to change after acceptance. When the next round of multi-satellite task planning is triggered, the task planning scheme for the subsequent task planning time period is generated. When the single-satellite measurement and control time point is reached, the several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction. That is, the task planning scheme within a measurement and control cycle of a single satellite can be divided into multiple batches and multiple task planning time periods. The specific scheduling sequence and mechanism design are as follows: Figure 2 shown.

[0080] This mechanism is designed to realize event-driven rolling planning. When the tracking and control point is not reached and the multi-satellite joint task planning is not triggered, the system maintains a dynamic task pool. When the multi-satellite joint task planning is triggered, the remote sensing tasks before the tracking and control point are planned in a short period of time based on the task priority, timeliness, expected benefits, task conflicts and other factors. The task requirements that cannot be arranged in a short period of time will be returned to the task pool, and the priority, timeliness and other parameters will be adjusted to improve the order level of the task in the task queue. It can effectively avoid the problems of long emergency adjustment time and large resource loss brought by the traditional mode. Based on this single-satellite scheduling kernel design, as long as the task is issued within a certain time before the tracking and control window, the task can be quickly inserted into the planning, so that the emergency task can be converted into a regular task, which greatly improves the rapid response capability of the control system to emergency tasks.

[0081] A dynamic task list is built and maintained in the mission planning system. When external requirements (tasks) are input, task planning and resource matching are performed according to task priority (importance), and the tasks are arranged in order of their planned execution time (the list is constantly updated). Multi-satellite joint mission scheduling is triggered through two driving modes:

[0082] (1) Short-cycle driven rolling planning mode. Different from the current method of allocating 2 or 3 relatively fixed-time measurement and control resources to each satellite, the system dynamically allocates sufficient measurement and control resources to each satellite based on historical conditions and task requirements. Each measurement and control resource is several hours apart and can be dynamically scheduled according to task requirements to ensure short-cycle dynamic planning. When a certain satellite measurement and control point is reached, all required tasks that arrive dynamically before the measurement and control point are summarized and coordinated, and multi-satellite joint task scheduling is carried out. Task scheduling is assigned to each satellite task execution queue. The task planning period of the satellite is the unplanned time period within the measurement and control cycle (from the measurement and control point to the next measurement and control point). The planning period of other satellites is the unplanned time period within the measurement and control cycle. As time goes by, the measurement and control plans of all satellites in the entire system are retrieved. When the next earliest measurement and control point is reached, the next round of multi-satellite joint task planning and scheduling is triggered. The rolling task planning cycle is determined to be the same as the previous round. Several satellite tracking and control points that are close in time can be combined as a trigger condition to avoid frequent triggering of multi-satellite task scheduling, which will result in too short a task planning period, making it difficult to execute rolling cycles and strategies, and ensuring task efficiency and resource utilization. A tracking and control cycle can be divided into multiple task planning time domains. A round of multi-satellite task planning is triggered according to a short cycle. When a single-satellite tracking and control time point is reached, several rounds of task planning schemes between the previous tracking and control point and this tracking and control point are combined to generate an instruction. That is, the task planning scheme within a single-satellite tracking and control cycle can be completed in multiple batches and multiple task planning periods.

[0083] (2) Important event-driven rolling planning mode. On the basis of short-cycle driven rolling planning, when an important event occurs and remote sensing needs to be arranged immediately, and it is deduced that there is an observation opportunity for the target in the planned time period, and the measurement and control resources can be dynamically scheduled to implement the above, the nearest measurement and control point of each satellite is dynamically scheduled, and the measurement and control cycle of a satellite closest to the measurement and control point is used as the planning period. All newly arrived remote sensing tasks and planned but unfinished remote sensing tasks before the measurement and control point are summarized and coordinated, and multi-satellite joint task planning and scheduling are carried out. After the planning is completed, the planning schemes previously completed by all satellites are replaced, and the short-cycle driven rolling planning mode will be restored later.

