Satellite group cooperative tracking observation interactive virtual simulation verification system

Abstract

The invention belongs to the technical field of constellation-level collaborative observation, virtual simulation verification and real-time simulation interaction, and particularly relates to a star group collaborative tracking observation interactive virtual simulation verification system which comprises a platform end used for providing an operating environment and computing power support for man-machine interaction software and a three-dimensional virtual scene; the platform end comprises a main control unit and a visual scene unit; the main control unit is mainly responsible for flow control and man-machine interaction of overall demonstration verification, sends a control instruction through network communication, collects experimental data and realizes data management based on a database; the visual scene unit constructs a plurality of different virtual scenes corresponding to the overall task process based on a Unity engine, provides real-time environment parameter information for the intelligent planning algorithm of the algorithm end, and finally verifies the effectiveness of the intelligent planning algorithm under the real-time environment parameter information. And three-dimensional visual scene demonstration and verification of in-orbit operation of the intelligent satellite group are realized.

Description

Technical Field

[0001] This invention belongs to the field of constellation-level collaborative observation, virtual simulation verification, and real-time simulation interaction technology, specifically involving an interactive virtual simulation verification system for constellation collaborative tracking observation. Background Technology

[0002] Modern small satellites not only possess advantages such as small size, light weight, and low development cost, but also have short development cycles and can continuously iterate and apply new technologies, thus having broad development and application prospects. On the other hand, as the application scope of satellites in communication, navigation, reconnaissance, and other fields continues to expand, single satellites cannot meet the increasingly complex mission requirements, and the collaborative work of multiple satellites forming formations is gradually becoming the mainstream direction of satellite technology development. To fully utilize space resources, space-distributed systems composed of multiple satellites are rapidly developing, and inter-satellite space collaboration technology also needs to adapt to more complex and diverse missions. For example, mission planning for multiple observation targets moving at high speeds in space, through multi-angle collaborative observation by small satellite formations, can improve constellation detection capabilities and space utilization, meeting future practical needs for reconnaissance and detection. Due to the ever-expanding demands of satellite space missions, the collaborative completion of complex mission objectives by multiple satellites is gradually becoming the mainstream direction of current space technology development. Furthermore, with the continuous increase in constellation size, mission objectives also exhibit diversity and complexity. Taking Starlink as an example, after deployment, the number of Starlink satellites per geocentric unit solid angle will reach 12,000. The problem of satellite constellation mission decision-making urgently needs to be solved. Faced with such large-scale and highly flexible constellations, traditional enemy satellite monitoring technologies are unable to quickly and comprehensively observe enemy anomalies. Unlike traditional control methods, the adaptive coalition formation algorithm optimizes the observation effects of multiple key satellites in a large-scale enemy constellation by constructing a cooperative coalition. It plans collaborative missions for GF constellations through adaptive coalition formation. The location distribution and motion state of the observed targets are characterized by randomness and uncertainty; therefore, the decision-making problem for constellation collaborative observation missions is a complex problem with multiple dimensions including time, space, and location, and possesses high research value.

[0003] Due to the high development costs and launch risks associated with satellites, satellite simulation technology has played a crucial role in the development of satellite technology. Traditional digital simulation technology typically uses simulation modules provided by MATLAB / Simulink to simulate and verify the entire physical model and diagnostic techniques. However, this approach has drawbacks: the simulation environment is relatively idealized, and purely digital simulation results are difficult to integrate with real-world application scenarios, thus limiting the simulation conclusions. With the gradual maturation of virtual simulation technology, the integration of satellite simulation technology with virtual simulation technology has become an inevitable trend. Traditional satellite simulation technology struggles to meet the demands for real-time performance, intuitiveness, and interactivity in satellite simulations, leading to the emergence of satellite visualization simulation technology. On one hand, based on the satellite's actual physical appearance and mission requirements, a 3D physical model of the satellite is designed using modeling software. A physics engine simulates the satellite's actual motion in space, creating a visualized virtual simulation scene to recreate the real-world satellite mission execution, improving the readability, intuitiveness, and persuasiveness of simulation conclusions. On the other hand, a human-computer interaction interface is designed to ensure the real-time performance and efficiency of satellite visualization simulation technology, allowing users to directly manage and control the satellite simulation process, thus improving the versatility and convenience of the simulation system.

[0004] In summary, this invention addresses the monitoring and verification needs of constellation satellites in complex and dynamic space environments by constructing an interactive virtual simulation verification system for collaborative tracking and observation of satellite clusters. This system enables three-dimensional visualization and verification of intelligent satellite clusters in orbit, demonstrating innovation and practical application value. Considering the cost of virtual simulation platforms, this invention adopts a dual-machine mode of a main control computer and a visual computer, which is cost-effective and provides real-time performance. The main control computer uses an intelligent planning algorithm written in Python to perform real-time satellite cluster task planning and sends the planning results to the visual computer. The visual computer uses Unity3D to develop an interactive virtual simulation verification platform, embedding human-computer interaction functions to visualize the satellite cluster's operational status, abnormal orbit changes, and collaborative tracking observations. This allows researchers to more quickly and accurately verify the feasibility and effectiveness of task planning.

[0005] The purpose of this invention is to provide an interactive virtual simulation verification system for constellation collaborative tracking and observation, integrating real-time simulation and visual display functions to verify the effectiveness of intelligent constellation mission planning algorithms. Considering the limitations of real-world flight verification of satellite constellations in engineering projects, this invention overcomes the shortcomings of traditional offline simulation in real-time verification. Combining the needs of satellite constellation technology verification, it employs an adaptive alliance formation algorithm based on Python for constellation observation mission planning. A real-time interactive virtual simulation verification platform is designed using Unity3D to build a realistic constellation model and space environment, enabling real-time data transmission with the algorithm and providing a more convenient human-computer interaction experience, thereby verifying the real-time performance and effectiveness of the designed mission planning technology. Summary of the Invention

[0006] The purpose of this invention is to provide an interactive virtual simulation verification system for satellite constellation collaborative tracking observation. An interactive virtual simulation platform was built to meet the needs of mission planning technology. The effectiveness of mission planning technology was verified in the virtual simulation environment. Based on real-time simulation data, human-computer interaction was carried out to realize a three-dimensional visualization demonstration of the satellite constellation collaborative observation process.

