CN-121998384-A - Satellite task planning method and system based on multiple constraint conditions

CN121998384ACN 121998384 ACN121998384 ACN 121998384ACN-121998384-A

Abstract

The application discloses a satellite task planning method and system based on multiple constraint conditions, and relates to the technical field of reinforcement learning, wherein the method comprises the steps of receiving task parameters of a plurality of observation tasks, and performing multiple constraint analysis on each observation task based on current satellite orbit data and the task parameters to obtain at least one candidate observation scheme corresponding to each task, wherein the multiple constraint analysis at least comprises orbit visibility analysis and meteorological constraint analysis; and generating a collaborative planning scheme integrating a plurality of constraint conditions through a collaborative decision model based on candidate observation schemes of all observation tasks, wherein the collaborative decision model takes the maximized overall task return as an objective function, the objective function fuses quality scores of all tasks and resource conflict penalty items among the tasks, and a task instruction sequence for driving the satellite to execute is generated based on the collaborative planning scheme. The method and the device remarkably improve the accuracy and the self-adaptive capacity of satellite task planning in a complex scene.

Inventors

CAI YULIN
WU QIANYU
WANG MI
HONG YONG
ZHANG TINGTING
GUO PEIPEI

Assignees

湖北珞珈实验室
武汉大学

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. The satellite mission planning method based on the multiple constraint conditions is characterized by comprising the following steps: Receiving task parameters of a plurality of observation tasks, wherein the task parameters at least comprise a target area of each task, an expected execution time window, a load type and a priority; Performing multi-constraint analysis on each observation task based on current satellite orbit data and task parameters to obtain at least one candidate observation scheme corresponding to each task, wherein the multi-constraint analysis at least comprises orbit visibility analysis and meteorological constraint analysis; Generating a collaborative planning scheme integrating a plurality of constraint conditions through a collaborative decision model based on the candidate observation schemes of all observation tasks, wherein the collaborative decision model takes the maximized overall task return as an objective function, and the objective function fuses quality scores of all tasks and resource conflict penalty items among the tasks; based on the collaborative planning scheme, a sequence of mission instructions for driving the satellite to execute is generated.
2. The method of claim 1, wherein the step of performing multi-constraint analysis on the basis of the current satellite orbit data and the task parameters for each of the observation tasks to obtain at least one candidate observation scheme corresponding to each task comprises: Carrying out orbit visibility analysis on each observation task based on current satellite orbit data and a target area corresponding to the observation task, calculating a transit time window of a satellite to the target area, and judging an observation mode; Acquiring real-time meteorological data which are matched with the transit time window in time and space according to a multi-source meteorological data acquisition mechanism; according to cloud sensitivity rules corresponding to the load types, cloud coverage in the real-time meteorological data is processed to obtain meteorological constraint scores; And generating candidate observation schemes based on the observation mode, the transit time window and the weather constraint score.
3. The method of claim 2, wherein the steps of performing an orbit visibility analysis for each observation task based on current satellite orbit data and a target area corresponding to the observation task, calculating an transit time window of the satellite to the target area, and discriminating an observation mode include: Inputting two lines of orbit data into a simplified perturbation model, and calculating a transit time window of a satellite to the target area; in the transit time window, calculating the transverse distance between a target point and a satellite lower point track, wherein the target point is the center point of the target area; And obtaining an observation mode according to the relation among the transverse distance, the half-width of the satellite load and the side sway coverage range, wherein the observation mode comprises a normal push-broom mode, a side sway observation mode and an unobservable mode.
4. The method of claim 3, wherein the obtaining an observation mode based on the relationship between the lateral distance, the half-width of the satellite load, and the roll coverage, the observation mode including a normal push broom mode, a roll observation mode, and a non-observable step further comprises: If the observation mode is a side swing observation mode, calculating the altitude angle at each moment in the transit period based on the simplified perturbation model, and forming altitude angle time sequence data; Screening all moments when the height angle is larger than or equal to the minimum observed height angle threshold value from the height angle time sequence data; combining the continuous time periods in the screened moments to obtain at least one available time period; for each available period, establishing a local coordinate system with a satellite as an origin, and calculating an observation vector pointing from the satellite to the target point, wherein an X-axis of the local coordinate system is along a speed direction, a Y-axis is perpendicular to an orbit plane, and a Z-axis is pointing to the earth center; and calculating a roll angle through the included angle between the observation vector and the Z axis, and determining the roll direction according to the projection of the observation vector on the Y axis.