[0084] Comprehensively consider the measurement and control resource conditions required for the recent tasks in the dynamic task list, and generate and inject instructions at an appropriate time (1 or 2 measurement and control circles) in advance (the kernel repeats and iterates continuously). The scheduling kernel is event-driven, that is, it starts to execute planning when an event that changes the system state occurs or when it is intervened manually, such as when a new task arrives, the satellite state changes, or the decision-making department proposes a planning requirement. When the number of new tasks is large, frequent task planning may be triggered, and the task planning and scheduling system will execute planning continuously, greatly increasing the time complexity of task planning. It brings hidden dangers to the consistency and security stability of the database system. In addition, the time interval between two adjacent measurement and control is long, generally at the hour level. It is only necessary to formulate a planning plan within a certain time range before the measurement and control opportunity (such as 15 minutes). There is no need to start task planning every time a task is received.

[0085] It can be seen that for event-driven, it does not mean that every new task can trigger task planning. It is only necessary to set a suitable planning back wall time, that is, the measurement and control window needs to reserve a period of time for task planning, plan compilation and instruction injection. Tasks arriving during this period cannot be scheduled in real time, but can only be accumulated in batches until the next scheduling cycle begins. Different from the significant feature of real-time scheduling, it can only respond in near real time. This period is called the frozen interval. The planning scheme within this interval is not affected by the newly added tasks. Task planning is performed according to the newly added tasks after the interval.

[0086] According to the method of the first aspect of the present invention, in the step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

[0087] Specifically, in terms of operation process design, the traditional fixed time period (24 hours in advance) for issuing instructions is changed to realize dynamic generation and issuance of instructions based on measurement and control resource constraints. The system autonomously mines and generates remote sensing requirements based on network open source information, historical task information, system real-time status and multi-source guidance information, or accepts user remote sensing requirements, and the requirements arrive in order; builds and maintains a dynamic task queue for existing remote sensing requirements. When the event timeliness threshold or measurement and control window threshold is reached, rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete parameter setting. Planning forms multiple sets of planning plans, and at the same time, the execution of planning plans is estimated to provide commanders with auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability. After the user or the on-board business agency adjusts the decision, a remote sensing plan is formed. The task queue is dynamically updated over time and driven by events. Comprehensively consider the measurement and control resource conditions required for the most recent task in the dynamic task list, dynamically schedule the most recent measurement and control resources, implement command generation and injection, change the traditional fixed time cycle injection command, and realize dynamic injection of commands based on task triggers, effectively avoiding the problems brought by the traditional model, such as long emergency adjustment time, large resource loss, and high risk of on-board plan adjustment. Monitor the task execution and system status throughout the process, and conduct system failure prediction based on historical data analysis to improve the reliability of satellite operation. Figure 3 shown.

[0088] The second aspect of the present invention provides a multi-platform collaborative perception intelligent planning system. The system includes a processing unit configured to execute:

[0089] Step S1: Establishing a dynamic intelligent task planning model; specifically including:

[0090] Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data;

[0091] Step S2: Establishing an intelligent planning and scheduling kernel; specifically including:

[0092] The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan;

[0093] Step S3: Establishing a dynamic intelligent task planning process; specifically including:

[0094] Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection non-periodically, monitor task execution throughout the entire process and issue system fault warnings.

[0095] According to the system of the second aspect of the present invention, in step S1:

[0096] Conduct remote sensing demand mining based on multi-source guidance; specifically include: mining network open source information to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and mining observation opportunities, observation needs and observation tasks in combination with real-time remote sensing data feedback, thereby generating the remote sensing demand;

[0097] Dynamically maintain the remote sensing demand queue; specifically, including: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution;

[0098] The task planning based on the remote sensing demand is triggered by an event-driven method; specifically, the task planning is started after the event-driven trigger reaches a specified threshold, the demand set in the current demand queue is used as the task source, an optimization strategy is selected to perform multi-satellite joint intelligent task planning, and the nearest measurement and control window after the task planning is triggered is scheduled for instruction injection;

[0099] The optimization strategy of the task planning is adjusted according to the quantitative evaluation result of the task planning; specifically, each time the task planning is rolled out, the task planning scheme is quantitatively evaluated to determine the reliability and execution cost of the task planning scheme to update the optimization strategy.

