A smart scheduling and control system and method for aerospace telemetry and tracking ground stations

By constructing an intelligent scheduling and control system for aerospace telemetry and control ground stations, and using large models and reinforcement learning algorithms for multi-task joint optimization, the system solves the problem of low efficiency in resource scheduling and task control in multi-satellite and multi-task scenarios, realizes autonomous task execution and continuous model optimization, and improves the system's autonomous and reliable operation capability.

CN123117618APending Publication Date: 2026-11-1710TH RES INST OF CETC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611252751.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing aerospace telemetry and control systems are inefficient in resource scheduling and mission management in multi-satellite and multi-mission scenarios, making it difficult to adapt to the needs of massive satellite telemetry and control missions and the elastic scheduling of heterogeneous resources. Traditional scheduling modes rely on human experience and are difficult to achieve dynamic mission autonomous decision-making, resulting in insufficient response timeliness and system autonomy.

Method used

A smart scheduling and control system for aerospace telemetry, tracking, and command (TT&C) ground stations is constructed, comprising a mission control unit, an intelligent model unit, a resource scheduling unit, a model object unit, a data-real symbiotic parallel deduction unit, and a human-computer interaction unit. This forms a resource scheduling loop and a mission control loop. A large model and reinforcement learning algorithms are used for multi-task joint optimization to achieve an end-to-end intelligent closed loop.

Benefits of technology

It improved resource utilization and task fulfillment rate, reduced the operational burden on personnel, enhanced the system's autonomous and reliable operation capability in high-concurrency and multi-constraint scenarios, solved the problems of frequent resource conflicts and long scheduling decision-making time in multi-star and multi-task scenarios, and realized autonomous task execution and continuous model optimization.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117618A_ABST
    Figure CN123117618A_ABST
Patent Text Reader

Abstract

This invention discloses an intelligent scheduling and control system and method for aerospace telemetry and control ground stations, belonging to the field of aerospace telemetry and control technology. The key technical points of its solution are: it includes a task control unit, an intelligent model unit, a resource scheduling unit, a model object unit, a data-real symbiotic parallel inference unit, and a human-computer interaction unit. The task control unit analyzes task characteristics and issues control commands; the intelligent model unit provides domain knowledge and reasoning support; the resource scheduling unit uses large models and / or reinforcement learning algorithms to perform multi-task joint optimization to generate resource scheduling commands; the model object unit manages antenna, baseband, and transmission resources and executes commands; the data-real symbiotic parallel inference unit constructs a digital twin environment to generate incremental simulation data to drive model iteration; and the human-computer interaction unit presents resource status and operation recommendations. This invention improves resource utilization and task fulfillment rate, reduces personnel burden, and enhances the system's autonomous and reliable operation capability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of aerospace telemetry and control technology, and more specifically, to an intelligent scheduling and control system and method for aerospace telemetry and control ground stations. Background Technology

[0002] Spaceborne telemetry, tracking, and command (TT&C) systems are crucial infrastructure for real-time tracking, telemetry, and remote control of satellites and other spacecraft, providing key support for ensuring the safe operation of various spacecraft in complex space environments. In recent years, with the rapid development of the space industry, the launch density of spacecraft has increased significantly, the number of on-orbit space platforms has grown exponentially, and the demand for TT&C missions has become increasingly strong, placing higher demands on the complexity and real-time performance of spacecraft management. Simultaneously, the accelerated deployment of TT&C infrastructure has driven the evolution of ground-based TT&C resources from isolated sites to wide-area distribution.

[0003] However, this trend also brings the dual challenges of a rapid expansion in the scale of telemetry, tracking, and command (TT&C) resources and a simultaneous increase in the difficulty of mission management: on the one hand, resources such as heterogeneous antennas, baseband, and transmission links are diverse and their states change dynamically, making it difficult for traditional pre-planning and manual scheduling modes to meet the needs of efficient resource scheduling; on the other hand, in multi-satellite, multi-mission scenarios, TT&C requirements frequently conflict, and the existing TT&C mode has low interaction efficiency, further increasing the difficulty of resource scheduling and mission management. To address these issues, existing solutions mostly focus on the scheduling optimization of single TT&C stations or homogeneous resources, limiting application scenarios and making it difficult to adapt to the needs of massive satellite TT&C missions and the elastic scheduling of heterogeneous resources. Furthermore, existing scheduling and management modes rely on manual experience to build static models, making it difficult to achieve dynamic mission autonomous decision-making, resulting in significant deficiencies in response timeliness and system autonomy.

[0004] Therefore, researching and designing an intelligent scheduling and control system and method for aerospace telemetry and control ground stations that can overcome the above-mentioned defects is an urgent problem to be solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent scheduling and control system and method for aerospace telemetry and control ground stations. This system enables an end-to-end intelligent closed loop for telemetry and control tasks, from requirement analysis to parameter issuance and from anomaly detection to autonomous handling. It improves resource utilization and task fulfillment rates, reduces the operational burden on personnel, and enhances the system's autonomous and reliable operation capabilities.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: Firstly, an intelligent scheduling and control system for aerospace telemetry, tracking, and command ground stations is provided, including: The task management unit is used to parse the task characteristics of the measurement and control task request, send the task characteristics to the resource scheduling unit and the intelligent model unit, and issue control commands to the model object unit. The intelligent model unit is used to provide domain knowledge and reasoning support to the resource scheduling unit and the mission management unit based on the aerospace telemetry and control knowledge base and telemetry and control base model. The resource scheduling unit is used to perform multi-task joint optimization based on the task characteristics and the resource status information fed back by the model object unit, and to generate resource scheduling instructions by using a large model and / or reinforcement learning algorithm. The model object unit is used to manage antenna resources, baseband resources, and transmission resources, execute the resource scheduling instructions and control instructions issued by the task management unit, and provide feedback on the execution results. The parallel simulation unit of digital-real coexistence is used to construct a digital twin environment for simulation based on the real operating data of the model object unit, generate incremental simulation data and feed it back to the intelligent model unit to drive model iteration; The human-computer interaction unit is used to present the resource status and operation recommendations to the user, and forward the user's confirmation command to the resource scheduling unit or the task management unit. The resource scheduling unit forms a small scheduling loop of perception, scheduling, evaluation, and optimization, while the task management unit forms a measurement and control loop of demand reception, task planning, process decision-making, parameter orchestration, equipment control, and result evaluation. The two closed loops are coordinated through the execution feedback of the model object unit and the knowledge support of the intelligent model unit.

[0007] Furthermore, the resource scheduling unit includes a resource scheduling agent, a resource and task perception module, a target and constraint analysis module, an air interface resource scheduling module, an antenna resource scheduling module, a baseband resource scheduling module, and a decision performance evaluation module; The resource and task perception module is used to receive resource status and the requirements of the measurement and control tasks to be executed. With the support of the resource scheduling agent, it parses the task type, priority, time window and required resource type, and transmits the task analysis results and resource status to the target and constraint analysis module and each resource scheduling module. The target and constraint analysis module is used to analyze constraints using the resource scheduling agent, form a structured constraint model, and transmit it to each resource scheduling module. The air interface resource scheduling module, the antenna resource scheduling module, and the baseband resource scheduling module transmit decision information bidirectionally, and use large models and / or reinforcement learning algorithms to perform multi-task collaborative optimization and comprehensively generate a resource scheduling scheme. The decision performance evaluation module is used to receive the decision results and performance feedback results of each resource scheduling module, perform performance evaluation to obtain a decision score level, and feed back the evaluation results to each resource scheduling module to adjust the resource scheduling scheme.

[0008] Furthermore, the operation of the scheduling loop includes: The resource and task awareness module triggers a scheduling event after detecting a change in resource status. Each resource scheduling module re-optimizes the multi-task joint operation based on the current resource status and task queue, generates a new scheduling strategy, and sends it to the model object unit for execution. The decision performance evaluation module collects the execution results and evaluates the scheduling effect. If the scheduling effect is lower than the preset threshold, the scheduling event is triggered again to form a closed loop iteration.

