Satellite load automatic scheduling method and system based on AI

Abstract

The embodiment of the invention provides an AI-based satellite load automatic scheduling method and system, and the method comprises the steps: inputting satellite load data into an intelligent decision model after obtaining the satellite load data, so as to obtain task load distribution data outputted by the intelligent decision model. And generating a scheduling command set according to the task load distribution data, and sending the scheduling command set to a satellite control system, so that the satellite control system schedules the satellite load according to the scheduling command set. According to the method, efficient configuration of satellite load resources can be realized through an intelligent decision-making process, and the task execution efficiency is improved. Through accurate task demand analysis and load matching, it is ensured that tasks can be efficiently and accurately completed, the reliability of the system is improved, and the problem that satellite loads are difficult to effectively schedule is solved.

Description

Technical Field

[0001] The present application relates to the field of aerospace intelligent technology, and in particular to an AI-based satellite payload automated scheduling method and system. Background Art

[0002] Artificial satellites mainly include power supply system, attitude control system, navigation system, remote sensing monitoring system, control system, communication system, structural system and thermal control system. Through the coordination of multiple systems, the satellite is controlled to operate in the space environment according to the preset trajectory and operation mode.

[0003] According to the purpose of artificial satellites, artificial satellites carry certain payloads. The payload of artificial satellites refers to the instruments, equipment, systems and data processing resources that directly complete the satellite's specific tasks. Taking the payload required for information acquisition tasks as an example, when performing earth observation, satellites need to carry sensors such as high-definition cameras to perform earth observation, and certain data processing resources need to be allocated in the control system to process earth observation data.

[0004] Since it is difficult to upgrade the hardware of artificial satellites after launch, the data processing capabilities of artificial satellites are almost not scalable. Therefore, how to schedule satellite payloads so that satellites can make full use of current resources has become a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] In view of this, an embodiment of the present application provides an AI-based satellite payload automatic scheduling method and system to solve the problem that satellite payloads are difficult to effectively schedule.

[0006] According to one aspect of the present application, a satellite payload automation scheduling method based on AI is provided, the method comprising:

[0007] Acquiring satellite payload data, wherein the satellite payload data includes real-time status information, mission requirement information, and resource constraint information;

[0008] The satellite payload data is input into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model, wherein the intelligent decision model is a machine learning model trained using payload sample data; the payload sample data includes a payload sample and a payload allocation label annotated for the payload sample; the intelligent decision model is configured to perform payload matching according to the satellite payload data;

[0009] generating a scheduling command set according to the task load distribution data;

[0010] The scheduling command set is sent to a satellite control system, so that the satellite control system schedules the satellite payload according to the scheduling command set.

[0011] Optionally, after acquiring the satellite payload data, the method further includes:

[0012] Reading the real-time status information, the mission requirement information and the resource constraint information from the satellite payload data;

[0013] Generate a task requirement item according to the task requirement information, wherein the task requirement item includes at least one of a task type, a task priority, and a task execution time;

[0014] Generate a current status item according to the real-time status information, wherein the current status includes at least one of resolution, coverage, and energy consumption;

[0015] The task analysis item and the load status item are set to the intelligent decision-making model as constraint conditions of the intelligent decision-making model.

[0016] Optionally, after acquiring the satellite payload data, the method further includes:

[0017] Generate a resource constraint item according to the resource constraint information, wherein the resource constraint item includes at least one of load working time, load power consumption, and storage space;

[0018] generating an objective function according to the resource constraint item, wherein the dependent variable of the objective function is the utilization rate of the resource constraint item;

[0019] The objective function is set to the intelligent decision-making model.

[0020] Optionally, after acquiring the satellite payload data, the method further includes:

[0021] Performing data cleaning on the satellite payload data according to preset cleaning items, wherein the cleaning items include one or more combinations of data deduplication, missing value supplementation, outlier processing, and standardization;

[0022] The satellite payload data after data cleaning is converted into a target format, wherein the target format is a data format set according to an input neural network node of the intelligent decision-making model.

[0023] Optionally, the method further includes:

[0024] Receive task instructions;

[0025] In response to the task instruction, parsing the task requirement information from the task instruction;

[0026] A task requirement report is generated according to the task requirement information.