[0007] The specific technical solution adopted by this invention is as follows: An interactive virtual simulation verification system for stellar cluster collaborative tracking observation, including both the platform and the algorithm. The platform runs on a high-performance server, providing a runtime environment and computing power support for human-computer interaction software and 3D virtual scenes. The platform includes a main control unit and a visual unit; The main control unit is the control and communication center of the threat target collaborative tracking and observation virtual simulation platform. It is mainly responsible for the overall demonstration and verification process control and human-computer interaction, sending control commands and collecting experimental data through network communication, and implementing data management based on the database. The visual unit is the demonstration and verification center of the UAV swarm decision-making virtual simulation platform. It constructs multiple virtual scenes corresponding to the overall task flow based on the Unity engine, and at the same time provides real-time environmental parameter information for the intelligent planning algorithm of the algorithm side. The algorithm acquires real-time environmental parameter information sent by the visual unit, generates decision planning results, and feeds the decision planning results back to the visual unit. The visual unit visualizes the decision planning results and sends them to the main control unit. The main control unit stores the decision planning results and ultimately verifies the effectiveness of the intelligent planning algorithm under real-time environmental parameter information, realizing a three-dimensional visualization demonstration and verification of the intelligent constellation's on-orbit operation.

[0008] Preferably, the specific process by which the system verifies the effectiveness of the intelligent planning algorithm under real-time environmental parameter information is as follows: Step 1: The algorithm is based on an adaptive alliance dynamic programming algorithm; when a threat from an enemy satellite is detected, research on real-time tracking technology for the enemy threat target needs to be carried out to complete the mission planning for the observation satellite constellation. Step 2: The platform design implements the real-time simulation architecture of the constellation collaborative observation interactive virtual simulation verification platform; Step 3: The platform-side design implements the visual software for the constellation collaborative observation interactive virtual simulation verification platform.

[0009] Preferably, in step 1, after a threat from an enemy satellite is detected, research on real-time tracking technology for the enemy threat target needs to be carried out. First, considering multiple constraints such as unobservable blind spots, double coverage, and space payload resources, a constellation collaborative intelligent mission planning model is established with the goal of maximizing the collaborative tracking of dynamic threat targets. Second, alliance switching rules between observation satellites and conditions for the formation of stable alliance partitions in the constellation are proposed. A constellation collaborative observation mission planning solution algorithm based on the alliance formation is designed to realize mission planning for the observation satellite constellation.

[0010] Preferably, in step 2, the design and implementation of the real-time simulation architecture of the constellation collaborative observation interactive virtual simulation verification platform are carried out. This invention addresses the real-time simulation verification requirements of constellation mission planning algorithms by implementing a real-time simulation architecture based on ML-Agents, building a real-time training environment, where the Unity client receives real-time data transmitted from the Python client for mission planning, and the Python client receives satellite status information fed back from the Unity client, using constellation intelligent mission planning technology for online real-time mission allocation.

[0011] Preferably, in step 3, the visual software of the constellation collaborative observation interactive virtual simulation verification platform is designed and implemented. This invention addresses complex spacecraft mission scenarios by creating a three-dimensional interactive virtual simulation scene based on Unity3D, constructing satellite and Earth models based on 3D Max, and driving satellite operation simulation through TLE data parsing. A clear functional interactive interface is constructed using WPF and the HandyControl UI library, and a graphical interface is developed based on UGUI components, thus realizing the construction of the interactive virtual simulation verification platform. Real-time satellite mission planning results are developed and simulated through the collaborative observation simulation interface to further verify the feasibility and reliability of the intelligent planning method.

[0012] The technical effects achieved by this invention are as follows: The interactive virtual simulation verification platform for stellar collaborative tracking and observation built in this invention provides a real-time on-orbit anomaly injection function for Starlink satellites to meet the needs of stellar mission planning technology. It calculates the optimal solution based on a real-time intelligent planning algorithm and transmits it to the simulation platform for virtual demonstration, thereby verifying the effectiveness of mission planning technology in the real-time operating environment of spacecraft and has high application value.

[0013] Compared with traditional virtual simulation of star cluster tracking observation, the advantages of this invention are as follows: (1) Traditional mission planning virtual simulation does not support real-time human-computer interaction. This invention develops real-time simulation interaction function by independently designing UI interface. Experimenters can select any satellite through UI interface to view its on-orbit operation status, randomly select any satellite at any time to inject abnormal maneuver information, interact with intelligent mission planning algorithm, and display planning results in real time.

[0014] (2) Traditional mission planning virtual simulations mostly display allocation results in the form of text or charts, which are cumbersome and cannot intuitively display the results. This invention uses Unity 3D to build an interactive virtual simulation platform, which displays the satellite's operating status in three dimensions through animation effects, thus getting rid of the previous complex chart data analysis.

[0015] In summary, this invention can bring corresponding benefits to the spacecraft field. Satellite mission planning has always been a key research issue in the spacecraft field. The threat target collaborative tracking and observation virtual simulation verification platform designed in this invention has the characteristics of good real-time performance and strong engineering practicality, and has high application value, providing a solid foundation for the feasibility analysis and verification of future mission allocation algorithms. In addition, this invention can also be applied as a complete verification platform to other fields, and has considerable practical significance. Attached Figure Description

[0016] Figure 1 This is a diagram showing the overall structure of the simulation verification system of this invention; Figure 2 This is a diagram of the alliance formation algorithm structure of this invention; Figure 3 This invention relates to the OSI and TCP / IP network models; Figure 4 This is a flowchart of the client-server communication process of the present invention; Figure 5 This is a code block call logic diagram for the ML-Agents invention; Figure 6 This is a functional structure diagram of the visual software of this invention; Figure 7 This is the startup interface of the simulation platform of this invention; Figure 8This is the normal operating interface of the satellite constellation of this invention; Figure 9 This invention relates to the Starlink satellite abnormal maneuver interface; Figure 10 This is the collaborative tracking and observation interface of the present invention; Figure 11 This is a diagram showing the real-time planning results of this invention. Detailed Implementation