[0009] Furthermore, the resource scheduling agent is constructed using a large model algorithm, providing model reasoning and optimization algorithm support for each module; The resource scheduling agent comprehensively analyzes the input information of each module in complex scenarios with multiple concurrent tasks, assists in generating resource scheduling strategies, and outputs risk warnings when resources are insufficient or conflicts cannot be resolved.

[0010] Furthermore, the task management unit includes a task management intelligent agent group, a requirement receiving module, a task planning module, a process decision module, a parameter arrangement module, an equipment control module, and a result evaluation module; The requirement receiving module is used to receive measurement and control task requests and parse the task type, priority and resource requirements. The task planning module is used to generate a task execution plan based on task characteristics, including time window allocation and resource configuration scheme; The process decision module is used to generate a sequence of equipment control actions based on the task execution plan, combined with the task management intelligent agent group and the knowledge base in the aerospace telemetry and control field, to form the task operation process definition. The parameter arrangement module is used to automatically arrange the control parameters of each measurement and control device according to the task planning and operation process definition, and generate a parameter configuration instruction set; The device control module is used to send control commands to the corresponding hardware devices via the resource scheduling unit according to the parameter configuration instruction set. The result evaluation module is used to obtain data returned by the device after the task is completed, to comprehensively evaluate the task completion, execution deviation and data quality, generate task execution evaluation results and feed them back to the demand receiving module to support dynamic adjustment of the process.

[0011] Furthermore, the operation of the measurement and control loop includes: After receiving a new task, the demand receiving module triggers the task planning module to generate a task execution plan. The process decision module determines the control process based on the task execution plan; The parameter orchestration module generates a parameter configuration instruction set, and the device control module sends the parameter configuration instruction set as a control instruction to the model object unit for execution. The result evaluation module continuously monitors the execution status during task execution. If an anomaly is detected, an anomaly handling process is triggered, a recommended handling strategy is generated, and the strategy is presented to the operator for confirmation via the human-computer interaction unit. After the task is completed, the result evaluation module performs a comprehensive evaluation and feeds back the evaluation results to the requirement receiving module for dynamic process adjustment.

[0012] Furthermore, the parallel simulation unit for digital-real coexistence is used to construct a digital twin environment that is deeply mapped to the real measurement and control system, so as to realize the parallel coexistence of digital virtual space and real physical space; The parallel simulation unit, which is based on real measurement and control scenarios, establishes a parallel virtual environment. During periods without actual measurement and control tasks, it performs virtual task deployment and closed-loop simulation, continuously generating incremental simulation data that includes normal operating conditions, boundary scenarios, and abnormal modes. This data is used to drive the fine-tuning and updating of the domain-wide model in the intelligent model unit, as well as the training and iteration of the models within the resource scheduling unit and the task management unit.

[0013] Furthermore, the intelligent model unit includes a domain knowledge base construction module, a model fine-tuning module, a model lightweighting module, and a measurement and control base model module; The domain knowledge base construction module is used to collect and process multi-source heterogeneous documents in the aerospace measurement, control and operation field to build a knowledge base for the aerospace measurement, control and operation field. The model fine-tuning module is used to collect operation and maintenance feedback data generated in real measurement and control tasks, construct a fine-tuning sample set, and use efficient parameter fine-tuning and incremental learning strategies to fine-tune and optimize the inference strategy of the large model. The model lightweighting module is used to adapt a finely tuned large model to the edge environment through model compression technology, and comprehensively adopts model quantization, pruning and knowledge distillation methods to reduce the model's computational complexity and memory usage; The measurement and control base model module is used to integrate the results of domain knowledge base construction, model fine-tuning, and model lightweighting to form a large model capability base for aerospace measurement and control services.

[0014] Furthermore, the human-computer interaction unit includes a display and control interaction module, a resource status presentation module, an operation recommendation agent, an interface operation recommendation module, and an interactive data transmission and reception module; The display and control interaction module is used to support operators in inputting measurement and control task requirements, dynamically displaying the task execution process and equipment action status, displaying the execution results and evaluation conclusions after the task is completed, and supporting visual retrieval and input of domain knowledge base. The resource status presentation module is used to display the status of telemetry and control resources and the status of mission progress on a visual interface, including real-time resource occupancy, target satellite information and mission execution stage; The operation recommendation agent is used to generate recommendations for the next operation steps based on the task type, operator identity, and current execution steps, combined with historical operation records and expert experience, and to assign a recommendation score to each recommended operation. The interface operation recommendation module is used to display a list of recommended operations on the operation interface, support operators to confirm and execute them, and trigger a manual confirmation and feedback mechanism at key decision nodes. The interactive data transceiver module is used to receive and forward information data generated during human-computer interaction, distribute operation change instructions to the task management unit or the model object unit, and forward the received task execution status and resource status data to the resource status presentation module and the display and control interaction module.

[0015] Secondly, a method for intelligent scheduling and control of aerospace telemetry and control ground stations is provided, which is applied to an intelligent scheduling and control system for aerospace telemetry and control ground stations as described in any one of the first aspects, including a training and deployment phase and an operation phase. The training deployment phase includes: Collect multi-source heterogeneous data in the aerospace measurement, tracking, and control field, and build a knowledge base for the aerospace measurement, tracking, and control field; The knowledge base is used to pre-train or retrieve large domain models to enhance the generation of adaptations, and the measurement and control base model is formed through model fine-tuning and lightweight processing. Based on the measurement and control base model and knowledge base, an intelligent agent is designed for resource scheduling, task management and human-computer interaction scenarios; The trained base model and intelligent agent are integrated into each unit to complete the deployment and configuration; The operational phase includes: Receive telemetry and control task requests and parse task characteristics to generate a task planning list; Real-time collection of resource status information to generate a status map of available resources; By integrating task requirements and resource status, a resource scheduling strategy is generated by invoking a scheduling algorithm. Automatically generate task operation procedures and arrange equipment control parameters based on resource scheduling results and domain knowledge base; The instruction set is sent to the measurement and control equipment to establish a measurement and control link and initiate task execution. The status is continuously monitored during task execution, and when an anomaly is detected, a recommended handling strategy is generated and presented to the human-computer interaction interface. After the task is completed, a performance evaluation is performed, and the evaluation results are fed back to the knowledge base; After the task is completed, a resource release operation is performed to restore the device to a schedulable state.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention provides an intelligent scheduling and control system for aerospace telemetry and control ground stations. It constructs an intelligent scheduling and control system with six interconnected units: a task control unit, an intelligent model unit, a resource scheduling unit, a model object unit, a data-real symbiotic parallel deduction unit, and a human-computer interaction unit. This system forms a dual-closed-loop collaborative architecture: the resource scheduling unit's perception, scheduling, evaluation, and optimization scheduling loop, and the task control unit's demand reception, task planning, process decision-making, parameter orchestration, equipment control, and result evaluation loop. This achieves an end-to-end intelligent closed loop for telemetry and control tasks, from demand analysis to parameter issuance, and from anomaly detection to autonomous handling. This invention improves the resource utilization and task fulfillment rate of aerospace telemetry and control ground stations, reduces the operational burden on personnel, and enhances the system's autonomous and reliable operation capability under high concurrency and multi-constraint scenarios.

[0017] 2. This invention forms a complete closed-loop scheduling mechanism through the refined division of labor and collaboration among the resource scheduling intelligent agent, resource and task perception module, target and constraint analysis module, air interface resource scheduling module, antenna resource scheduling module, baseband resource scheduling module, and decision performance evaluation module within the resource scheduling unit. The resource and task perception module acquires resource status and task requirements in real time. The target and constraint analysis module uses the resource scheduling intelligent agent to analyze multi-dimensional constraints to form a structured constraint model. The air interface resource scheduling module, antenna resource scheduling module, and baseband resource scheduling module bidirectionally transmit decision information and employ large-scale models and / or reinforcement learning algorithms for multi-task collaborative optimization. The decision performance evaluation module provides feedback on the scheduling results. This layered, decoupled, and collaboratively optimized architecture enables the system to quickly generate globally optimal resource allocation schemes under complex constraints and continuously approach the best scheduling effect through closed-loop iteration. This solves the problems of frequent heterogeneous resource conflicts and long scheduling decision-making time in multi-satellite, multi-task scenarios, improving the efficiency and accuracy of resource scheduling.