[0100] Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically, the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing requirements are met;

[0101] Executing the observation task according to the task plan and feeding back the observation result data; specifically comprising: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task according to the observation data and the situation awareness information;

[0102] Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

[0103] According to the system of the second aspect of the present invention, in step S2, a measurement and control cycle is divided into multiple task planning time domains, a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan, a single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and is not allowed to change after acceptance, after the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated, and after reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection, and the task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

[0104] According to the system of the second aspect of the present invention, in the step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

[0105] The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in the multi-platform collaborative perception intelligent planning method described in the first aspect of the present disclosure are implemented.

[0106] Figure 4 is a structural diagram of an electronic device according to an embodiment of the present invention, such as Figure 4 As shown, the electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, near field communication (NFC) or other technologies. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covered on the display screen, or a button, a trackball or a touch pad set on the housing of the electronic device, or an external keyboard, touch pad or mouse, etc.

[0107] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a structural diagram of the part related to the technical solution of the present disclosure, and does not constitute a limitation on the electronic device to which the technical solution of the present application is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0108] The fourth aspect of the present invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the multi-platform collaborative perception intelligent planning method described in the first aspect of the present disclosure are implemented.

[0109] In summary, the architecture adopted by the technical solution of the present invention is relatively flexible and easy to expand; it can realize autonomous demand generation, rapid task response and intelligent task planning, and support heterogeneous platform resource expansion; compared with the traditional model, it has the advantages of flexible emergency adjustment and low resource loss, and has good application prospects.

[0110] Please note that the technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. The above-mentioned embodiments only express several implementation methods of the present application, and their descriptions are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present application, several variations and improvements can be made, which all belong to the scope of protection of the present application. Therefore, the scope of protection of the patent in this application shall be based on the attached claims.

Claims

1. A multi-platform collaborative perception intelligent planning method, characterized in that: The method comprises: Step S1: Establishing a dynamic intelligent task planning model; specifically including: Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data; Step S2: Establishing an intelligent planning and scheduling kernel; specifically including: The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan; Step S3: Establishing a dynamic intelligent task planning process; specifically including: Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection aperiodically, monitor task execution throughout the entire process, and issue system failure warnings; Wherein, in step S1: Conduct remote sensing demand mining based on multi-source guidance; specifically include: mining network open source information to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and mining observation opportunities, observation needs and observation tasks in combination with real-time remote sensing data feedback, thereby generating the remote sensing demand; Dynamically maintain the remote sensing demand queue; specifically, including: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution; The task planning based on the remote sensing demand is triggered by an event-driven method; specifically, the task planning is started after the event-driven trigger reaches a specified threshold, the demand set in the current demand queue is used as the task source, an optimization strategy is selected to perform multi-satellite joint intelligent task planning, and the nearest measurement and control window after the task planning is triggered is scheduled for instruction injection; Adjusting the optimization strategy of the task planning according to the quantitative evaluation result of the task planning; specifically comprising: performing a quantitative evaluation on the task planning scheme each time the task planning is rolled out, determining the reliability and execution cost of the task planning scheme, and updating the optimization strategy; Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically, the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing requirements are met; Executing the observation task according to the task plan and feeding back the observation result data; specifically comprising: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task according to the observation data and the situation awareness information; Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

2. According to claim 1, a multi-platform collaborative perception intelligent planning method is characterized in that: In step S2, a measurement and control cycle is divided into multiple task planning time domains, and a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan. A single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and no changes are allowed after acceptance. When the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated. After reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection. The task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

3. The multi-platform collaborative perception intelligent planning method according to claim 2 is characterized in that: In step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