[0017] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0018] like Figure 1 As shown, this invention takes complex satellite constellations as the research object, proposes a constellation intelligent mission planning technology based on adaptive alliance formation, and designs an interactive virtual simulation verification platform to meet the verification needs of the planning algorithm. The overall structure diagram of the constellation cooperative tracking and observation interactive virtual simulation verification system of this invention is shown below. Figure 1 As shown, the entire virtual simulation system includes the platform side and the algorithm side; The platform runs on a high-performance server, providing a runtime environment and computing power support for human-computer interaction software and 3D virtual scenes. The platform includes a main control unit and a visual unit; The main control unit is the control and communication center of the threat target collaborative tracking and observation virtual simulation platform. It is mainly responsible for the overall demonstration and verification process control and human-computer interaction, sending control commands and collecting experimental data through network communication, and implementing data management based on the database. The visual unit is the demonstration and verification center of the UAV swarm decision-making virtual simulation platform. It constructs multiple virtual scenes corresponding to the overall task flow based on the Unity engine, and at the same time provides real-time environmental parameter information for the intelligent planning algorithm of the algorithm side. The algorithm acquires real-time environmental parameter information sent by the visual unit, generates a decision planning result, and feeds the result back to the visual unit. The visual unit visualizes the decision planning result and sends it to the main control unit. The main control unit stores the decision planning result, ultimately verifying the intelligent planning algorithm under real-time environmental parameter information. The specific process by which the system verifies the effectiveness of the intelligent planning algorithm under real-time environmental parameter information is as follows: Step 1: The algorithm is based on an adaptive alliance dynamic programming algorithm; when a threat from an enemy satellite is detected, research on real-time tracking technology for the enemy threat target needs to be carried out to complete the mission planning for the observation satellite constellation. This invention proposes an intelligent task planning technology, establishes a constellation-cooperative intelligent task planning system, and puts forward an adaptive alliance-forming dynamic programming algorithm, which has a good effect on conflict resolution and can effectively solve the task allocation problem in multi-target cooperative observation of constellations. It enables real-time relay tracking and observation of multiple dynamic enemy threat targets.

[0019] First, based on constraints such as deep space background and distance, the visibility window of our constellation to enemy satellites with unusual movements is calculated. Second, considering the visibility conditions of the space-based early warning system constellation satellites to enemy threat targets, double coverage, and mission urgency, multiple constraints are established, including unique observation, continuous observation constraints, and double coverage. A comprehensive mission benefit function is established using overall double coverage, system switching times, relative motion trends, and stereo observation angles, and a coalition formation model is built. Then, coalition switching rules between observation satellites and conditions for the formation of stable coalition partitions in the constellation are proposed. Finally, a coalition-based collaborative observation mission decision-making algorithm is designed to achieve mission decision-making for the observation satellite constellation. The specific technical solution is as follows: 1. Establishment of a constellation mission planning model based on dynamic target observation In the process of constellation mission planning, how to schedule observation satellites according to the observation mission plan and ensure the rational allocation of observation satellite resources is a key issue that needs to be addressed. The constraints involved in the constellation space dynamic multi-target collaborative observation problem include dual observation constraints, unique observation constraints, continuous observation constraints, near-edge observation angle constraints, and observation distance constraints.

[0020] Constraint Analysis Double observation constraint: To acquire position and motion information of an observation target, a satellite constellation requires at least two satellites to be allocated observation resources for that target simultaneously. (Definition) Indicates observation satellite Select observation task , Indicates the selection of observation tasks The set of observation satellites, specifically represented as:

[0021] In the formula: Indicates observation satellite Select observation task , Indicates the selection of observation tasks A collection of observation satellites.

[0022] Numerical Relationship Constraints Considering the limited observation resources of the agile satellites and their inability to execute multiple observation tasks simultaneously, each observation satellite is restricted to selecting only one observation task within a scheduling cycle, specifically as follows:

[0023] In the formula Indicates the first The number of observation tasks existing within a scheduling cycle.

[0024] Relay observation constraints Considering the need for continuous observation of the target, the start time of the observation satellite's observation of the target should be the end time of the previous observation satellite's observation of the same target, specifically expressed as follows:

[0025] In the formula: and They represent observation satellites. and observation satellites The start time of observation, Indicates observation satellite The duration of continuous observation of the target.

[0026] Visibility constraints For an observation satellite to observe a target, its line of sight must be above the edge of the Earth's atmosphere; otherwise, the target cannot be observed. Secondly, satellites have a maximum observation range, expressed as:

[0027] In the formula: It is the observation angle at the edge. This is the satellite's maximum observation distance.

[0028] Inter-satellite communication constraints Due to the influence of dual observation, satellites conducting collaborative observations need to perform inter-satellite communication to lock their positions. Let B be the number of satellites. Multiply by the number of tasks A 0, 1 communication matrix. Communication constraints are represented as follows:

[0029] definition Indicates observation satellite With observation satellites Communication status between them Objective function establishment To maximize global gains, the objective function is determined by the total number of successful observations of the target by the satellite constellation. More observations result in greater gains, but more frequent switching between observation targets increases costs. Therefore, the global objective function is defined as follows:

[0030] In the formula: This represents the total number of dynamically observed targets. Indicates the first One observation satellite, Indicates the first The observation satellite for the first The observation coverage of each observation target Indicates the overall double observation coverage. This represents the number of times the satellite constellation switches during the relay observation process of all dynamic observation targets. and These represent the two weighting coefficients.

[0031] Optimization model establishment Combining the above constraints and objective function, an optimization model for the dynamic multi-objective cooperative observation problem of star clusters can be established, specifically expressed as:

[0032] The goal of this optimization model is to assign appropriate observation tasks to each observation satellite while satisfying mission requirements, numerical relationships, relay observation constraints, visibility constraints, and inter-satellite communication constraints, thereby maximizing the sum of the benefits of all observation combinations.

[0033] Design of adaptive consortium-based collaborative observation algorithm This invention proposes an adaptive coalition formation mission planning technique based on an intelligent mission planning model. A coalition formation model is established based on the task allocation problem of multi-target collaborative observation of satellite constellations; coalition switching rules between observation satellites and conditions for stable coalition partitioning of satellite constellations are proposed; and a task planning algorithm based on coalition formation is designed to realize task planning for observation satellite constellations. With the continuous development of space-distributed systems composed of multiple satellites, the mission execution capabilities of a single satellite are insufficient to meet the increasingly complex space mission requirements. Therefore, multi-satellite collaboration is needed to better complete missions. The process of satellite networking and collaborative cooperation is similar to the coalition formation process; therefore, the coalition formation concept can be used to solve the task allocation problem of multi-target collaborative observation of satellite constellations.