[0018] 3. This invention constructs a complete lifecycle management closed loop for measurement and control tasks through the organic coordination of a task management intelligent agent group within the task management unit and modules for demand reception, task planning, process decision-making, parameter orchestration, equipment control, and result evaluation. The demand reception module parses task requests, the task planning module generates execution plans, the process decision-making module generates equipment action sequences based on a domain knowledge base, the parameter orchestration module automatically orchestrates control parameters, the equipment control module issues instructions via the resource scheduling unit, and the result evaluation module comprehensively evaluates the execution effect and provides feedback for adjustments. This fully automated and intelligent task management mechanism enables routine measurement and control tasks to be completed autonomously without human intervention, while supporting manual confirmation and anomaly handling at key nodes. It achieves a shift from manual dominance to autonomous operation, effectively solving the problems of low interaction efficiency, rigid processes, and slow response to temporary tasks in traditional measurement and control modes, thus improving task execution efficiency and accuracy.

[0019] 4. This invention constructs a digital twin environment deeply mapped to the real measurement and control system through a parallel simulation unit that coexists with the real system. During periods without actual measurement and control tasks, it continuously performs multi-type simulations, generating incremental simulation data encompassing normal operating conditions, boundary scenarios, and abnormal modes. This data is then fed back to the intelligent model unit to drive fine-tuning and updates of the large model, as well as the training iterations of models within the resource scheduling and task management units. This mechanism of virtual-real coexistence and continuous evolution allows the system to accumulate a large amount of training data offline, continuously optimizing model performance and generalization ability. It solves the problems of scarce real measurement and control data and difficulty in covering extreme scenarios, providing continuous optimization momentum for the long-term stable operation of the measurement and control system.

[0020] 5. This invention constructs a dedicated large-scale model capability base for aerospace telemetry and control operations through the collaborative work of the domain knowledge base construction module, model fine-tuning module, model lightweighting module, and telemetry and control base model module within the intelligent model unit. The domain knowledge base construction module stores multi-source heterogeneous telemetry and control professional knowledge in a structured manner; the model fine-tuning module continuously optimizes the model inference strategy using operation and maintenance feedback data; the model lightweighting module compresses the large model into a lightweight version that can be deployed at the edge through techniques such as model quantization, pruning, and knowledge distillation; and the telemetry and control base model module provides knowledge retrieval and inference decision-making services to the outside world through a unified interface. This end-to-end model construction and optimization system enables the large model to deeply adapt to the professional needs of the telemetry and control field, while achieving real-time inference at resource-constrained edge devices. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a system block diagram in Embodiment 1 of the present invention; Figure 2 This is a flowchart from Embodiment 2 of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0023] Example 1: An intelligent scheduling and control system for aerospace telemetry and control ground stations, such as Figure 1 As shown, it includes a task management unit, an intelligent model unit, a resource scheduling unit, a model object unit, a data-real symbiotic parallel deduction unit, and a human-computer interaction unit. These six units work together to form a scheduling loop centered on the resource scheduling unit and a measurement and control loop centered on the task management unit, jointly realizing an end-to-end intelligent closed loop from task requirement analysis to parameter issuance, and from anomaly detection to autonomous handling.

[0024] The task management unit (TMU) parses the task characteristics of measurement and control task requests, sends these characteristics to the resource scheduling unit and the intelligent model unit, and issues control commands to the model object unit. The TMU is the task entry point and the core of the entire system's lifecycle management. It receives measurement and control task requests submitted by users in natural language or structured forms, and automatically converts unstructured descriptions into structured measurement and control task requirements through semantic understanding and intent recognition, including information such as task type, priority, time window, and required resource types. The parsed task characteristics are simultaneously sent to the resource scheduling unit for resource allocation decisions and to the intelligent model unit for acquiring domain knowledge and reasoning support. The TMU is also responsible for issuing control commands to the model object unit during task execution, driving the antenna, baseband, and transmission equipment to establish the measurement and control link and execute the task.

[0025] The intelligent model unit provides domain knowledge and reasoning support to the resource scheduling unit and mission management unit based on the aerospace telemetry and control knowledge base and telemetry and control foundation model. The intelligent model unit is the system's intelligent brain, integrating professional knowledge from the aerospace telemetry and control field with the reasoning capabilities of a large model. When the resource scheduling unit performs multi-task joint optimization, the intelligent model unit provides services such as semantic parsing of task constraints, reasoning of resource matching rules, and diagnostic suggestions for abnormal situations. When the mission management unit makes process decisions, the intelligent model unit provides support such as historical case retrieval, operation sequence generation, and parameter rationality verification. The intelligent model unit provides services externally through a unified interface, allowing the resource scheduling unit and mission management unit to operate without needing to concern themselves with the specific implementation details of the underlying knowledge base and large model.

[0026] The resource scheduling unit, based on task characteristics and resource status information fed back by the model object unit, employs large-scale model algorithms and / or reinforcement learning algorithms for multi-task joint optimization to generate a resource scheduling strategy. The resource scheduling unit is the core of the scheduling loop. It continuously monitors the occupancy and health status of antenna resources, baseband resources, and transmission resources. Combining this with task characteristics from the task management unit, it uses large-scale model algorithms to perform semantic understanding and constraint extraction of unstructured task requirements, and then uses reinforcement learning algorithms to search for the optimal resource allocation scheme under constraints. The resource scheduling strategy generated by the resource scheduling unit includes specific parameters such as antenna number, baseband channel, frequency, polarization, and transmission link for each task, and is issued to the model object unit for execution in the form of resource scheduling instructions.

[0027] The Model Object Unit (MAU) manages antenna, baseband, and transmission resources, executes resource scheduling instructions and control commands issued by the Task Management Unit, and provides feedback on the execution results. As the system's physical execution layer, the MAU maintains a real-time state database of all schedulable resources, including antenna pointing angle, rotation speed, available frequency bands, baseband processing channels, modulation schemes, coding rates, and transmission link bandwidth, latency, and bit error rate. Upon receiving a resource scheduling instruction, the MAU decomposes the instruction into specific operation commands for each hardware device and drives the devices to execute them. Upon receiving control commands from the Task Management Unit, the MAU sequentially completes operations such as antenna pointing adjustment, baseband parameter setting, and link establishment according to the process sequence. After execution, the MAU feeds back execution results, including telemetry data and link quality monitoring results, to the Resource Scheduling Unit and the Task Management Unit.

[0028] The parallel simulation unit, which integrates digital and physical data, constructs a digital twin environment based on the real operational data of the model object unit for simulation. It generates incremental simulation data and feeds it back to the intelligent model unit to drive model iteration. This unit builds a digital twin environment deeply mapped to the real measurement and control system, achieving parallel coexistence between the digital virtual space and the real physical space. Based on real measurement and control scenarios, this unit establishes a parallel virtual environment, performing virtual task deployment and closed-loop simulation during periods without actual measurement and control tasks, continuously generating incremental simulation data including normal operating conditions, boundary scenarios, and abnormal modes. After confidence verification, this incremental simulation data is fed back to the intelligent model unit to drive fine-tuning and updating of the domain-wide model, as well as training and iterating reinforcement learning or deep learning models within the resource scheduling unit and task management unit.

