Satellite control ground station intelligent task automatic operation method based on large model

Abstract

The application provides a satellite TT&C ground station intelligent task automatic operation method based on a large model, relates to the technical field of satellite TT&C and large language model (LLM) application, and solves the limitation problem of preset logic automatic TT&C. According to the demand analysis information of a TT&C task, a knowledge base for different subsystems is formed; the LLM is trained based on the field knowledge of a satellite TT&C ground station, a multi-role station control agent for automatic operation of a TT&C task is constructed based on the LLM after training, and the agent is deployed in the satellite TT&C ground station and calls multiple types of business equipment; when the TT&C task is executed, the multi-role station control agent publishes the TT&C task demand, generates a TT&C task plan, executes actions according to the knowledge base corresponding to different subsystems, organizes and completes the capture, tracking, TT&C, data reception and recording of an overflight satellite, automatically analyzes the TT&C task execution situation and generates a prompt and a solution when an exception occurs.

Description

Technical Field

[0001] This invention relates to the field of satellite telemetry, tracking, and command (TT&C) and large language model (LLM) application technology, specifically to a method for automatic operation of intelligent missions of satellite TT&C ground stations based on large models. Background Technology

[0002] In the field of aerospace telemetry, tracking, and command (TT&C), with continuous technological advancements and increasing mission complexity, the requirements for the automation level of ground stations are also rising. Currently, the development trend of intelligent equipment is to achieve comprehensive environmental perception, automatic control, trend prediction, and optimized decision-making. These systems provide powerful auxiliary decision support for equipment operation and maintenance through unified collection, management, and in-depth analysis of various data. However, in the actual operation of aerospace TT&C ground stations, although research on artificial intelligence (AI) and deep learning algorithms is quite mature and there is a large body of literature serving as a theoretical foundation, the direct application of advanced AI technologies such as Large Language Modeling (LLM) to practical scenarios, especially specific application schemes for aerospace TT&C ground stations, is still in the exploratory stage.

[0003] Traditional aerospace telemetry, tracking, and command (TT&C) ground station management software relies on pre-defined, fixed procedures to execute satellite TT&C tasks. While this logic-based automation can meet the needs of routine missions, it often falls short in handling unexpected situations such as satellite anomalies or equipment malfunctions. Once an anomaly occurs, existing automated systems struggle to automatically return to normal operation, typically requiring intervention from experienced personnel. This not only increases the cost of human intervention but can also negatively impact mission success rates due to slow response times.

[0004] Furthermore, the differences in geographical location, functional requirements, and equipment configuration among various monitoring and control sites present challenges in adapting to automated processes. This means that to ensure efficient operation of each site, a customized automation solution must be developed, which undoubtedly increases the complexity and cost of development and maintenance. At the same time, this customization process also increases the possibility of errors, further enhancing management difficulty.

[0005] In summary, how to leverage the latest artificial intelligence technologies and large language models to enhance the intelligence level of aerospace telemetry and control ground stations, especially their adaptability in abnormal situations, has become an urgent problem to be solved. However, this problem can be addressed by designing an automated mission operation mode that incorporates large language models from scratch. Therefore, by introducing more flexible and intelligent technologies, the need for manual intervention can be effectively reduced, the reliability and efficiency of the system can be improved, and thus, the challenges brought by future space missions can be better met. Summary of the Invention

[0006] Based on the current state of the technology, the purpose of this invention is to provide artificial intelligence technology and large language models to support the automatic operation of satellite tracking and control missions. Therefore, a method for the automatic operation of intelligent missions at satellite tracking and control ground stations based on large models is proposed. This invention abandons the traditional mechanized process design approach and instead designs intelligent agents for different roles based on station management methods. These agents collaborate to complete the automatic operation control of various tracking and control tasks. This invention thus solves the limitations of traditional pre-set logic-based automated tracking and control, and is applicable to various business functions in satellite tracking and control ground stations, successfully implementing artificial intelligence technology and large language models in the aerospace tracking and control field.