[0034] Adaptive Alliance Formation Model To solve the problem of task allocation for multi-target collaborative observation of star clusters using the concept of coalition formation, it is first necessary to establish an adaptive star cluster coalition formation model. The coalition formation model mainly includes two elements: the set of participants and the overall coalition benefit, defined as follows: Participant set

[0035] All participants in the collaborative observation mission As participants in the formation of a coalition, each observation satellite represents a set of all observation satellites in the corresponding coalition formation model.

[0036] Alliance Revenue

[0037] Coalition revenue is a criterion for measuring the quality of a coalition formed among observation satellites. Based on the constraints and objective function proposed in Chapter 1, we design the coalition revenue during the formation process of observation satellite coalitions. Each observation mission corresponds to one observation satellite coalition; therefore, the coalition revenue formed by observation satellites should be the sum of the individual revenues of all observation satellites that selected that observation mission, specifically expressed as:

[0038] Constellation coalition formation focuses more on the cooperative behavior among observation satellites, maximizing the total benefit of all coalitions by forming a stable coalition structure. The objective of constellation coalition formation is defined as follows:

[0039] In the formula: This indicates the final observation satellite alliance formed. According to (10), the final solution objective of the star cluster alliance is consistent with the optimization objective of the star cluster multi-target collaborative observation mission. The final observation satellite alliance result is the allocation result of the star cluster multi-target collaborative observation mission.

[0040] Cluster Alliance Formation Switching Rules and Stable Alliance Partitions Switching rules In constellation formation games, a coalition represents a group of observation satellites. Assuming each coalition corresponds to one observation mission, within a satellite scheduling cycle, each observation satellite can only choose one coalition. The allocation process divides the observation satellites into disjoint coalitions, known as coalition partitioning.

[0041] League Division In the formation of constellation alliances based on the problem of allocating multi-target collaborative observation tasks, the alliance partition consists of sets of non-overlapping observation satellite alliances. ,in This indicates the observation satellite consortium corresponding to the empty mission. Indicates observation task The corresponding observation satellite alliance, Indicates the first The number of observation tasks within a scheduling cycle. Coalition partitions. All alliances satisfy the following relationship:

[0042] In the formation of a constellation alliance, each observation satellite initially selects an empty mission, and the resulting alliance structure is designated as the initial alliance partition. Based on this initial alliance partition, an observation satellite can choose to leave its current alliance and join a new one, at which point both the individual satellite's gain and the alliance's overall gain will change. When changing alliances, observation satellites must adhere to certain switching rules, typically designed based on the observation satellite's alliance preferences; these rules are therefore also known as preference relationships.

[0043] Preference Relationship For any observation satellite Given two alliances and Preference Relationship Defined as an observation satellite In the league division Choose to join any two alliances in the game. The willingness to join the alliance is greater than or equal to the willingness to join the alliance. The will, This indicates that the observation satellite Prefer to join the league .

[0044] In the formation of constellation alliances, each observation satellite chooses among observation missions, thus joining or leaving an alliance based on preferences. This paper chooses the overall benefit of all alliances as the preference relationship, i.e., for any observation satellite... Given two arbitrary alliances and The preference relationship is expressed as:

[0045] In the formula: Indicates equivalence relation, Indicates observation satellite Not a member of the alliance Based on the aforementioned preference relationships, the switching rules that must be followed during the game are given.

[0046] Switching rules: Given the current set of alliance partitions If and only if At that time, the observation satellite Will choose to leave the league Join the alliance Meanwhile, observation satellites This switching operation will cause changes to the existing set of alliance partitions, further updating them to... .

[0047] Nash Stable Coalition Partitions: The ultimate goal of constellation coalition formation is to create a stable set of coalition partitions, maximizing the overall benefit of all coalitions and achieving a stable state. This is true if and only if for any observation satellite... If all satellites tend to choose their current alliance, and switching alliances cannot improve the overall alliance's benefits by observing the satellites themselves, then this is called an alliance partition. It is a stable Nash-style league division Based on the above definition, the objective of constellation alliance formation is the same as the objective of the constellation multi-target collaborative observation task allocation problem, and the resulting Nash-stable alliance partition is the solution to the task allocation problem.

[0048] Algorithm for solving star cluster alliance formation To obtain a Nash-stable coalition partition, firstly, each observation satellite needs to select an appropriate observation task, i.e., join a corresponding coalition, based on preference relationships and switching rules. Then, each coalition resolves conflicts among its observation satellites, eliminating those exceeding the limit or offering minimal benefits. Finally, this process is iterated until a stable coalition partition converges. The flowchart for the coalition formation algorithm is shown below. Figure 2 As shown The iterative process based on the star cluster alliance formation algorithm is shown in Algorithm 1.

[0049] Algorithm 1 Iterative process based on the star cluster alliance formation algorithm Input: Set of observed satellites Observation task set Initial Alliance Partition ; Output: Nash's stable league division That is, the final result of the multi-target collaborative observation task allocation for the constellation. 1: Initial Alliance Division In China, each observation satellite has chosen the Space Mission Alliance. ; 2: Assume the current iteration number is , No. The results of the next iteration of the alliance partitioning are: ; 3: If passed If the algorithm still fails to form a stable coalition partition after the [number]th iteration, then proceed to the [number]th iteration. Next iteration: 4: / / Get the alliance partitioning results from the previous iteration 5: ; / / Update each alliance / / Task selection phase 6: For each observation satellite : 7: Retrieve the current alliance partition results from the decision center. ; 8: For each observation task : 9: Calculate observation satellites Perform observation tasks Alliance benefits to be obtained Compared with the total revenue of the alliance, observation satellites The preferred observation task is selected based on the preference relationship, and then the switching operation is performed based on the switching rules; 10: End the loop; 11: End the loop; 12: The decision center updates the results for each coalition based on the decision actions of each observation satellite. ; / / Conflict resolution phase 13: For each observation task : 14: If the alliance Number of observation satellites / / Exceeding the limit on the number of observation satellites 15: Against the Alliance The observation satellites in the alliance are distributed according to the alliance's revenue. Sort from largest to smallest ; 16: For each observation satellite : 17: When observing satellites index hour: 18: / / That is, select the observation satellite Perform observation tasks ; 19: Otherwise: 20: / / That is, select the observation satellite Execute an empty task ; 21: End judgment; 22: End the loop; 23: The new alliance results and Update to League Division Results middle; 24: Otherwise: 25: [This likely refers to a specific alliance or alliance structure.] Update to League Division Results middle; 26: End judgment; 27: End the loop; 28: Determine if the consortium partitioning result is Nash stable. If a Nash stable consortium partition has been formed, the current algorithm ends, and the consortium partitioning result is determined. This represents the constellation mission allocation result within this scheduling cycle; if a Nash-stable coalition partition is not formed, the next iteration will continue. 29: End the iteration process; The specific steps of the above star cluster alliance formation algorithm are as follows: (a) Initialization phase The initialization phase of the algorithm mainly involves the observation satellite set. Perform initial parameter settings, and then configure the alliance partition. Initialization is performed, meaning all observation satellites select to perform no-mission tasks. .