5. The method of claim 2, wherein the step of acquiring real-time weather data that matches the transit time window space-time according to a multi-source weather data acquisition mechanism comprises: acquiring gridded cloud cover data from at least one external meteorological data source according to a multi-source meteorological data acquisition mechanism, wherein the external meteorological data source comprises a main data source and a standby data source, the multi-source meteorological data acquisition mechanism comprises switching to the standby data source if the main data source fails to acquire, and switching to the latest cache data if the main data source and the standby data source fail to acquire; And matching the gridded cloud cover coverage data through a space-time interpolation algorithm based on the specific moment in the transit time window and the specific position of the target area to obtain the cloud cover coverage matched with the transit time window in a space-time mode, wherein the space-time interpolation algorithm comprises linear interpolation of a time dimension and bilinear interpolation of a space dimension.
6. The method of claim 2, wherein after the step of generating candidate observation plans based on the observation mode, the transit time window, and the weather constraint score, further comprising: When a single observation task corresponds to a plurality of candidate observation schemes, calculating the quality scores of the candidate observation schemes through a multi-factor weighting scoring model based on the observation modes, the altitude angles, the lateral sway angles and the weather constraint scores corresponding to the candidate observation schemes; sorting the plurality of candidate observation schemes according to the quality scores to obtain sorting results; And screening a preset number of schemes from the plurality of candidate observation schemes according to the sequencing result to obtain a candidate scheme set corresponding to a single observation task.
7. The method of claim 6, wherein the collaborative planning scheme includes a plurality of target observation schemes, the collaborative planning scheme integrating a plurality of constraints is generated by a collaborative decision model based on the candidate observation schemes of all observation tasks, the collaborative decision model uses a maximization overall task return as an objective function, and the objective function merges quality scores of tasks and resource conflict penalty items among tasks, the method comprises the steps of: Constructing a state space based on current satellite state information, the candidate scheme set of each task and task parameters of each task, wherein the state space is also used for carrying out time sequence prediction through a long-short-term memory network to obtain future environment information, and modeling constraint relations among the tasks through a graph neural network to obtain task relation information; Inputting the state space into a deep reinforcement learning decision model, outputting an initial scheduling scheme, and maximizing the overall task return through an optimization strategy, wherein the initial scheduling scheme comprises a plurality of intermediate observation schemes, the deep reinforcement learning decision model is an Actor-Critic architecture, an Actor network outputs actions, and the Critic network evaluates state values, and the actions are used for distributing the intermediate observation schemes for each observation task; and performing multi-objective optimization processing based on the initial scheduling scheme to generate a pareto optimal solution set, wherein the pareto optimal solution set comprises a plurality of objective observation schemes.
8. A satellite mission planning system based on multiple constraint conditions is characterized in that, the satellite mission planning system based on the multiple constraint conditions comprises: The task configuration module is used for receiving task parameters of a plurality of observation tasks, wherein the task parameters at least comprise target areas of the tasks, expected execution time windows, load types and priorities; The multi-constraint analysis module is used for carrying out multi-constraint analysis on each observation task based on current satellite orbit data and task parameters to obtain at least one candidate observation scheme corresponding to each task, and the multi-constraint analysis at least comprises orbit visibility analysis and meteorological constraint analysis; The scheduling decision module is used for generating a collaborative planning scheme integrating a plurality of constraint conditions through a collaborative decision model based on the candidate observation schemes of all the observation tasks, the collaborative decision model takes the maximized whole task return as an objective function, and the objective function fuses the quality scores of all the tasks and the resource conflict penalty items among the tasks; And the result generation module is used for generating a task instruction sequence for driving the satellite to execute.
9. A multi-constraint based satellite mission planning system as claimed in claim 8, wherein said multi-constraint analysis module further comprises a visibility analysis unit and a meteorological data integration unit; the visibility analysis unit is used for carrying out orbit visibility analysis on each observation task based on current satellite orbit data and a target area corresponding to the observation task; The meteorological data integration unit is used for acquiring real-time meteorological data matched with the transit time window in time and space according to a multi-source meteorological data acquisition mechanism.
10. A multi-constraint based satellite mission planning system as claimed in claim 8, further comprising a visualization module for three-dimensional visual presentation and verification of the collaborative planning scheme, the visualization module comprising: The load simulation unit is used for dynamically calculating and generating a push-broom strip model fitting the curved surface of the earth based on satellite load parameters, real-time orbit data and observation geometric parameters in a collaborative planning scheme; The synchronous control unit is used for keeping the simulation display of the push-broom strip model and the planned observation period in time synchronization and keeping the posture of the observation view cone and the planned side swing angle in synchronization; and the interactive demonstration unit is used for displaying the push-broom strip model and the observation view cone in real time in the three-dimensional scene and providing an interactive interface for a user to carry out multi-scheme contrast verification.