[0029] The human-machine interface (HMI) unit presents resource status and operational recommendations to the user and forwards user confirmation commands to the resource scheduling unit or task control unit. The HMI unit graphically displays the real-time occupancy status of ground station antennas, baseband, and transmission link resources, the on-orbit position of the target satellite and its telemetry and control visible arc, the current mission execution stage, and abnormal alarm status. During mission scheduling or anomaly handling, the HMI unit recommends the optimal operation option to the operator based on the reasoning results of the intelligent model unit and assigns a recommendation score to each recommended operation. The operator can view and confirm the recommended operation on the interface or intervene manually at key decision points. The operator's confirmation command is forwarded to the resource scheduling unit or task control unit, which then continues the subsequent process.

[0030] The resource scheduling unit forms a small scheduling loop encompassing perception, scheduling, evaluation, and optimization; the task control unit forms a measurement and control loop encompassing demand reception, task planning, process decision-making, parameter orchestration, equipment control, and result evaluation. These two closed loops collaborate through execution feedback from the model object unit and knowledge support from the intelligent model unit. The scheduling loop focuses on the dynamic allocation and optimization of resources, while the measurement and control loop focuses on the full lifecycle management of tasks. Together, they efficiently execute measurement and control tasks.

[0031] I. Detailed Composition of the Resource Scheduling Unit.

[0032] The resource scheduling unit includes a resource scheduling agent, a resource and task perception module, a target and constraint analysis module, an air interface resource scheduling module, an antenna resource scheduling module, a baseband resource scheduling module, and a decision performance evaluation module.

[0033] The resource scheduling agent is constructed using a large-scale model algorithm, providing model inference and optimization algorithm support for the resource and task perception module, the target and constraint analysis module, the air interface resource scheduling module, the antenna resource scheduling module, the baseband resource scheduling module, and the decision performance evaluation module. In complex multi-task concurrent scenarios, the resource scheduling agent comprehensively analyzes the input information from each module to assist in generating resource scheduling strategies and outputs risk warnings when resources are insufficient or conflicts cannot be resolved. The resource scheduling agent continuously accumulates experience during task execution and optimizes the scheduling strategy for subsequent tasks through feedback learning.

[0034] The resource and task awareness module receives the resource status of antenna resources, transmission resources, baseband resources, and air interface resources, as well as the requirements of pending telemetry and control tasks. Supported by the resource scheduling agent, it parses the task type, priority, time window, and required resource type, and transmits the task analysis results and resource status to the target and constraint analysis module and each resource scheduling module. The resource and task awareness module is the starting point for the scheduling loop; it monitors in real-time resource status change events reported by the model object units and new task notifications from the task management unit, formatting the perceived information and distributing it to downstream modules.

[0035] The objective and constraint analysis module utilizes the resource scheduling agent to analyze the resource combination required to complete the objective task, the expected objective effect, current resource availability, time window conflicts, and external environmental constraints. This results in a structured constraint model, which is then passed to each resource scheduling module. The constraint model is a mathematical description containing a task set, a resource set, a time window matrix, and a conflict relationship matrix, providing formalized input for subsequent optimization decisions.

[0036] The air interface resource scheduling module, antenna resource scheduling module, and baseband resource scheduling module communicate bidirectionally to exchange decision information. They employ large-scale model algorithms and / or reinforcement learning algorithms for multi-task collaborative optimization, comprehensively generating a resource scheduling scheme. These three modules are responsible for different types of resource scheduling: the air interface resource scheduling module allocates air interface parameters such as frequency, polarization, and power; the antenna resource scheduling module allocates antenna parameters such as antenna pointing, tracking strategy, and beamforming; and the baseband resource scheduling module allocates baseband processing parameters such as modulation scheme, coding rate, and channelization. The three modules achieve collaborative optimization through bidirectional information exchange. For example, the decisions of the antenna resource scheduling module influence the frequency selection of the air interface resource scheduling module, and the coding rate selection of the baseband resource scheduling module is affected by the air interface signal-to-noise ratio. Together, they constitute a multi-agent cooperative system, where each module is an independent reinforcement learning agent, achieving global optimization through a shared reward function.

[0037] The decision performance evaluation module receives the decision results and performance feedback from each resource scheduling module, performs performance evaluation to obtain a decision score level, and feeds the evaluation results back to each resource scheduling module to adjust the resource scheduling plan. The decision score level is divided into four levels: excellent, good, average, and poor. The scoring criteria include a weighted comprehensive assessment of multiple dimensions such as task fulfillment rate, resource utilization rate, and scheduling latency. When the score level is average or poor, each resource scheduling module needs to re-optimize the scheduling until the score reaches good or above, or the maximum number of retries is reached.

[0038] II. The operation of the scheduling loop includes the following steps: The first step involves the resource and task awareness module detecting a change in resource status and triggering a scheduling event. Changes in resource status include the release of an antenna after completing its previous task, the idle state of a baseband channel, the restoration of a transmission link, the arrival of a new task, or a change in task priority.

[0039] The second step involves the air interface resource scheduling module, antenna resource scheduling module, and baseband resource scheduling module re-performing multi-task joint optimization based on the current resource status and task queue, generating a new resource scheduling strategy, and distributing it to the model object unit for execution. During this multi-task joint optimization process, each resource scheduling module employs a large-scale model algorithm to perform semantic parsing and rule extraction on the natural language descriptions of the tasks, transforming unstructured task requirements into structured constraints. Then, a deep reinforcement learning algorithm is used to search for the optimal resource allocation scheme under these constraints. The state space of the deep reinforcement learning algorithm includes resource occupancy status, task queue characteristics, and constraints; the action space includes resource allocation decisions; and the reward function is constructed based on task satisfaction rate, resource utilization rate, and scheduling latency.

[0040] The third step involves the decision performance evaluation module collecting execution results and assessing the scheduling effect. If the scheduling effect falls below a preset threshold, the scheduling event is triggered again, forming a closed-loop iteration. Evaluation metrics for scheduling effect include task completion rate, resource idle rate, and average waiting time. The preset threshold can be dynamically adjusted based on historical data from system operation.

[0041] III. Detailed Composition of the Task Control Unit.

[0042] The task management unit includes a task management intelligent agent group, a requirement receiving module, a task planning module, a process decision module, a parameter orchestration module, an equipment control module, and a result evaluation module.

[0043] The task management intelligent agent swarm includes a requirement receiving agent, a task planning agent, a process decision-making agent, a parameter orchestration agent, an equipment control agent, and a result evaluation agent, providing support for each module through multi-agent collaboration. Each agent corresponds to the function of a module, shares the domain knowledge and reasoning capabilities provided by the intelligent model unit, and achieves collaborative work through a message passing mechanism.

[0044] The demand receiving module receives telemetry, tracking, and command (TT&C) task requests and parses the task type, priority, and resource requirements. As the entry point to the TT&C loop, it converts user natural language requests or structured forms into standardized task descriptions, including task identifier, target satellite, task type (e.g., tracking, data transmission, remote control), priority level, expected time window, and required resource types (e.g., antenna aperture, frequency band, bandwidth).

[0045] The task planning module generates task execution plans based on task characteristics, including time window allocation and resource configuration schemes. Combining orbital forecast data and resource availability, the module assigns specific execution time periods and pre-selected resource combinations to each task, generating a task planning list.

[0046] The process decision module is used to generate a sequence of equipment control actions based on the mission execution plan, combined with the mission management intelligent agent swarm and the aerospace telemetry and control knowledge base, thus forming the mission's operation process definition. The operation process definition is a series of ordered equipment operation instructions, such as first adjusting the antenna's azimuth angle pointing towards the target satellite, then setting the baseband receiving frequency and modulation mode, then establishing the uplink remote control link, and finally initiating data reception.

[0047] The parameter orchestration module is used to automatically orchestrate the control parameters of each measurement and control device according to the task plan and operation procedure definition, and generate a parameter configuration instruction set. The parameter configuration instruction set contains the specific configuration parameters of each device, such as the azimuth angle, elevation angle, and tracking rate of the antenna; the intermediate frequency, symbol rate, and decoding method of the baseband; and the IP address, port number, and data format of the transmission link.