[0050] (b) Task selection phase The iterative process based on the constellation alliance formation algorithm is mainly divided into a mission selection phase and a conflict resolution phase. In the mission selection phase, each observation satellite first obtains the current alliance partitioning result from the decision center. Then calculate the coalition revenue that the observation satellite will receive for each observation mission it performs. Then, based on preference relationships, a decision is made on whether to execute a switching operation according to the switching rules. Finally, the decision center updates the results for all alliances based on the decision actions of each satellite. .

[0051] (c) Conflict resolution phase During the mission selection phase, there may be a situation where certain observation satellites concentrate on performing a particular observation mission. To resolve this issue, the conflict resolution phase requires iterating through each observation mission to determine the coalition. Has the number of observation satellites exceeded the upper limit? If the number of observation satellites exceeds the upper limit, then the alliance will... The observation satellites are sorted from highest to lowest according to their alliance benefits, and the current observation mission alliances are cleared. Then, the corresponding number of observation satellites are selected according to the upper limit, and other observation satellites exceeding the upper limit are changed to empty mission alliances. Finally, the new alliance results will be announced. and Update to League Division Results If the number of observation satellites is less than or equal to the upper limit, it proves that there is no conflict in the alliance, and therefore the alliance can be directly added to the list. Update to League Division Results Medium is fine.

[0052] (d) Nash Stable Union Partition Determination Phase After each iteration, if the current coalition partitioning result... Compared to the previous league partition results If no change occurs, it is considered that a Nash-stable coalition partition has been formed, and the algorithm iteration process ends; otherwise, return to step (2) and continue to the next iteration.

[0053] When a Nash-stable coalition partition is finally formed after continuous iteration, meaning that all observation satellites tend to choose the current observation mission coalition and it is impossible to increase the total benefit of all coalitions through unilateral coalition switching operations of observation satellites, then the swarm coalition formation algorithm converges, and the finally formed coalition partition is the final solution to the swarm multi-target collaborative observation mission allocation problem.

[0054] The second step involves the design and implementation of a real-time simulation architecture for a constellation intelligent collaborative observation interactive virtual simulation verification platform.

[0055] The constellation intelligent mission planning technology designed in the first step only performs mission planning for constellations at the algorithm level, but it cannot intuitively reflect the feasibility of the algorithm. The design verification platform's real-time simulation architecture consists of three main parts: a main control unit, a visual unit, and an algorithm unit. It aims to achieve a real-time simulation environment with functional separation, high cohesion, and low coupling. The platform supports satellite mission planning, threat scenario simulation, algorithm verification, and result visualization. Through a front-end and back-end separation model, it ensures the system's flexibility and scalability. The main control unit provides a graphical user interface (GUI) for operators to configure, start, monitor, and control simulation experiments. It is responsible for sending user commands to the visual unit and retrieving experimental data from the database. The visual unit receives and executes commands from the main control unit, rendering 3D scenes such as satellites, Earth, and orbits in real time; packaging scene state information and sending it to the algorithm unit; receiving the calculation results from the algorithm unit and visualizing them; and feeding back key results to the main control unit for recording. The algorithm unit receives environmental information from the visual unit, runs core mission planning, resource allocation, or adversarial decision-making algorithms, and returns the calculation results to the visual unit.

[0056] Detailed Explanation of the Communication Mechanism Between the Main Control Unit and the View Unit The communication function is developed based on a client / server architecture. The main control software acts as the server, adopts the TouchSocket framework, and uses TCP / IP as the data transmission protocol to realize data interaction with the visual unit and the algorithm unit.

[0057] Network communication model selection The OSI model (Open Systems Interconnection Model) is a layered architecture for network communication proposed by the International Organization for Standardization (ISO). Its aim is to standardize network communication functions and enable interoperability between different systems. The OSI model divides network communication functions into seven layers, each undertaking specific tasks and interacting with the layers above and below it. Figure 3 As shown, the TCP / IP model defines the physical layer and data link layer of the OSI model as the network interface layer, merges the application layer, presentation layer, and session layer into the application layer, and keeps the transport layer and network layer with the same definition.

[0058] Each layer of the TCP / IP network model performs different tasks: the network interface layer manages the physical network interface and sends and receives data frames on the network; it includes all hardware and protocols and is responsible for physical data transmission. The network layer is responsible for device routing and packet forwarding to transmit data from the source to the destination in complex networks; the IP protocol is the main protocol operating at this layer. The transport layer is responsible for providing data transmission services; TCP and UDP are the main protocols at this layer, used to ensure reliable and fast data transmission, respectively. The application layer is responsible for directly providing network services to users and defines the communication specifications and interaction methods between applications.

[0059] Network communication model selection The network interface layer and network layer of the TCP / IP network model are mainly related to hardware devices, and their functions can be implemented using existing equipment such as routers and switches. Therefore, the development of the communication module focuses more on the selection of transport and application layer protocols at the software level. The main control software uses the communication module to interact with the visual unit and algorithm unit, displaying the received data to the user and storing it in the database for subsequent analysis. High requirements are placed on the security, orderliness, and reliability of data communication; therefore, TCP should be selected as the transport layer protocol.