4. A multi-platform collaborative perception intelligent planning system, characterized in that: The system comprises a processing unit configured to perform: Step S1: Establishing a dynamic intelligent task planning model; specifically including: Conduct remote sensing demand mining based on multi-source guidance; dynamically maintain remote sensing demand queues; trigger task planning based on the remote sensing demand with event-driven; adjust the optimization strategy of the task planning according to the quantitative evaluation results of the task planning; configure remote sensing information acquisition terminals to obtain multi-party remote sensing information in real time; execute observation tasks according to the task planning and feedback observation result data; autonomously control each terminal / platform based on the observation result data; Step S2: Establishing an intelligent planning and scheduling kernel; specifically including: The intelligent planning and scheduling kernel sequentially accepts the continuously arriving remote sensing demands, determines a short-cycle measurement and control plan for each satellite according to the historical mission execution data, and each satellite maintains a queue of mission plans to be executed according to the short-cycle measurement and control plan; Step S3: Establishing a dynamic intelligent task planning process; specifically including: Dynamically schedule measurement and control resources based on measurement and control resource constraints, implement dynamic instruction generation and injection aperiodically, monitor task execution throughout the entire process, and issue system failure warnings; Wherein, in step S1: Conduct remote sensing demand mining based on multi-source guidance; specifically include: mining network open source information to extract hot events, mining the information download volume and usage of the hot events to analyze user preferences, and mining observation opportunities, observation needs and observation tasks in combination with real-time remote sensing data feedback, thereby generating the remote sensing demand; Dynamically maintain the remote sensing demand queue; specifically, including: maintaining the remote sensing demand queue, and generating the next remote sensing demand through event evolution prediction, dynamically adjusting the optimization level of the remote sensing demand in the remote sensing demand queue according to situation changes, and deleting the remote sensing demand from the remote sensing demand queue after triggering execution; The task planning based on the remote sensing demand is triggered by an event-driven method; specifically, the task planning is started after the event-driven trigger reaches a specified threshold, the demand set in the current demand queue is used as the task source, an optimization strategy is selected to perform multi-satellite joint intelligent task planning, and the nearest measurement and control window after the task planning is triggered is scheduled for instruction injection; Adjusting the optimization strategy of the task planning according to the quantitative evaluation result of the task planning; specifically comprising: performing a quantitative evaluation on the task planning scheme each time the task planning is rolled out, determining the reliability and execution cost of the task planning scheme, and updating the optimization strategy; Configuring a remote sensing information acquisition terminal to acquire multi-party remote sensing information in real time; specifically, the remote sensing information acquisition terminal acquires aerospace remote sensing information, on-board processing information, and situation awareness information in real time to determine whether the remote sensing requirements are met; Executing the observation task according to the task plan and feeding back the observation result data; specifically comprising: the remote sensing satellite implements the observation task according to the task plan, transmits the observation data to the ground control center, and the control center determines whether to execute the next observation task according to the observation data and the situation awareness information; Autonomous control of each terminal / platform is performed based on the observation result data; specifically, the observation result data obtained based on event-driven is used for subsequent autonomous control of satellite-ground integration.

5. The multi-platform collaborative perception intelligent planning system according to claim 4 is characterized in that: In step S2, a measurement and control cycle is divided into multiple task planning time domains, and a round of multi-satellite task planning is triggered according to the short-cycle measurement and control plan. A single satellite accepts the task planning scheme corresponding to the task planning cycle assigned by the scheduling, and no changes are allowed after acceptance. When the next round of multi-satellite task planning is triggered, a task planning scheme for the subsequent task planning time period is generated. After reaching the single-satellite measurement and control time point, several rounds of task planning schemes between the previous measurement and control point and this measurement and control point are merged to generate an instruction injection. The task planning scheme within a measurement and control cycle of a single satellite is completed in multiple batches and multiple task planning time periods.

6. The multi-platform collaborative perception intelligent planning system according to claim 5, characterized in that: In step S3, the remote sensing demand is generated based on network open source information, historical task information, system real-time status and multi-source guidance information, and the remote sensing demand arrives in sequence; a queue of the task planning to be executed is constructed and maintained, and when the event timeliness threshold or the measurement and control window threshold is reached, the rolling intelligent planning is triggered, and the imaging parameter optimization module is called to complete the parameter setting, and multiple sets of planning plans are planned. At the same time, the execution estimation of the planning scheme is carried out, and auxiliary decision-making information such as task completion timeliness, meteorological conditions, task loss and reliability is provided. After the decision is adjusted, a remote sensing plan is formed, and the queue of the task planning to be executed is dynamically updated over time and event-driven.

7. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps in the multi-platform collaborative perception intelligent planning method described in any one of claims 1-3 are implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps in the multi-platform collaborative perception intelligent planning method described in any one of claims 1 to 3 are implemented.

Images