[0048] The device control module, based on the parameter configuration instruction set, issues control commands to the corresponding hardware devices via the resource scheduling unit. The device control module sends the control commands sequentially to the resource scheduling unit according to the process sequence, and the resource scheduling unit forwards them to the specific hardware devices in the model object unit. This indirect issuance method ensures that the resource scheduling unit can monitor the control status of the devices in real time, facilitating subsequent resource conflict detection and scheduling adjustments.

[0049] The results evaluation module is used to acquire data returned from the device after task execution, comprehensively evaluate task completion, execution deviation, and data quality, generate task execution evaluation results, and feed them back to the requirements receiving module to support dynamic process adjustments. Evaluation results include metrics such as whether the task was successfully completed, the deviation between actual execution time and plan, and the signal-to-noise ratio and bit error rate of the received data. After being fed back to the requirements receiving module, these results can be used to adjust the planning parameters for subsequent similar tasks.

[0050] III. The operation of the measurement and control loop includes the following steps: The first step involves the requirement receiving module receiving a new task, which triggers the task planning module to generate a task execution plan. Based on task priority and time window constraints, the task planning module inserts the new task into the existing task queue and updates resource reservation information.

[0051] The second step involves the process decision module determining the control flow based on the task execution plan. The process decision module invokes the process decision agent within the task management intelligent agent group, combines the domain knowledge provided by the intelligent model unit, retrieves a standard operation process template matching the task type from the knowledge base, and adaptively adjusts it according to the current resource status and task parameters to generate a specific sequence of equipment control actions.

[0052] The third step involves the parameter orchestration module generating a parameter configuration instruction set, which the device control module then sends as control commands to the model object units for execution. Based on the operation flow definition and device interface specifications, the parameter orchestration module automatically calculates the specific parameter values ​​for each device and assembles them into a standard format command message. The device control module sends commands sequentially according to the flow sequence and waits for confirmation of each command's execution.

[0053] The fourth step involves the results evaluation module continuously monitoring the execution status during task execution. If an anomaly is detected, an anomaly handling process is triggered, generating a recommended handling strategy which is then presented to the operator for confirmation via the human-machine interface unit. Anomalies include equipment failure, link interruption, signal loss, and task timeout. The anomaly handling process begins with the intelligent model unit performing fault diagnosis to determine the anomaly type and severity level. Then, relevant agents in the task management intelligent agent group generate handling strategies, such as switching to backup equipment, adjusting parameters, or terminating the task. These handling strategies are presented as a list of recommended operations on the human-machine interface, allowing the operator to choose to adopt, modify, or reject them.

[0054] Fifth, after the task is completed, the results evaluation module performs a comprehensive evaluation and feeds the evaluation results back to the requirements receiving module for dynamic process adjustments. The comprehensive evaluation includes not only the execution quality of this task but also an assessment of the applicability of the scheduling strategy, parameter configuration, and operational procedures. The feedback information is stored in a knowledge base for optimization of similar tasks in the future.

[0055] V. Detailed Composition of the Parallel Deduction Unit of Data-Real Coexistence.

[0056] The parallel simulation unit, which integrates digital and physical systems, constructs a digital twin environment that deeply maps to the real telemetry, tracking, and command (TT&C) system, enabling the parallel coexistence of digital virtual space and real physical space. Based on real TT&C scenarios, this unit establishes a parallel virtual environment that supports various types of simulations for satellite TT&C missions, including single-satellite orbit simulations, collision warning and avoidance simulations, multi-satellite network collaborative simulations, and satellite-ground resource scheduling simulations. During periods without actual TT&C missions, the parallel simulation unit performs virtual mission deployment and closed-loop simulations within the twin environment, continuously generating incremental simulation data encompassing normal operating conditions, boundary scenarios, and abnormal modes. This data drives the fine-tuning and updating of the domain-wide model in the intelligent model unit, as well as the training and iteration of models within the resource scheduling and mission management units.

[0057] The workflow of the parallel simulation unit for digital-real coexistence includes: First, based on the real physical device parameters of the model object unit and the model parameters provided by the intelligent model unit, a digital twin synchronized with the real system is constructed, including antenna models, baseband models, transmission link models, satellite orbit models, etc. Then, various operating conditions are simulated in the digital twin environment, including normal transit tracking, multi-satellite concurrent collisions, equipment failure degradation, electromagnetic interference, and other scenarios, generating a large amount of simulation data. Finally, the simulation data, after confidence verification, is fed back to the intelligent model unit to expand the training dataset, verify the effectiveness of the algorithm, and identify potential problems.

[0058] V. Detailed Composition of Intelligent Model Units

[0059] The intelligent model unit includes a domain knowledge base construction module, a model fine-tuning module, a model lightweighting module, and a measurement and control base model module.

[0060] The domain knowledge base construction module is used to collect and process multi-source heterogeneous documents in the aerospace measurement, control, and operations (TT&O) field, constructing a domain knowledge base for TT&O. This module uses document parsing, text extraction, and semantic understanding technologies to perform format conversion and entity unification on materials such as operating procedures, equipment manuals, historical cases, and expert experience, extracting and annotating knowledge points from these documents. A knowledge graph is used to store each knowledge entity, relationship, and attribute; a corpus is built to store the original documents; and a dense vector index is constructed to achieve similar semantic search, ultimately forming a structured domain knowledge base for TT&O.

[0061] The model fine-tuning module collects operational feedback data generated during real-world measurement and control tasks, constructs a fine-tuning sample set, and employs efficient parameter fine-tuning and incremental learning strategies to optimize the inference strategy of the large model. The fine-tuning sample set includes successful cases, failed cases, anomaly handling records, and operator correction records from the task execution process. The model fine-tuning module periodically evaluates and compares the model before and after fine-tuning. When key indicators such as inference accuracy and response time reach preset thresholds, the model is updated, thus establishing a continuous evolution loop from operational feedback to model optimization.

[0062] The model lightweighting module is used to adapt large, fine-tuned models to edge environments through model compression technology. It comprehensively employs model quantization, pruning, and knowledge distillation methods to reduce the model's computational complexity and memory footprint. Model quantization converts floating-point parameters into low-bit integers, pruning removes redundant network connections, and knowledge distillation transfers knowledge from large models to smaller models. After lightweighting, the model can run in real time on resource-constrained edge devices, meeting the low-latency requirements of measurement and control systems.

[0063] The Measurement and Control Base Model Module integrates the results of domain knowledge base construction, model fine-tuning, and model lightweighting, forming a large-scale model capability base for aerospace measurement and control operations. This module provides a unified interface for domain knowledge base retrieval and question-answering services, as well as large-scale model inference and decision-making services. Resource scheduling units, task management units, and human-computer interaction units can obtain the necessary knowledge and inference results through this interface.

[0064] VI. Detailed Composition of the Human-Computer Interaction Unit.

[0065] The human-computer interaction unit includes a display and control interaction module, a resource status presentation module, an operation recommendation agent, an interface operation recommendation module, and an interactive data transmission and reception module.

[0066] The display and control interaction module supports operators in inputting measurement and control task requirements, dynamically displays the task execution process and equipment action status, displays execution results and evaluation conclusions after the task is completed, and supports visual retrieval and input of domain knowledge base. The display and control interaction module is the main interface for operator interaction with the system; it adopts a graphical user interface design and supports mouse, keyboard, and touchscreen operation.

[0067] The resource status presentation module displays the status of telemetry, tracking, and command (TT&C) resources and mission progress on a visual interface, including real-time resource occupancy, target satellite information, and mission execution stage. The module uses various chart formats, such as maps, Gantt charts, and dashboards, to display the real-time occupancy of ground station antennas, baseband, and transmission link resources, their respective mission identifiers, and idle resource periods; the target satellite's on-orbit position, attitude mode, remaining power, and TT&C visible arc; the current mission execution stage, completion percentage, and abnormal alarm status. It also supports displaying views of the relationships between resources and missions, TT&C event timelines, and historical status retrospective information.