[0060] Development of a communication module based on the client-server model Once the network communication model and protocol are determined, the next step is to develop specific network communication modules. The client-server communication model is a common network architecture model where applications on the network are divided into two categories: clients and servers. This model is widely used in various network services and distributed systems and is one of the fundamental architectures of the Internet.

[0061] The client-server communication process is as follows: Figure 4 As shown. The server is the main control software, and the client is the visual software or algorithm unit. The specific execution steps are as follows: ① After the main control software starts, it instantiates the TcpService class, sets the IP address of the client to be listened to, and then starts the server to continuously check whether the client is connected.

[0062] ② After the visual software or algorithm unit is started by the main control software through an external program call module, it sets the IP address and port number of the server to be connected and initiates a connection request.

[0063] ③ After the connection is established, the client sends simulation data to the server, and the server sends a response receipt to the client after receiving it. This process is repeated continuously.

[0064] ④ The connection will be automatically disconnected after the client or server program is closed.

[0065] Detailed Explanation of the Communication Mechanism Between the Algorithm and the Viewpoint Architecture The simulation data calculated by the algorithm needs to be transmitted in real time to the visual computer for visualization and simulation verification. A controllable data transmission timing interaction interface is required between the algorithm software (Python) and the visual software (Unity). The ML-Agents plugin, released by Unity3D in 2018, makes real-time interaction between Python and Unity3D possible. Because newer versions of ML-Agents have stronger encapsulation, especially in terms of algorithm openness, the earliest version of ML-Agents was chosen. The ML-Agents framework is as follows: Figure 5 As shown.

[0066] ML-Agents Real-time Simulation Architecture The relationship between the three high-level components of the ML-Agents framework is as follows: the simulation environment (Unity3D) communicates with the Python API through ExternalCommunicator. The following describes the composition of the simulation environment and the transmission principle of ExternalCommunicator.

[0067] The learning environment includes three add-on components to help organize the Unity3D scene: the Agent component, the Brain component, and the Academy component. The Agent component can be attached to a Unity Game Object, corresponding one-to-one with a satellite in the scene. It is responsible for collecting environmental information and executing control strategies. The Brain component encapsulates the Agent's decision-making logic and is responsible for receiving information collected by the Agent. Agents with similar behaviors can share a single Brain component. To reduce coupling and facilitate data processing, we still maintain a one-to-one correspondence between Agents and Brains. The Academy component is the control center of the entire interactive environment, responsible for directing the Agent's observation and decision-making processes and coordinating data between different Brains. It can also specify several environmental parameters, and the communication interface with Python is also located in this component.

[0068] A socket is a software abstraction layer located between the application layer and the TCP / IP protocol suite. Employing the facade pattern, it hides the complexities of the protocol suite from the user, allowing programmers to focus on business logic development while the socket handles data organization to meet specified communication protocols. A combination of IP address and port number uniquely identifies a socket. First, the server initializes the socket parameters, then binds it to the specified port, and subsequently calls `accept` to block the process, waiting for client connections. When a client makes a service request, it first initializes a socket, specifying the target IP address and port number, and then calls `connect` to connect to the server. If the connection is successfully established, the server can process the client's request, and both parties begin data exchange. Since the number of tasks that can be trained simultaneously by a training algorithm is limited, typically a one-to-one relationship, the ML-Agents server uses a blocking approach to handle client requests. That is, the main thread blocks and waits for client connection requests; when a client connects, the main thread immediately creates a new sub-thread to communicate with that client connection, while the main thread continues to wait for new client connections.

[0069] Setting up a real-time simulation environment The training environment setup revolves around the three components mentioned above: Academy, Brain, and Agent. The scene setup steps are as follows: The Academy component configures the IP address and port number of the algorithm software. As the central controller in the Unity scene, it is responsible for configuring the connection (IP and port) with the Python server and coordinating the entire data interaction lifecycle. All Brain models in the scene are managed centrally under the Academy component.

[0070] Brain is responsible for distributing control variables to the Agent downwards and summarizing the environment state to the algorithm upwards. Therefore, depending on the specific task, the dimensions of the environment state and the control variables need to be specified in Brain. The calculation formula is: dimension × number of states. The data structure for interaction between Unity and Python is defined. It specifies the dimensions of the "State" and "Action" spaces. In this design, the "State" space is used to pass scene situations from Unity to Python, and the "Action" space is used to pass the algorithm's calculation results from Python to Unity.

[0071] Bind an Agent component to each agent in the scene. The Agent component is responsible for acquiring specific environmental data and executing the control variables distributed by Brain.

[0072] The following section focuses on the implementation of the functionalities provided in the Agent component. The method call logic diagram is as follows: Figure 5 As shown. When the scenario receives a reset command from the algorithm, it calls the AgentReset() function, which is responsible for initializing the agent's state at the start or reset of the simulation. The scenario calls the CollectObservations() function once per time step, adding information one by one using the built-in AddVectorObs() function, and then combining and packaging the observation vectors to send them to the Python server through the ML-Agents framework. The scenario calls the AgentAction() function to execute the control variables stored in the action array vectorAction.

[0073] In summary, by configuring the three components of ML-Agents and implementing the scenario logic within the constructed virtual simulation environment, a real-time simulation environment can be efficiently built. This ensures that the algorithm's planning results can be executed promptly after real-time addition of anomaly information.

[0074] The third step involves the design and implementation of the visual software for the interactive virtual simulation verification platform for stellar cluster collaborative tracking observation.

[0075] To meet the mission environment and demonstration requirements of satellite constellations, this invention receives satellite parameter data selected by the main control computer and drives a virtual simulation model of the Earth and satellites in a 3D scene to create the satellite constellation's operating environment. This invention utilizes the Unity3D physics engine to build the satellite constellation's working environment. Unity is a real-time 3D interactive content creation and operation platform with powerful cross-platform capabilities, a stunning special effects system, a complete basic framework, sophisticated performance analysis tools, an extensible editor, a convenient resource management system, and a wealth of plugins, making it suitable for building this working environment and showcasing the visuals.

[0076] The overall structure of the following visual software is as follows: Figure 6 As shown, the three-dimensional virtual scene can be divided into three parts: constellation operation module, human-computer interaction module, and task planning module. The technical solution is described below.