Description

Satellite task planning method and system based on multiple constraint conditions Technical Field The application relates to the technical field of reinforcement learning, in particular to a satellite mission planning method and system based on multiple constraint conditions. Background Along with the rapid development of the aerospace technology, the number of in-orbit satellites is continuously increased, and the in-orbit satellites have the characteristics of high scale, high frequency and timeliness requirements on earth observation tasks. The core of the satellite mission planning is to efficiently and reasonably arrange satellite resources to execute observation missions on the premise of meeting various physical constraints. In recent years, research begins to explore application of artificial intelligence technologies such as deep learning to task planning, for example, resource allocation is performed by using deep reinforcement learning, but traditional satellite task planning cannot be suitable for high-timeliness scenes such as multi-task collaborative planning, emergency response, ultra-large-scale task scheduling and the like, and the accuracy of task planning is insufficient, so that the practicability and decision efficiency of a system are seriously affected. Therefore, in a complex scenario, how to improve the adaptability and accuracy of satellite mission planning is a problem that needs to be solved at present. Disclosure of Invention The application mainly aims to provide a satellite task planning method and system based on multiple constraint conditions, and aims to solve the technical problem of how to improve the adaptability and accuracy of satellite task planning in a complex scene. In order to achieve the above object, the present application provides a satellite mission planning method based on multiple constraint conditions, the method comprising: Receiving task parameters of a plurality of observation tasks, wherein the task parameters at least comprise a target area of each task, an expected execution time window, a load type and a priority; Performing multi-constraint analysis on each observation task based on current satellite orbit data and task parameters to obtain at least one candidate observation scheme corresponding to each task, wherein the multi-constraint analysis at least comprises orbit visibility analysis and meteorological constraint analysis; Generating a collaborative planning scheme integrating a plurality of constraint conditions through a collaborative decision model based on the candidate observation schemes of all observation tasks, wherein the collaborative decision model takes the maximized overall task return as an objective function, and the objective function fuses quality scores of all tasks and resource conflict penalty items among the tasks; based on the collaborative planning scheme, a sequence of mission instructions for driving the satellite to execute is generated. In an embodiment, the step of performing multi-constraint analysis on each observation task based on current satellite orbit data and the task parameters to obtain at least one candidate observation scheme corresponding to each task includes: Carrying out orbit visibility analysis on each observation task based on current satellite orbit data and a target area corresponding to the observation task, calculating a transit time window of a satellite to the target area, and judging an observation mode; Acquiring real-time meteorological data which are matched with the transit time window in time and space according to a multi-source meteorological data acquisition mechanism; according to cloud sensitivity rules corresponding to the load types, cloud coverage in the real-time meteorological data is processed to obtain meteorological constraint scores; And generating candidate observation schemes based on the observation mode, the transit time window and the weather constraint score. In an embodiment, the step of performing, for each observation task, an orbit visibility analysis based on current satellite orbit data and a target area corresponding to the observation task, calculating an transit time window of the satellite to the target area, and determining an observation mode includes: Inputting two lines of orbit data into a simplified perturbation model, and calculating a transit time window of a satellite to the target area; in the transit time window, calculating the transverse distance between a target point and a satellite lower point track, wherein the target point is the center point of the target area; And obtaining an observation mode according to the relation among the transverse distance, the half-width of the satellite load and the side sway coverage range, wherein the observation mode comprises a normal push-broom mode, a side sway observation mode and an unobservable mode. In an embodiment, the obtaining an observation mode according to the relation among the lateral dis