[0068] The action recommendation agent generates recommendations for the next action based on task type, operator identity, and current execution step, combined with historical operation records and expert experience. It assigns a recommendation score to each recommended action. The recommendation score is a percentage score from 0 to 100, representing the confidence level or priority of the recommended action. The score calculation is based on a weighted average of multiple factors, including historical success rate, current context matching, and operator preferences.

[0069] The interface operation recommendation module displays a list of recommended operations on the user interface, allowing operators to confirm and execute them. It also triggers a manual confirmation and feedback mechanism at critical decision points. The recommended operation list is presented in card or menu format, with each option accompanied by a percentage indicating its recommendation level. Operators can click to view detailed information on multiple recommended operations and choose to adopt, modify, or reject them. For high-risk operations such as switching between primary and backup equipment or changing critical parameters, the system mandates manual confirmation from the operator before execution.

[0070] The interactive data transceiver module receives and forwards information data generated during human-computer interaction, distributes operation change instructions to the task management unit or model object unit, and forwards received task execution status and resource status data to the resource status presentation module and the display and control interaction module. As the data communication hub of the human-computer interaction unit, the interactive data transceiver module employs a publish-subscribe model to achieve efficient data distribution, ensuring timely and accurate information synchronization between modules.

[0071] Example 2: An intelligent scheduling and control method for aerospace telemetry and control ground stations, applied to the intelligent scheduling and control system for aerospace telemetry and control ground stations described in Example 1. For example... Figure 2 As shown, the method includes a training and deployment phase and an operation phase. The training and deployment phase is used to build the domain knowledge base, large model, intelligent agent, and scheduling algorithm required for system operation, while the operation phase is used to execute the complete closed-loop process from task reception to resource release in actual measurement and control tasks.

[0072] S1: Training and deployment phase.

[0073] The training and deployment phase includes seven steps: knowledge acquisition, knowledge modeling, knowledge verification, knowledge base construction, large model construction, agent design, and scheduling algorithm deployment, which are described in detail below.

[0074] Step S11: Knowledge Acquisition.

[0075] Multi-source heterogeneous data from the aerospace telemetry, tracking, and command (TT&C) field are collected, including equipment technical manuals, expert experience rules, TT&C schemes, mission requirement specifications, system design reports, interface control documents, and historical operation logs of traditional TT&C stations, serving as raw materials for knowledge construction. Knowledge acquisition is the foundation for building a domain knowledge base, and the comprehensiveness and accuracy of the data directly affect the quality of subsequent models. During the collection process, preliminary screening and classification of the data are required to remove duplicate and invalid data, ensuring the usability of the raw materials.

[0076] Step S12: Knowledge Modeling.

[0077] Based on ontology engineering and hybrid knowledge representation methods, the acquired standardized knowledge is formalized and structured into models. Through techniques such as relation extraction and attribute imputation, unstructured text is transformed into a knowledge representation containing conceptual hierarchy, semantic relationships, and constraint rules. A reasonable, queryable, and reusable domain knowledge system is constructed by integrating knowledge graphs, corpora, and vectorization to support real-time instruction generation and reasoning decision-making needs. The knowledge modeling results comprise three layers: a knowledge graph layer for storing structured information about entities and relations; a corpus layer for storing original documents for rapid indexing of original text paragraphs; and a dense vector index layer for enabling similar semantic search.

[0078] Step S13: Knowledge verification.

[0079] The structured knowledge modeled after knowledge modeling undergoes quality assessment using a combination of formal verification and empirical testing to ensure the correctness and adaptability of the knowledge base. Formal verification includes ontology consistency checks, logical conflict detection, and constraint rule satisfaction checks, used to identify and eliminate redundant, contradictory, or domain-axiom-violating knowledge entries. Empirical testing, based on historical measurement and control cases or expert-annotated samples, reverse-verifies the reasoning conclusions of the knowledge, evaluating its coverage and accuracy in real-world business scenarios. Only knowledge that passes verification is added to the knowledge base.

[0080] Step S14: Knowledge base construction.

[0081] Based on validated structured knowledge, a knowledge base for the aerospace telemetry, tracking, and command (TT&C) domain is established. The knowledge base supports persistent storage, incremental updates, and efficient retrieval of knowledge, providing a knowledge foundation for model training and operational reasoning. The knowledge base's storage structure includes a relational database for storing structured knowledge, a document database for storing raw corpora, and a vector database for storing semantic vector indexes. The knowledge base supports online updates; new cases and experiences generated during the operational phase can be added to the knowledge base in real time.

[0082] Step S15: Large model construction.

[0083] By leveraging structured knowledge and raw corpora from a knowledge base, a domain-specific large-scale model is pre-trained or enhanced using retrieval, resulting in a measurement and control base model that balances inference accuracy with edge deployment requirements. The large-scale model construction comprises three stages: the first stage employs retrieval-enhanced generation technology, injecting structured knowledge from the knowledge base as an external knowledge source into the large-scale model's inference process; the second stage utilizes efficient parameter fine-tuning technology, using domain-annotated data to perform domain-specific training on the large-scale model; and the third stage employs model quantization, pruning, and knowledge distillation techniques to lightweight the fine-tuned large-scale model, reducing computational complexity and memory consumption, enabling its deployment on edge computing nodes of measurement and control stations.

[0084] Step S16: Agent design.

[0085] Based on the measurement and control base model and knowledge base, intelligent agents are designed for scenarios such as resource scheduling, task management, and human-computer interaction. Their frameworks, including perception input, reasoning logic, tool invocation, and action output, are defined. The resource scheduling agent is responsible for resource status perception, constraint analysis, and scheduling decisions. The task management agent group includes a demand receiving agent, a task planning agent, a process decision-making agent, a parameter orchestration agent, an equipment control agent, and a result evaluation agent, corresponding to the six modules of the task management unit. The operation recommendation agent is responsible for generating operation recommendations based on contextual information. Each agent encapsulates specific domain knowledge and reasoning strategies, and collaborates through a message passing mechanism.

[0086] Step S17: Deploy the scheduling algorithm.

[0087] The trained base model and intelligent agents are integrated into the resource scheduling unit, task management unit, and human-computer interaction unit to complete the deployment, configuration, and joint testing of the algorithm and model in the production environment. The deployment process includes model loading, interface registration, parameter initialization, connectivity testing, and performance stress testing. Joint testing ensures that data interaction between units is normal, model inference response time meets real-time requirements, and intelligent agents work together without conflict.

[0088] The domain knowledge base, large model, intelligent agent, and scheduling algorithm formed during the training and deployment phase are all used to assist in each step of the operation phase.

[0089] S2: Operation phase.

[0090] The operation phase includes ten steps: receiving external tasks, task parsing, resource status analysis, resource scheduling, process decision-making, parameter orchestration, autonomous operation, result evaluation, exception handling, and resource release, which are described in detail below.

[0091] Step S21: Receive external tasks.

[0092] The system receives telemetry, tracking, and command (TT&C) task requests submitted by users in natural language or other formats. Through semantic understanding and intent recognition, it automatically converts unstructured descriptions into structured TT&C task requirements. These requirements cover various task types, including planned tasks, ad-hoc tasks, on-demand tasks, and emergency tasks. For natural language input, the system invokes the TT&C base model within the intelligent model unit for semantic parsing, extracting key elements such as task type, target satellite, time window, priority, and special requirements. This generates standardized task requirements, which are then passed to the next stage.

[0093] Step S22: Task analysis.

[0094] A deep analysis and requirement assessment of various task demands are conducted, utilizing the domain knowledge base constructed in step S14 as knowledge support. Combined with the orbit calculation module, the task intent and execution conditions are accurately understood. The result of task analysis is a task planning list containing elements such as task priority, time window, and resource requirements. Each task record in the task planning list includes fields such as task identifier, target satellite designation, task type, priority level, expected start and end times, required resource type and quantity, and special constraints.