[0077] Constellation Operation Module: This invention primarily involves the simulation of satellite operation in a space scenario. Developed using the Unity engine, it is suitable for scenarios such as orbit visualization and task allocation demonstration. This module details the technical process of calculating satellite orbits based on Two-Line Element (TLE) data and performing simulations within the Unity visual software. With the rapid development of satellite constellation technology, especially the emergence of large-scale satellite constellations like Starlink, the surge in the number of satellites has posed a significant challenge to traditional orbit analysis methods. The sheer number of Starlink satellites means that directly using the TLE data of each satellite for simulation would result in an extremely high computational burden, increasing computational complexity and severely limiting the feasibility of real-time simulation. Therefore, how to reduce the load on the simulation system without sacrificing accuracy through reasonable clustering and simplification methods has become an important research topic. This paper proposes a simplified simulation method based on satellite orbital parameter clustering analysis. By clustering and classifying the satellite orbital planes, the average orbital parameters within each cluster are calculated, and the feasibility is verified through simulation to achieve efficient analysis and demonstration of large-scale satellite constellation orbits.

[0078] Data extraction and preliminary classification Data Extraction: For large satellite constellations like Starlink, the first step is to extract orbital parameters associated with each satellite from the satellite database, particularly the inclination and right ascension of the ascending node. Inclination and right ascension are crucial parameters of the satellite's orbital plane, determining its orbital inclination and the location of its nodes on the equatorial plane. Extracting these parameters provides data support for subsequent orbital plane classification and simulation.

[0079] Preliminary Classification: Based on the extracted orbital parameters, satellites are initially classified into coarse groups. This preliminary classification primarily relies on inclination and right ascension of the ascending node, dividing satellites into different orbital plane groups to form initial subsets. This process simplifies subsequent cluster analysis steps by reducing the orbital variability among individual satellites.

[0080] K-means clustering orbital plane classification Clustering Algorithm Selection: Based on the initial classification, the K-means clustering algorithm is further employed to refine the classification of satellite orbital planes. The goal of clustering algorithms is to group satellites with similar orbital parameters into the same cluster, thereby reducing the number of orbital planes that need to be simulated individually. The K-means algorithm, through iterative optimization, assigns each satellite to the nearest cluster center, ensuring high consistency of orbital parameters within each cluster.

[0081] Cluster center selection: The K-value, i.e., the number of cluster centers, is an important parameter affecting the clustering effect. Choosing an appropriate K-value requires comprehensive consideration of constellation size and simulation accuracy. For large constellations, appropriately increasing the number of cluster centers can improve simulation accuracy, while reducing the K-value can significantly reduce the computational load.

[0082] Calculation and simulation of average orbital parameters Orbit parameter averaging: For each cluster, the system calculates the average of all satellite orbit parameters (including inclination, right ascension of the ascending node, argument of perigee, and semi-major axis) within the cluster. The purpose of averaging is to simplify the orbit parameters of satellites within each cluster without significantly sacrificing the accuracy of the orbit simulation, so that an averaged orbital plane can be used to represent the orbital plane of the entire cluster.

[0083] Orbital plane simulation: By calculating the average orbital parameters, the simulation system can generate representative orbital planes for each cluster. These orbital planes can be used for subsequent simulation analysis, thereby reducing the complexity of satellite orbital plane simulation.

[0084] Performance Evaluation: The satellite operation within each cluster is evaluated, with key indicators including orbital plane accuracy, observation mission completion rate, and orbital overlap rate. The effectiveness of the orbital parameter averaging method can be verified by comparing the simulation results of the original TLE data with those of the simplified clustering simulation.

[0085] Simulation demonstration in Unity scene Importing Orbit Parameters to Unity: The average orbit parameters and their operational data calculated in STK are imported into Unity. The system extracts orbit parameters by parsing TLE data, calculates the satellite's position and velocity in the geocentric inertial coordinate system (ECI) using an orbital dynamics model, and maps the results to a Unity 3D scene, displaying the satellite's trajectory and orbital path in real time. The system includes the following core modules: TLE Parsing Module: Parses TLE data and extracts orbit parameters using the Decoder class; Orbit Calculation Module: Calculates satellite orbit parameters and ECI position using the Orbit, SolutionD, and SolutionG classes; Visual Demonstration: ECI coordinates are mapped to the Unity scene using the SatelliteScript class, enabling visualization of satellite motion, orbit rendering, and ground station observations. Basic elements of the Earth and satellite virtual scene are added, with the satellite model responsible for receiving simulation data and simulating the control process. Accurate and aesthetically pleasing 3D models are essential for a good demonstration. Modeling software such as 3ds Max, Blender, and Maya are used, and the models are then converted to the standard .fbx model format supported by Unity3D. Real-time demonstration is then performed in the 3D virtual scene. Through real-time rendering of the satellite's operational status, users can intuitively observe the coordinated movement of the satellite constellation in its orbit, orbital coverage, and orbital plane adjustments.

[0086] Interactive Simulation: Utilizing Unity's interactive features, users can observe the satellite's operational status in real-time across different orbital planes, adjust the viewing angle to observe from different perspectives, or manually switch orbital plane groups to verify their operational effects. This step enhances the interactivity of the simulation, allowing users to gain a deeper understanding of the actual effects of orbital plane clustering simulation. Human-Computer Interaction Module: Built using WPF and the HandyControl UI library, as shown in ViewMain.xaml. The interface uses the Ribbon control, with clearly defined functional areas, including task control, scene selection, task planning, and parameter settings. User interaction is implemented through Command binding in the MVVM pattern, and data is sent to the viewport via communication. Text boxes and buttons are developed using Unity 3D's built-in UGUI and TextMeshPro components, while charts are implemented using the Xcharts component. The graphical interface is mainly divided into three parts: the satellite status information panel, the injected anomaly information panel, and the task planning results panel. The satellite status information panel includes the selected satellite's number, location information, and allows users to interact by clicking on satellites to view their on-orbit status. The injected anomaly information panel includes the selection of the abnormal satellite number, satellite maneuver acceleration in each direction, and modification of the abnormal maneuver time. The task planning results panel mainly displays the real-time task planning results given by the constellation intelligent planning algorithm, as well as the overall task planning results.