[0007] The present invention employs the following technical solutions to achieve its objective:

[0008] A method for automated operation of intelligent missions at satellite tracking, telemetry, and command (TT&C) ground stations based on a large model includes: forming multiple knowledge bases for different subsystems based on the requirements analysis information of the TT&C mission; training a large language model (LLM) based on the domain knowledge of the satellite TT&C ground station to construct a multi-role station control agent with the trained LLM as the basic architecture for automated operation of TT&C missions; deploying the multi-role station control agent in the satellite TT&C ground station, where the multi-role station control agent calls and controls various types of service equipment; when the satellite TT&C ground station executes TT&C missions, the multi-role station control agent publishes TT&C mission requirements, generates TT&C mission plans, executes actions based on the knowledge bases corresponding to different subsystems, organizes and completes the acquisition, tracking, TT&C, data reception and recording of passing satellites, automatically analyzes the execution status of TT&C missions, and generates prompts and solutions when anomalies occur.

[0009] Furthermore, the requirements analysis information for the measurement and control tasks is determined by technical personnel with measurement and control expertise. The requirements analysis information includes process documents recorded in natural language. The LLM is trained based on the process documents, and after training, when executing measurement and control tasks, it uses the logical description in the process documents as a basis to obtain the status of multiple types of business devices under the subsystem and complete the corresponding call control actions.

[0010] Preferably, the process document includes an overall process and subsystem processes. The overall process is used to maintain the global control content of the telemetry, tracking and command (TT&C) mission unchanged, while the subsystem processes are used to deal with the independent control content of different satellite TT&C ground stations or different business equipment in a satellite TT&C ground station. Both the overall process and the subsystem processes include control commands for business equipment. The lower limit of the control command level is any specific control action or operation of any specific business equipment.

[0011] Furthermore, the multi-role station control intelligent agent includes a station manager intelligent agent, a task scheduling intelligent agent, a satellite management intelligent agent, and a subsystem intelligent agent. The station manager intelligent agent receives and publishes telemetry and control (TT&C) task requests from users, continuously responding to new tasks or maintenance needs. The task scheduling intelligent agent responds to newly published TT&C task requests by establishing TT&C task plans, calculating orbits, and allocating resources, while simultaneously observing existing TT&C task plans for actions that meet time requirements and executing them. The satellite management intelligent agent receives the scheduled TT&C task requests and provides the corresponding satellite and task information. Based on receiving TT&C task requests, satellite information, and task information, the subsystem intelligent agent performs operational control on the satellite TT&C ground station's equipment, completing specific operational actions.

[0012] Specifically, before executing the measurement and control task, based on the knowledge base of different subsystems, each intelligent agent in the multi-role station control intelligent agent is pre-configured with corresponding task objectives and prompt word templates, and associated with the role attributes of each intelligent agent; each intelligent agent solves the specific problems corresponding to its own role when executing the measurement and control task according to its configured task objectives and prompt word templates.

[0013] Preferably, each agent in the multi-role station control agent is configured with an independent running process. When performing measurement and control tasks, each agent listens for the appearance of input corresponding to its own role. If the input appears, it processes the input based on the configured task objective and LLM to obtain the corresponding result. The result is then returned to form a new input, which is then listened to and processed by other agents.

[0014] In summary, due to the adoption of this technical solution, the beneficial effects of this invention are as follows:

[0015] This invention proposes an operational framework based on a large model and intelligent agents. Through training and learning, the intelligent agents can extract patterns from data and autonomously make decisions and predictions. They only need to be provided with tools that provide ultimate control capabilities, such as various devices at the site, to automatically run telemetry and control tasks. The organization and application of equipment and tools, and the fulfillment of business scenario requirements, are all determined collaboratively by the intelligent agents. Developers previously only needed to perform business requirement analysis and configure the corresponding professional knowledge base for the intelligent agents. Compared with traditional automated operation modes, although the application scenarios are the same, the application logic of the large model LLM differs from fixed preset programming code, resulting in significant differences in subsequent scalability. With the continuous development of artificial intelligence technology and computing hardware, the aerospace telemetry and control solutions brought about by artificial intelligence interaction and control guided by this invention will have increasingly obvious advantages.

[0016] This invention allows end-users or system designers to directly participate in the implementation and modification of system task flows without overly relying on coding skills and programmers; operators can control the system by directly inputting natural language. During the development phase, developers no longer need to code specific task scenarios or processes; instead, they only need to ensure the corresponding intelligent agent understands the process's knowledge or steps and provide equipment monitoring tools, allowing the intelligent agent to automatically run and complete the task.

[0017] Therefore, the method of this invention improves the flexibility and variability of satellite telemetry and control ground stations, lowers the technical threshold, reduces reliance on operator proficiency, and reduces the risk of operational errors, providing new possibilities for the optimization and improvement of mission processes for aerospace telemetry and control ground stations.