[0095] Step S23: Resource status analysis.

[0096] By integrating a domain knowledge base to perform a fusion analysis of the surrounding electromagnetic environment and system status, the system collects real-time data on the occupancy and idle status of resources such as antennas, baseband, and transmission links. It also acquires information on the target spacecraft's available telemetry and control arcs, orbital parameters, and platform constraints. Combined with spectrum monitoring and environmental perception data, this generates a current system resource availability status map and conflict warning information. The result of the resource status analysis is a resource status matrix, where rows represent various resource types, columns represent time slices, and matrix elements represent the resource's occupancy status within the corresponding time slice. This step can be executed synchronously with step S21 to shorten the overall response latency.

[0097] Step S24: Resource scheduling.

[0098] Integrating telemetry and control mission requirements, resource awareness, and feedback results, and comprehensively considering various complex constraints such as antenna status, baseband processing capabilities, satellite orbit constraints, transmission bandwidth, and mission priorities, the scheduling algorithm deployed in step S17 is invoked to globally optimize various resources, including beam resources, transmission links, signal synthesis channels, and baseband boards, generating an optimal resource scheduling strategy suitable for the current mission concurrency scenario. The resource scheduling process employs a combination of large-scale model algorithms and reinforcement learning algorithms. The large-scale model algorithm is responsible for semantic parsing and rule extraction of the natural language description of the mission, transforming unstructured mission requirements into structured constraints. The reinforcement learning algorithm searches for the optimal resource allocation scheme under the constraints. Its state space includes resource occupancy status, mission queue characteristics, and constraints; its action space includes resource allocation decisions; and its reward function is constructed based on mission satisfaction rate, resource utilization rate, and scheduling latency. The output of the resource scheduling strategy is a resource allocation table, which records the specific resource identifier, usage time period, and configuration parameters allocated to each mission.

[0099] Step S25: Process Decision.

[0100] Based on the current mission type, resource scheduling results, and mission plan, and combined with the aerospace telemetry and control knowledge base, historical execution cases, and expert rules constructed in step S14, a sequence of equipment actions is automatically generated. This sequence includes operational steps such as antenna tracking switching timing, baseband operating mode switching, link establishment, and node release, forming a structured mission operation flow. The result of the flow decision is an operation flow definition file, which lists the execution time, operation object, operation content, and expected result of each operation step in chronological order.

[0101] Step S26: Parameter arrangement.

[0102] Based on equipment interface specifications, operating procedures, and task requirements, the system automatically orchestrates equipment control parameters for various measurement and control tasks. Control parameters include antenna pointing angle sequences, baseband demodulation and decoding parameters, transmission rates, and encoding methods. For scenarios with complex parameter combinations, the measurement and control base model constructed in step S15 is invoked for inference verification to ensure the rationality, compatibility, and security of the parameters. The result of parameter orchestration is a parameter configuration instruction set, where each instruction includes a device identifier, parameter name, parameter value, and execution timestamp.

[0103] Step S27: Autonomous Operation.

[0104] Once resource scheduling, process decision-making, and parameter orchestration are complete, the intelligent agent automatically distributes the instruction set to each measurement and control device, establishes the measurement and control link, and initiates task execution. During operation, it monitors device status and task progress in real time, generates confirmation prompts for key operation nodes, supports parallel task execution and dynamic resource switching, and ensures the continuity and reliability of measurement and control tasks. During autonomous operation, the system continuously monitors the status information transmitted back by the devices, including antenna pointing error, signal lock status, and data reception rate. If any deviation from expectations is detected, an adjustment mechanism is immediately triggered.

[0105] Step S28: Result evaluation.

[0106] After the task is completed, a multi-dimensional performance evaluation is conducted based on the telemetry data and link quality detection results returned by the equipment. Evaluation dimensions include task objective achievement, resource utilization efficiency, execution process deviations, and data quality. The measurement and control base model from step S15 is used to compare and analyze the execution effect, generating a task evaluation report. Typical cases and optimization experiences are fed back to the knowledge base to support continuous iterative updates of the model. The task evaluation report includes basic task information, execution timeline, resource usage statistics, abnormal event records, and improvement suggestions.

[0107] Step S29: Exception handling.

[0108] Throughout the task execution process, the system's operating status and link quality are continuously monitored. Upon detecting anomalies such as device alarms, link interruptions, or parameter exceeding limits, fault diagnosis is immediately initiated. The intelligent agent designed in step S16 is invoked to analyze the anomaly type, severity level, and impact scope, generating recommended handling strategies and intelligently presenting them to the human-machine interface, supporting rapid operator decision-making and closed-loop handling. Anomaly handling strategies include options such as fault isolation, backup equipment switching, parameter adjustment, task degradation, or task termination. Operators can view detailed information and expected effects of each recommended strategy on the human-machine interface, choosing to adopt it or develop their own handling plan.

[0109] Step S210: Resource release.

[0110] After the measurement and control mission is completed, the scheduling algorithm deployed in step S17 is invoked to perform a resource release operation, reclaiming the antenna, baseband, and transmission link resources occupied by this mission and restoring the equipment status to a schedulable state. The resource release operation includes steps such as unlocking the antenna, releasing the baseband channel, disconnecting the transmission connection, and updating the resource status database. The released resources can be promptly allocated to respond to other pending tasks, forming a closed-loop management of resource allocation, occupation, release, and reallocation.

[0111] In this embodiment, the scheduling loop and the measurement and control loop do not operate independently, but rather collaborate through execution feedback from model object units and knowledge support from intelligent model units. The scheduling loop focuses on dynamic resource allocation and optimization, with a short operating cycle, typically completing one closed-loop iteration in seconds or minutes. The measurement and control loop focuses on the full lifecycle management of tasks, with its operating cycle related to task execution duration, typically ranging from minutes to hours. The two closed loops collaborate in the following way: when the result evaluation module in the measurement and control loop detects a deviation in task execution, it feeds the evaluation result back to the scheduling loop, triggering a re-optimization of the resource scheduling strategy; when the decision performance evaluation module in the scheduling loop detects low resource utilization, it transmits optimization suggestions to the measurement and control loop, triggering dynamic adjustments to task planning. This dual-closed-loop collaborative mechanism ensures that the system maintains a highly efficient operating state in a dynamically changing measurement and control environment.

[0112] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0114] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0116] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent scheduling and control system for aerospace telemetry and control ground stations, characterized in that, include: The task management unit is used to parse the task characteristics of the measurement and control task request, send the task characteristics to the resource scheduling unit and the intelligent model unit, and issue control commands to the model object unit. The intelligent model unit is used to provide domain knowledge and reasoning support to the resource scheduling unit and the mission management unit based on the aerospace telemetry and control knowledge base and telemetry and control base model. The resource scheduling unit is used to perform multi-task joint optimization based on the task characteristics and the resource status information fed back by the model object unit, and to generate resource scheduling instructions by using a large model and / or reinforcement learning algorithm. The model object unit is used to manage antenna resources, baseband resources, and transmission resources, execute the resource scheduling instructions and control instructions issued by the task management unit, and provide feedback on the execution results. The parallel simulation unit of digital-real coexistence is used to construct a digital twin environment for simulation based on the real operating data of the model object unit, generate incremental simulation data and feed it back to the intelligent model unit to drive model iteration; The human-computer interaction unit is used to present the resource status and operation recommendations to the user, and forward the user's confirmation command to the resource scheduling unit or the task management unit. The resource scheduling unit forms a small scheduling loop of perception, scheduling, evaluation, and optimization, while the task management unit forms a measurement and control loop of demand reception, task planning, process decision-making, parameter orchestration, equipment control, and result evaluation. The two closed loops are coordinated through the execution feedback of the model object unit and the knowledge support of the intelligent model unit.

2. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 1, characterized in that, The resource scheduling unit includes a resource scheduling agent, a resource and task perception module, a target and constraint analysis module, an air interface resource scheduling module, an antenna resource scheduling module, a baseband resource scheduling module, and a decision performance evaluation module. The resource and task perception module is used to receive resource status and the requirements of the measurement and control tasks to be executed. With the support of the resource scheduling agent, it parses the task type, priority, time window and required resource type, and transmits the task analysis results and resource status to the target and constraint analysis module and each resource scheduling module. The target and constraint analysis module is used to analyze constraints using the resource scheduling agent, form a structured constraint model, and transmit it to each resource scheduling module. The air interface resource scheduling module, the antenna resource scheduling module, and the baseband resource scheduling module transmit decision information bidirectionally, and use large models and / or reinforcement learning algorithms to perform multi-task collaborative optimization and comprehensively generate a resource scheduling scheme. The decision performance evaluation module is used to receive the decision results and performance feedback results of each resource scheduling module, perform performance evaluation to obtain a decision score level, and feed back the evaluation results to each resource scheduling module to adjust the resource scheduling scheme.

3. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 2, characterized in that, The operation of the scheduling loop includes: The resource and task awareness module triggers a scheduling event after detecting a change in resource status. Each resource scheduling module re-optimizes the multi-task joint operation based on the current resource status and task queue, generates a new scheduling strategy, and sends it to the model object unit for execution. The decision performance evaluation module collects the execution results and evaluates the scheduling effect. If the scheduling effect is lower than the preset threshold, the scheduling event is triggered again to form a closed loop iteration.

4. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 2, characterized in that, The resource scheduling agent is constructed using a large model algorithm, providing model reasoning and optimization algorithm support for each module. The resource scheduling agent comprehensively analyzes the input information of each module in complex scenarios with multiple concurrent tasks, assists in generating resource scheduling strategies, and outputs risk warnings when resources are insufficient or conflicts cannot be resolved.

5. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 1, characterized in that, The task management unit includes a task management intelligent agent group, a requirement receiving module, a task planning module, a process decision module, a parameter arrangement module, an equipment control module, and a result evaluation module. The requirement receiving module is used to receive measurement and control task requests and parse the task type, priority and resource requirements. The task planning module is used to generate a task execution plan based on task characteristics, including time window allocation and resource configuration scheme; The process decision module is used to generate a sequence of equipment control actions based on the task execution plan, combined with the task management intelligent agent group and the knowledge base in the aerospace telemetry and control field, to form the task operation process definition. The parameter arrangement module is used to automatically arrange the control parameters of each measurement and control device according to the task planning and operation process definition, and generate a parameter configuration instruction set; The device control module is used to send control commands to the corresponding hardware devices via the resource scheduling unit according to the parameter configuration instruction set. The result evaluation module is used to obtain data returned by the device after the task is completed, to comprehensively evaluate the task completion, execution deviation and data quality, generate task execution evaluation results and feed them back to the demand receiving module to support dynamic adjustment of the process.

6. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 5, characterized in that, The operation of the measurement and control loop includes: After receiving a new task, the demand receiving module triggers the task planning module to generate a task execution plan. The process decision module determines the control process based on the task execution plan; The parameter orchestration module generates a parameter configuration instruction set, and the device control module sends the parameter configuration instruction set as a control instruction to the model object unit for execution. The result evaluation module continuously monitors the execution status during task execution. If an anomaly is detected, an anomaly handling process is triggered, a recommended handling strategy is generated, and the strategy is presented to the operator for confirmation via the human-computer interaction unit. After the task is completed, the result evaluation module performs a comprehensive evaluation and feeds back the evaluation results to the requirement receiving module for dynamic process adjustment.

7. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 1, characterized in that, The parallel simulation unit for digital-real coexistence is used to construct a digital twin environment that is deeply mapped to the real measurement and control system, so as to realize the parallel coexistence of digital virtual space and real physical space; The parallel simulation unit, which is based on real measurement and control scenarios, establishes a parallel virtual environment. During periods without actual measurement and control tasks, it performs virtual task deployment and closed-loop simulation, continuously generating incremental simulation data that includes normal operating conditions, boundary scenarios, and abnormal modes. This data is used to drive the fine-tuning and updating of the domain-wide model in the intelligent model unit, as well as the training and iteration of the models within the resource scheduling unit and the task management unit.

8. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 1, characterized in that, The intelligent model unit includes a domain knowledge base construction module, a model fine-tuning module, a model lightweighting module, and a measurement and control base model module; The domain knowledge base construction module is used to collect and process multi-source heterogeneous documents in the aerospace measurement, control and operation field to build a knowledge base for the aerospace measurement, control and operation field. The model fine-tuning module is used to collect operation and maintenance feedback data generated in real measurement and control tasks, construct a fine-tuning sample set, and use efficient parameter fine-tuning and incremental learning strategies to fine-tune and optimize the inference strategy of the large model. The model lightweighting module is used to adapt a finely tuned large model to the edge environment through model compression technology, and comprehensively adopts model quantization, pruning and knowledge distillation methods to reduce the model's computational complexity and memory usage; The measurement and control base model module is used to integrate the results of domain knowledge base construction, model fine-tuning, and model lightweighting to form a large model capability base for aerospace measurement and control services.

9. The intelligent scheduling and control system for aerospace telemetry and control ground stations according to claim 1, characterized in that, The human-computer interaction unit includes a display and control interaction module, a resource status presentation module, an operation recommendation agent, an interface operation recommendation module, and an interactive data transmission and reception module. The display and control interaction module is used to support operators in inputting measurement and control task requirements, dynamically displaying the task execution process and equipment action status, displaying the execution results and evaluation conclusions after the task is completed, and supporting visual retrieval and input of domain knowledge base. The resource status presentation module is used to display the status of telemetry and control resources and the status of mission progress on a visual interface, including real-time resource occupancy, target satellite information and mission execution stage; The operation recommendation agent is used to generate recommendations for the next operation steps based on the task type, operator identity, and current execution steps, combined with historical operation records and expert experience, and to assign a recommendation score to each recommended operation. The interface operation recommendation module is used to display a list of recommended operations on the operation interface, support operators to confirm and execute them, and trigger a manual confirmation and feedback mechanism at key decision nodes. The interactive data transceiver module is used to receive and forward information data generated during human-computer interaction, distribute operation change instructions to the task management unit or the model object unit, and forward the received task execution status and resource status data to the resource status presentation module and the display and control interaction module.

10. A method for intelligent scheduling and control of aerospace telemetry and control ground stations, characterized in that, The system is applied to an intelligent scheduling and control system for aerospace telemetry and control ground stations as described in any one of claims 1-9, including a training and deployment phase and an operation phase. The training deployment phase includes: Collect multi-source heterogeneous data in the aerospace measurement, tracking, and control field, and build a knowledge base for the aerospace measurement, tracking, and control field; The knowledge base is used to pre-train or retrieve large domain models to enhance the generation of adaptations, and the measurement and control base model is formed through model fine-tuning and lightweight processing. Based on the measurement and control base model and knowledge base, an intelligent agent is designed for resource scheduling, task management and human-computer interaction scenarios; The trained base model and intelligent agent are integrated into each unit to complete the deployment and configuration; The operational phase includes: Receive telemetry and control task requests and parse task characteristics to generate a task planning list; Real-time collection of resource status information to generate a status map of available resources; By integrating task requirements and resource status, a resource scheduling strategy is generated by invoking a scheduling algorithm. Automatically generate task operation procedures and arrange equipment control parameters based on resource scheduling results and domain knowledge base; The instruction set is sent to the measurement and control equipment to establish a measurement and control link and initiate task execution. The status is continuously monitored during task execution, and when an anomaly is detected, a recommended handling strategy is generated and presented to the human-computer interaction interface. After the task is completed, a performance evaluation is performed, and the evaluation results are fed back to the knowledge base; After the task is completed, a resource release operation is performed to restore the device to a schedulable state.