[0027] Optionally, inputting the satellite payload data into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model includes:

[0028] Extracting task-related features and load-related features from the satellite payload data, wherein the task-related features are semantic feature vectors extracted from the task requirement information; and the load-related features are semantic feature vectors extracted from the resource constraint information;

[0029] matching a suitable load based on the task-related characteristics and the load-related characteristics;

[0030] An association relationship between the suitable load and the task-related feature is established to generate the task load distribution data.

[0031] Optionally, matching a suitable load based on the task association feature and the load association feature includes:

[0032] Extracting state-related features from the satellite payload data, wherein the state-related features are semantic feature vectors extracted from the real-time state information;

[0033] Determine an optional load set and a non-optional load set based on the state association feature and the load association feature;

[0034] The suitable load is extracted from the optional load set according to the task association feature.

[0035] Optionally, the method further includes:

[0036] Obtaining load sample data and calling a trained model, wherein the machine learning neural network structure of the trained model is the same as that of the intelligent decision-making model;

[0037] Inputting the load sample data into the trained model to obtain training load distribution data output by the trained model;

[0038] Calculating training loss according to the training load distribution data and the load distribution label;

[0039] Iteratively optimizing model parameters of the trained model based on the training loss;

[0040] When the training loss reaches a preset loss requirement, the model parameters are output to obtain the intelligent decision-making model.

[0041] Optionally, the method further includes:

[0042] Obtaining scheduling result data, wherein the scheduling result data includes task execution progress and external environment parameters;

[0043] Extracting an optimized feature set from the scheduling result data based on an autonomous learning algorithm, wherein the optimized feature set includes one or more combinations of clustering features, dimensionality reduction features, and association rule features;

[0044] Using the optimized feature set to evaluate the task load distribution data to obtain evaluation result data;

[0045] Adjust the model parameters of the intelligent decision-making model according to the evaluation result data.

[0046] According to another aspect of the present application, there is provided an AI-based satellite payload automation scheduling system, the system comprising:

[0047] A payload resource management module, used to obtain satellite payload data, wherein the satellite payload data includes real-time status information, mission requirement information, and resource constraint information;

[0048] An intelligent scheduling algorithm module is used to input the satellite payload data into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model, wherein the intelligent decision model is a machine learning model trained using payload sample data; the payload sample data includes a payload sample and a payload allocation label annotated for the payload sample; the intelligent decision model is configured to perform payload matching according to the satellite payload data;

[0049] A command generation module, used to generate a scheduling command set according to the task load distribution data;

[0050] The execution and monitoring module is used to send the scheduling command set to the satellite control system so that the satellite control system schedules the satellite payload according to the scheduling command set.

[0051] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned AI-based satellite payload automatic scheduling method when executing the program.

[0052] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned AI-based satellite payload automatic scheduling method is implemented.

[0053] By means of the above technical solution, the embodiment of the present application provides an AI-based satellite payload automatic scheduling method and system. After acquiring the satellite payload data, the method can input the satellite payload data into an intelligent decision-making model to obtain the task payload allocation data output by the intelligent decision-making model. Then, a scheduling command set is generated according to the task payload allocation data, and the scheduling command set is sent to the satellite control system so that the satellite control system schedules the satellite payload according to the scheduling command set. The method can realize the efficient configuration of satellite payload resources and improve the efficiency of task execution through an intelligent decision-making process. Through precise task requirement analysis and payload matching, it is ensured that the task can be completed efficiently and accurately, the reliability of the system is improved, and the problem of difficult effective scheduling of satellite payloads is solved.

[0054] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0056] Figure 1 A schematic diagram of the ground monitoring center and satellite communication structure provided in an embodiment of the present application is shown;

[0057] Figure 2 A schematic diagram of a flow chart of an AI-based satellite payload automation scheduling method provided in an embodiment of the present application is shown;

[0058] Figure 3 A schematic diagram of a satellite payload scheduling process provided by an embodiment of the present application is shown;

[0059] Figure 4 A schematic diagram of the intelligent decision-making model training process provided in an embodiment of the present application is shown;

[0060] Figure 5 A schematic diagram of a load matching process provided in an embodiment of the present application is shown;

[0061] Figure 6 A schematic diagram of the process of setting model constraints and objective functions provided in an embodiment of the present application is shown;

[0062] Figure 7 A schematic diagram of the autonomous learning process of the model provided in the embodiment of the present application is shown;

[0063] Figure 8A schematic diagram of the structure of an AI-based satellite payload automation scheduling system provided in an embodiment of the present application is shown;

[0064] Fig. 9 A schematic diagram of the computer device structure provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0065] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0066] In the embodiment of the present application, the satellite refers to an artificial satellite, which may be an artificial earth satellite, an artificial external celestial satellite, an artificial space vehicle, etc. The satellite may include a power supply system, an attitude control system, a navigation system, a remote sensing monitoring system, a control system, a communication system, a structural system, and a thermal control system. Through the coordination between multiple systems, the satellite is controlled to operate in a space environment according to a preset trajectory and operation mode.