[0018] The method of this invention can also unify the automatic operation implementation mode of various telemetry and control ground stations. Regardless of whether their mission requirements or specific configurations are the same, the intelligent agent itself has professional aerospace telemetry and control knowledge, can handle various mission requirements, and can also acquire the design characteristics, equipment configuration and operation experience of various telemetry and control ground stations to realize differentiated execution processes. This solves the limitation problem of different manufacturers and inconsistent operation processes, and helps to accumulate operation experience, summarize lessons, and optimize the mission management capabilities of telemetry and control ground stations.

[0019] In summary, the method of this invention has advantages such as strong adaptability, good scalability, simple deployment, short development cycle, and low-code development. Its application will enhance customer experience, reduce operation and maintenance costs, and change the current software development model in the aerospace telemetry, tracking, and command (TT&C) field. It allows subsystem or overall designers to directly modify the operational logic, processes, and external interfaces of site software by modifying documents, without requiring specialized programming languages ​​and program deployment knowledge. This approach helps accelerate the development process, reduce costs, and allows system designers or customers to directly participate in software design and implementation without affecting the system's flexibility and stability. AI technology can thus be successfully implemented in the aerospace TT&C field, providing technical reserves and core competitiveness for subsequent related technological fields. Attached Figure Description

[0020] Figure 1 A schematic diagram of the architecture of the multi-role station control intelligent agent constructed by the method of the present invention;

[0021] Figure 2 This is a schematic diagram of the dynamic process editing process in the method of the present invention;

[0022] Figure 3 This is a schematic diagram of the dynamic expansion device process in the method of the present invention;

[0023] Figure 4 This is a schematic diagram of the health management process in the method of the present invention;

[0024] Figure 5 This is a flowchart illustrating the business equipment monitoring function in the method of the present invention;

[0025] Figure 6 This is a schematic diagram of the planned automatic operation process in the method of the present invention;

[0026] Figure 7 This is a schematic diagram of the automatic operation process for task preparation in the method of the present invention;

[0027] Figure 8 This is a schematic diagram of the automatic task execution process in the method of the present invention;

[0028] Figure 9 This is a schematic diagram of the automatic execution process upon task completion in the method of the present invention;

[0029] Figure 10 This is a schematic diagram illustrating the automatic start-up process of the tracking function in the method of the present invention;

[0030] Figure 11 This is a schematic diagram illustrating the automatic operation process of tracking the end of the process in the method of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0032] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0033] Example

[0034] A method for automated operation of intelligent tasks at satellite tracking, telemetry, and command (TT&C) ground stations based on a large model is proposed. This method includes: forming multiple knowledge bases for different subsystems based on TT&C task requirements analysis information; training a large language model (LLM) based on domain knowledge of the satellite TT&C ground station to construct a multi-role station control agent with the trained LLM as its foundational architecture, used for automated TT&C task operation; deploying the multi-role station control agent in the satellite TT&C ground station, where it calls and controls various types of service equipment; during TT&C task execution at the satellite TT&C ground station, the multi-role station control agent publishes TT&C task requirements, generates TT&C task plans, executes actions based on the knowledge bases corresponding to different subsystems, organizes and completes the acquisition, tracking, TT&C, data reception and recording of passing satellites, automatically analyzes the TT&C task execution status, and generates prompts and solutions in case of anomalies.

[0035] Traditional telemetry, tracking, and command (TT&C) task flow control requires process coding, configuration, or graphical drag-and-drop for specific task scenarios. Due to diverse requirements and varying equipment, task flows cannot be standardized across different satellite TT&C ground stations, resulting in repetitive development work. Furthermore, users cannot modify the flow or define new flows for control. Therefore, this embodiment uses a Large Language Model (LLM) for related design processing, and focuses on preventing the "illusion" phenomenon caused by unclear flows during the design process, thereby avoiding the problem of unstable flow control.

[0036] In this embodiment, the requirements analysis information for the measurement and control task is determined by technical personnel with expertise in measurement and control. This information includes process documents recorded in natural language. These process documents should be written with clear logic to ensure that the LLM can utilize its natural language processing capabilities. The LLM is trained based on these process documents, and upon completion of training, when executing the measurement and control task, it uses the logical descriptions in the process documents to obtain the status of various business devices under the subsystem and complete the corresponding call control actions.