[0087] 3. Task Planning Module: To meet the information display needs of the virtual simulation platform, the main demonstration area includes a simulation time zone and a satellite status bar. The simulation time zone displays the real-time simulation time. Each satellite calculates its current position based on the simulation time for reference by simulation personnel. During the simulation, personnel can adjust the simulation system's running speed using the slider in the human-computer interaction area of ​​the main control software, thereby accelerating the simulation process. The satellite status bar mainly displays status information of the currently selected satellite, including satellite name, elevation angle, azimuth angle, latitude, longitude, altitude, as well as observation target and mission completion status, facilitating understanding of the current satellite status and mission execution status for simulation personnel.

[0088] After receiving task allocation data from the algorithm, each task needs to be assigned to its corresponding satellite to complete the visualization of target observations. Since a single satellite may need to perform multiple observation tasks, all the processed data is first stored in a task list. Each row of data represents an observation task, consisting of four parts: target number, satellite number, start time, and end time. Each satellite matches a task in the task list based on its own number. In Unity, the GameObject of the corresponding satellite is obtained, and the static target to be observed is added to the corresponding Place of the GameObject. Then, the task is removed from the task list to prevent duplicate execution. Finally, the visualization of static target observations is performed based on the task start and end times.

[0089] The satellite constellation visual demonstration process is as follows. Figure 7 shows the simulation platform startup interface. Experimenters can interact with the virtual simulation platform to select functions such as viewing constellation configuration, starting simulation, or exiting the simulation platform. When the experimenter selects to start simulation, the visual software enters the interface shown in Figure 8 when the satellite constellation is working normally in orbit. Observation satellites and Starlink satellites can be added. Experimenters can select typical threat scenarios from the drop-down menu and modify the abnormal satellite number, satellite maneuvering acceleration in each direction, and abnormal maneuvering time. Figure 9 This is the interface after the anomaly injection. It shows that the experimenter injected a star-sensor fault, causing the satellite to turn yellow and exhibit abnormal maneuvers. After the intelligent planning algorithm finishes running, a relay observation scenario is demonstrated. Figure 10 , Figure 11 This is the real-time constellation mission planning result display interface. The full-view demonstration process verifies the effectiveness of the constellation intelligent mission planning algorithm proposed in this invention.

[0090] The following are specific examples: .

[0091] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. An interactive virtual simulation verification system for stellar cluster collaborative tracking observation, characterized in that: This includes both the platform side and the algorithm side; The platform runs on a high-performance server, providing a runtime environment and computing power support for human-computer interaction software and 3D virtual scenes. The platform includes a main control unit and a visual unit; The main control unit is the control and communication center of the threat target collaborative tracking and observation virtual simulation platform. It is mainly responsible for the overall demonstration and verification process control and human-computer interaction, sending control commands and collecting experimental data through network communication, and implementing data management based on the database. The visual unit is the demonstration and verification center of the UAV swarm decision-making virtual simulation platform. It constructs multiple virtual scenes corresponding to the overall task flow based on the Unity engine, and at the same time provides real-time environmental parameter information for the intelligent planning algorithm of the algorithm side. The algorithm acquires real-time environmental parameter information sent by the visual unit, generates decision planning results, and feeds the decision planning results back to the visual unit. The visual unit visualizes the decision planning results and sends them to the main control unit. The main control unit stores the decision planning results and ultimately verifies the effectiveness of the intelligent planning algorithm under real-time environmental parameter information, realizing a three-dimensional visualization demonstration and verification of the intelligent constellation's on-orbit operation.

2. The interactive virtual simulation verification system for stellar cluster collaborative tracking observation according to claim 1, characterized in that: The specific process by which the system verifies the effectiveness of the intelligent planning algorithm under real-time environmental parameter information is as follows: Step 1: The algorithm is based on an adaptive alliance dynamic programming algorithm; when a threat from an enemy satellite is detected, research on real-time tracking technology for the enemy threat target needs to be carried out to complete the mission planning for the observation satellite constellation. Step 2: The platform design implements the real-time simulation architecture of the constellation collaborative observation interactive virtual simulation verification platform; Step 3: The platform-side design implements the visual software for the constellation collaborative observation interactive virtual simulation verification platform.

3. The interactive virtual simulation verification system for stellar cluster collaborative tracking observation according to claim 2, characterized in that: In step 1, once a threat from an enemy satellite is detected, research on real-time tracking technology for the enemy threat target needs to be conducted. First, considering multiple constraints such as unobservable blind spots, double coverage, and space payload resources, a constellation collaborative intelligent mission planning model is established with the goal of maximizing the collaborative tracking of dynamic threat targets. Second, alliance switching rules between observation satellites and conditions for the formation of stable alliance partitions in the constellation are proposed. Finally, a constellation collaborative observation mission planning solution algorithm based on the alliance formation is designed to realize mission planning for the observation satellite constellation.

4. The interactive virtual simulation verification system for stellar cluster collaborative tracking observation according to claim 3, characterized in that: In step 2, the design and implementation of the real-time simulation architecture of the constellation collaborative observation interactive virtual simulation verification platform are carried out. In response to the real-time simulation verification requirements of the constellation mission planning algorithm, a real-time simulation architecture is implemented based on ML-Agents, and a real-time training environment is built. The Unity client receives real-time data transmitted from the Python client to perform mission planning, while the Python client receives satellite status information fed back from the Unity client. The constellation intelligent mission planning technology is used for online real-time mission allocation.

5. The interactive virtual simulation verification system for stellar cluster collaborative tracking observation according to claim 4, characterized in that: In step 3, the visual software of the constellation collaborative observation interactive virtual simulation verification platform was designed and implemented. For the mission scenarios of complex spacecraft, a three-dimensional interactive virtual simulation scene was created based on Unity3D, and satellite models and Earth models were constructed based on 3D Max. The satellite operation simulation was driven by parsing TLE data. The functional interactive interface was clearly constructed using WPF and HandyControl UI library, and a graphical interface was developed and designed based on UGUI components, thus realizing the construction of the interactive virtual simulation verification platform. Develop real-time satellite mission planning results and verify the feasibility and reliability of the intelligent planning method through a collaborative observation simulation interface.

Images