[0037] In this preferred embodiment, the process document includes an overall process and subsystem processes. The overall process maintains the global control content of the telemetry, tracking, and command (TT&C) mission unchanged, while the subsystem processes address the independent control content of different satellite TT&C ground stations or different service devices within a single satellite TT&C ground station. Both the overall process and the subsystem processes include control commands for service devices. The lower limit of the control command level is any specific control action or operation of any specific service device, clearly specifying what control action or operation a service device should perform at a given time, ensuring that the LLM does not experience illusions due to the inability to find the accurate tool corresponding to the device.

[0038] The following is a detailed introduction to the multi-role station control intelligent agent constructed and used in this embodiment. In terms of business capabilities, the multi-role station control intelligent agent is responsible for the monitoring and control of the entire satellite telemetry, tracking, and command (TT&C) ground station, which is equivalent to using a Large Language Model (LLM) to complete the functions of traditional TT&C ground station system station control software.

[0039] like Figure 1 As shown, the multi-role station control intelligent agent includes a station manager intelligent agent, a task scheduling intelligent agent, a satellite management intelligent agent, and a subsystem intelligent agent. The station manager intelligent agent receives and publishes telemetry and control (TT&C) task requests from users, continuously responding to new tasks or maintenance needs. The task scheduling intelligent agent responds to newly published TT&C task requests by establishing TT&C task plans, calculating orbits, and allocating resources, while simultaneously observing existing TT&C task plans for actions that meet time requirements and executing them. The satellite management intelligent agent receives scheduled TT&C task requests and provides the corresponding satellite and task information. Based on the received TT&C task requests, satellite information, and task information, the subsystem intelligent agent performs operational control on the satellite TT&C ground station's equipment, completing specific operational actions.

[0040] Figure 1 In this context, LLM stands for Large Language Model, and Agent refers to an intelligent agent. A Large Language Model is an artificial intelligence system capable of processing and generating natural language. It learns from large amounts of text data to understand human language and can generate coherent and meaningful responses or content based on context. An intelligent agent refers to a software entity with a certain degree of autonomy, capable of perceiving its state, performing tasks, making decisions, and taking actions to achieve goals in a specific environment. Intelligent agents typically possess learning capabilities and can optimize their behavioral strategies over time. In addition to the specific intelligent agent roles mentioned earlier in this embodiment, Figure 1 The virtual company software agent and health management agent shown in the document will be described in detail later.

[0041] From a software architecture perspective, multi-role station control intelligent agents, because they have multiple specific intelligent agents, include elements such as corresponding roles, actions, and knowledge bases. Their basic architecture is a common LLM, which is used for unified integration.

[0042] In a multi-role station control intelligent agent, each role is an intelligent agent, configured with its own knowledge base and actions before construction or during use, and publishes its own task requirements, which are accepted and completed by other corresponding intelligent agents. Therefore, a multi-role station control intelligent agent is equivalent to a measurement and control station management team, with each specific intelligent agent role being equivalent to the personnel configuration for each position, the knowledge base being equivalent to the professional knowledge and job descriptions possessed by each position, and the actions being equivalent to the skills and executable operations possessed by each position.

[0043] In this embodiment, before executing the measurement and control task, based on the knowledge base of different subsystems, corresponding task objectives and prompt word templates are pre-configured for each agent in the multi-role station control agent, and these are associated with the role attributes of each agent. Each agent solves the specific problems corresponding to its role when executing the measurement and control task according to its configured task objectives and prompt word templates. Table 1 below lists and explains the specific roles in the multi-role station control agent, which can be consulted simultaneously and its content can be referenced to configure the agent accordingly in actual applications.

[0044] Table 1. Role Allocation Diagram for Multi-Role Station Control Agents

[0045]

[0046]

[0047] The above method utilizes the definition of multiple roles to gradually realize the initial task requirements. In the application of the multi-role station control agent, the overall entry point is the requirements of the station manager agent, and then each subsequent role will process these requirements according to the decomposed workflow steps.

[0048] Furthermore, as a preferred embodiment, each agent in the multi-role station control agent is configured with an independent running process. When executing measurement and control tasks, each agent listens for inputs corresponding to its own role. If such input appears, it processes the input based on the configured task objective and LLM to obtain the corresponding result, and returns the result to form a new input for other agents to listen for and process. In this way, once other roles in the system hear inputs related to their own tasks, they can acquire and execute the tasks. This method is similar to the personnel management and organizational structure of current measurement and control stations, with each position responding in real time to its corresponding new task requirements.

[0049] In this embodiment, before and during the execution of the measurement and control task, additional role assistance components are pre-configured for the multi-role station control agent to assist it in executing the task. These role assistance components may include typical search role components and prompt decomposition roles. Prompt decomposition refers to breaking down a complex prompt into smaller, more manageable and processable components to better understand and optimize the role of each part. This method can help improve the quality and relevance of content generated by large language models.

[0050] In this embodiment, before and during the execution of the measurement and control task, additional role helper components are pre-configured for the multi-role station control agent to assist the multi-role station control agent in executing the measurement and control task. For the multiple agent roles of the multi-role station control agent and the multiple specific component roles in the role helper components, each has a py file corresponding to its own skills. The py file predefines templates for the corresponding skills. Each role uses its own skills to process the input and output of the corresponding measurement and control task part and complete the measurement and control task of its own part.

[0051] Example 2

[0052] Based on Example 1, this example provides a more detailed description of some specific agent roles in the multi-role station control agent and the corresponding processes in the method implementation.

[0053] Firstly, in a multi-role station control intelligent agent, the dynamic process editing of the subsystem intelligent agent is divided into a development phase and an operational phase, which can be viewed simultaneously. Figure 2 This is an illustration. The process is used to configure and edit the execution sequence of actions for various types of business equipment in a satellite telemetry, tracking, and command (TT&C) ground station. First, during the development phase, the knowledge base for the new process is configured. The client, by running the RAG terminal, inputs process documents related to requirements analysis information. The embedded model then vectorizes the process documents and stores them as the knowledge base of the corresponding subsystem. Next, during the execution phase, the sequence design of the new process is completed. The station manager agent, in conjunction with the task scheduling agent, inputs the TT&C task requirements. The embedded model reads the corresponding knowledge base and returns the relevant text of the TT&C task requirements. The task scheduling agent then integrates the process requirements, generates corresponding prompts, and hands them over to the LLM for processing. The LLM generates the process and forms a stack of actions to be executed, determines the next business equipment tool to be called, and hands it over to the task scheduling agent and the subsystem agent to execute the relevant actions until the execution phase ends.

[0054] In this embodiment, the multi-role station control intelligent agent also includes a virtual software company intelligent agent, which can be viewed synchronously. Figure 3 The flowchart illustrates how a virtual software company's intelligent agent dynamically expands new business equipment and develops corresponding business functions in satellite telemetry and control ground stations, thereby enabling the editing function of equipment monitoring interfaces. Based on the telemetry and control task requirements corresponding to the new business equipment and functions, the virtual software company's intelligent agent generates prompts for writing and testing code, which are then handed over to LLM to generate monitoring interface code and test code. After testing and confirmation by the virtual software company's intelligent agent, the corresponding code is added to the list of callable control tools, and the tool list of the subsystem intelligent agent corresponding to the business equipment is updated for use when executing telemetry and control tasks.

[0055] In this embodiment, the multi-role station control intelligent agent also includes a health management intelligent agent, which can be viewed simultaneously. Figure 4The flowchart illustrates the use of a health management agent to perform health management and testing on various agents within a multi-role station control agent and on various types of operational equipment in a satellite telemetry and control ground station. Based on health detection instructions from the station manager agent, the health management agent performs fault detection operations according to a preset cycle, retrieves health management knowledge from a preset health management knowledge base, generates corresponding health management prompts, and submits them to the LLM (Local Management Module) to generate the health management process. The task scheduling agent and subsystem agents, based on the health management process, obtain corresponding task plan information and detection point status, automatically perform fault detection, and obtain detection results. The health management agent then generates a health management report, ending the health management process.

[0056] The following is an introduction to the various satellite telemetry and control mission functions that can be realized by combining the specific intelligent agent roles in the multi-role station control intelligent agent in this embodiment.

[0057] The monitoring workflow for various types of operational equipment in satellite telemetry, tracking, and command (TT&C) ground stations can be found here. Figure 5 As illustrated, the station manager agent here acts as the recipient of user input device control requests. It adds context and generates corresponding prompts based on this information. LLM decomposes the task, extracts the business device name and related commands, calls the tools of the corresponding subsystem agent, realizes the call of the business device interface, and controls it to complete the corresponding measurement and control operations.

[0058] For information on receiving and automatically executing the measurement and control plan, please refer to [link / reference]. Figure 6 As illustrated, after a plan is created through the station manager's intelligent agent terminal, the LLM organizes the plan. The satellite management intelligent agent queries satellite information and default parameters, and then the LLM generates the corresponding plan. The task scheduling intelligent agent adds the newly created plan generated by the LLM to the list and checks the time nodes. While executing the plan, it continuously traverses the plan to complete the operation execution at each time node.

[0059] For the preparation process of the telemetry and control mission, the automatic operation flow can be found in [link to documentation]. Figure 7 As illustrated, the task scheduling agent generates preparation steps based on the LLM and the knowledge base corresponding to the task preparation; after obtaining the telemetry, tracking, and command (TT&C) task information, the satellite management agent queries the TT&C task information and macro parameters, the LLM takes over and generates macro control steps, forms relevant commands and matches the macro control tools corresponding to each subsystem agent; the subsystem agents thus complete their respective business functions and prepare to implement relevant controls to execute the task.

[0060] The start and end processes of a telemetry and control mission are similar and can be viewed simultaneously. Figure 8 and Figure 9As illustrated, both are initiated by a task scheduling agent. The LLM generates relevant steps based on a knowledge base corresponding to the start or end of the measurement and control task. After the task begins, the subsystem agent processes each business device based on the process steps; after the task ends, the task scheduling agent records the task work report and clears the relevant plans.

[0061] During the telemetry, tracking, and command (TT&C) mission, taking satellite tracking as an example, Figure 10 and Figure 11 The processes for starting and ending tracking are illustrated separately. After the task scheduling agent determines the start of tracking, the LLM generates the tracking start process steps based on the corresponding knowledge base. At this time, the subsystem agent acquires relevant feature parameters. After the LLM organizes these feature parameters, the subsystem agent sets up and starts the corresponding service equipment to realize tracking-related functions such as information scanning and reporting. When tracking ends, the LLM generates the tracking end process steps based on the corresponding knowledge base. The subsystem agent then controls various service equipment to stop data processing and transmission, the information is stored, and various service equipment enters standby mode, waiting for the next measurement and control task.

Claims

1. A method for intelligent automatic operation of satellite tracking, telemetry, and command (TT&C) ground stations based on a large model, characterized in that, include: Based on the requirements analysis information of the measurement and control mission, multiple knowledge bases are formed for different subsystems; Based on knowledge of satellite telemetry, tracking, and command (TT&C) ground stations, a large language model (LLM) is trained to construct a multi-role station control intelligent agent with the trained LLM as the basic architecture, which is used for the automatic operation of TT&C missions. A multi-role station control intelligent agent is deployed in the satellite tracking and control ground station. The multi-role station control intelligent agent calls and controls the various subsystem devices in the satellite tracking and control ground station. When the satellite tracking and control ground station performs tracking and control tasks, the multi-role station control intelligent agent issues tracking and control task requirements, generates tracking and control task plans, executes actions based on the knowledge base corresponding to different subsystems, organizes and completes the acquisition, tracking, control, data reception and recording of passing satellites, automatically analyzes the tracking and control task execution status and generates prompts and solutions when anomalies occur. The multi-role station control intelligent agent includes a station manager intelligent agent, a task scheduling intelligent agent, a satellite management intelligent agent, and a subsystem intelligent agent. The dynamic process editing of the subsystem intelligent agent is divided into a development phase and an operation phase. This process is used to configure and edit the action execution sequence of various business equipment in the satellite telemetry, tracking, and command (TT&C) ground station. First, in the development phase, the knowledge base of the new process is configured. The client inputs process documents related to requirements analysis information by running the RAG terminal. After the process documents are vectorized using the embedded model, they are stored as the knowledge base of the corresponding subsystem. Next, in the operation phase, the sequence design of the new process is completed. The station manager intelligent agent, in conjunction with the task scheduling intelligent agent, inputs the TT&C task requirements, reads the corresponding knowledge base from the embedded model, and returns the relevant text of the TT&C task requirements. The task scheduling intelligent agent integrates the process requirements, forms corresponding prompt words, and hands them over to the LLM for processing. The LLM generates the process and forms a stack of actions to be executed, determines the tool corresponding to the next business equipment to be called, and hands it over to the task scheduling intelligent agent and the subsystem intelligent agent to execute the relevant actions until the end of the operation phase.

2. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 1, characterized in that: The requirements analysis information for measurement and control tasks is determined by technical personnel with measurement and control expertise. The requirements analysis information includes process documents recorded in natural language. The LLM is trained based on the process documents. After training is completed, when executing measurement and control tasks, it uses the logical description in the process documents as a basis to obtain the status of each subsystem device under the subsystem and complete the corresponding call control actions.

3. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 2, characterized in that: The process document includes an overall process and subsystem processes. The overall process is used to maintain the global control content of the telemetry, tracking, and command (TT&C) mission unchanged, while the subsystem processes are used to deal with the independent control content of different satellite TT&C ground stations or different business equipment in a satellite TT&C ground station. Both the overall process and the subsystem processes include control commands for each subsystem equipment. The lower limit of the control command level is any specific control action or operation of any specific business equipment.

4. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 1, characterized in that: The station administrator intelligent agent receives and publishes telemetry and control (TT&C) task requests from users, continuously responding to new tasks or maintenance needs. The task scheduling intelligent agent responds to newly published TT&C task requests, scheduling TT&C tasks by establishing TT&C task plans, calculating orbits, and allocating resources. Simultaneously, it observes existing TT&C task plans for actions that meet time requirements and executes them. The satellite management intelligent agent receives the scheduled TT&C task requests and provides the corresponding satellite and task information. Based on receiving TT&C task requests, satellite information, and task information, the subsystem intelligent agent performs business control on the operational equipment of the satellite TT&C ground station to complete specific business actions.

5. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 4, characterized in that: Before executing the measurement and control task, based on the knowledge base of different subsystems, the corresponding task objectives and prompt word templates are pre-configured for each intelligent agent in the multi-role station control intelligent agent, and are associated with the role attributes of each intelligent agent. Each intelligent agent solves the specific problems corresponding to its own role when executing the measurement and control task according to its configured task objectives and prompt word templates.

6. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 5, characterized in that: Each agent in the multi-role station control agent is configured with an independent running process. When performing measurement and control tasks, each agent listens for the appearance of input corresponding to its own role. If the input appears, it processes the input based on the configured task objective and LLM to obtain the corresponding result, and returns the result to form a new input, which is then listened to and processed by other agents.

7. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 4, characterized in that: Before and during the execution of measurement and control tasks, additional role helper components are pre-configured for the multi-role station control agent to assist the multi-role station control agent in executing the measurement and control tasks. For the multiple agent roles of the multi-role station control agent and the multiple specific component roles in the role helper components, each has a py file corresponding to its own skills. The py file predefines templates for the corresponding skills. Each role uses its own skills to process the input and output of the corresponding measurement and control task part and complete the measurement and control task of its own part.

8. The intelligent mission automatic operation method for satellite telemetry and control ground stations according to claim 4, characterized in that: The multi-role station control intelligent agent also includes a virtual software company intelligent agent, which is used to dynamically expand new business equipment and develop corresponding business functions in satellite telemetry and control ground stations. Based on the telemetry and control task requirements corresponding to the new business equipment and functions, the virtual software company intelligent agent generates prompts for writing and testing code, which are then handed over to LLM to generate monitoring interface code and test code. After testing and confirmation by the virtual software company intelligent agent, the corresponding code is added to the list of callable control tools, and the tool list of the subsystem intelligent agent corresponding to the business equipment is updated for use when executing telemetry and control tasks.

9. The intelligent mission automatic operation method for satellite telemetry, tracking, and command ground stations according to claim 4, characterized in that: The multi-role station control intelligent agent also includes a health management intelligent agent, which is used to perform health management and detection of various intelligent agents in the multi-role station control intelligent agent and various types of business equipment in the satellite telemetry and control ground station; Based on the health detection instructions from the station manager's intelligent agent, the health management intelligent agent performs fault detection operations according to a preset cycle, retrieves health management knowledge from the preset health management knowledge base, generates corresponding health management prompts, and submits them to the LLM to generate the health management process; Based on the health management process, the task scheduling agent and the subsystem agent obtain the corresponding task plan information and the status of the detection points, automatically perform fault detection and obtain the detection results, and then the health management agent generates a health management report and ends the health management process.

Images