[0067] Among them, the power supply system may include solar panels (solar wings) and batteries. Solar panels can convert solar energy into electrical energy. Batteries are used to store electrical energy. The attitude control system includes attitude sensors, reaction wheels, thrusters, etc., which are used to determine the direction of the satellite in space through the attitude sensor, change the rotation of the satellite using the reaction wheels, and use thrusters to fine-tune the position and attitude of the satellite.

[0068] The navigation system includes GPS receivers and star sensors, which are used to determine the precise position of satellites and the orientation of satellites by identifying star maps. The remote sensing monitoring system includes remote sensing monitoring sensors, such as cameras, multispectral sensors, radar systems, etc., which are used to capture images of specific areas and collect spectral data in different spectral ranges.

[0069] The communication system includes communication antennas and transponders for sending and receiving signals, as well as signal processing during the sending and receiving process. The structural system includes mechanical structures such as satellite shells and solar panel brackets, which are used to protect internal equipment and provide support for other systems.

[0070] The control system may include an on-board computer (OBC) or other device or module with data processing capabilities, which is used to control the overall operation process of the satellite, and collect, process, store, and send data generated by the navigation system, remote sensing monitoring system, and communication system in the satellite. Therefore, in some embodiments, the on-board computer corresponding to the satellite control system may include a processor, a memory, an electronic function module, a communication interface, a command interface, a signal processing module, a memory, and the like.

[0071] The satellite payload refers to the instruments, equipment, systems and data processing resources that directly complete the satellite's specific mission. For example, when a satellite performs earth observation, it needs to carry sensors such as high-definition cameras to perform earth observation, and it is necessary to allocate certain data processing resources in the control system, such as memory space, CPU processing power, storage space, etc., for processing earth observation data. Based on this, for earth observation tasks, the satellite payloads that need to be allocated include image sensors and data processing resources for image collection, processing, and transmission.

[0072] It should be noted that the satellite payload may be one or more combinations of instruments, equipment, systems, and data processing resources. Since most instruments in the satellite have been set with specific executable tasks before launch, in some embodiments of the present application, the satellite payload may specifically refer to data processing resources. The corresponding satellite payload scheduling is used to refer to the scheduling of satellite data processing resources.

[0073] After being launched into the space environment, the satellite can also maintain communication connection with the ground monitoring center. The ground monitoring center can be deployed with output processing equipment, and the data processing equipment of the ground monitoring center can have strong data processing capabilities. Therefore, Figure 1 As shown in the figure, the ground monitoring center can obtain the data collected by the satellite through the communication connection, and perform complex operations such as storage, analysis, and conversion on the data collected by the satellite. The ground monitoring center can also send control instructions to the satellite according to actual control needs to control the operating status of the satellite. For example, when a new task needs to be issued to the satellite, the ground monitoring center can send a task instruction to the satellite. After receiving the task instruction, the satellite can execute the corresponding load scheduling according to the task instruction to realize the functions related to the new task.

[0074] During the operation of the satellite, the payload can be scheduled according to the specific tasks to be performed. Among them, payload scheduling refers to the process of rationally arranging and managing the payload in the satellite mission. Payload scheduling can ensure that the satellite can efficiently perform tasks in its orbit and maximize the functions and performance of the satellite. In some embodiments, when performing payload scheduling, the satellite needs to perform mission planning, that is, formulate a payload usage plan based on the satellite's mission requirements and mission priorities. The usage plan may include when and how to use different payload equipment to meet the needs of scientific observations, communications and other tasks. Satellites also need to manage resources. That is, during the scheduling process, it is necessary to consider the satellite's resource limitations, such as power, bandwidth and data storage, to ensure that the use of the payload does not exceed these limitations.

[0075] When performing payload scheduling, the satellite can also perform time scheduling. By reasonably arranging the working time of the payload, it ensures that data collection or signal transmission is performed at the best time to improve the effectiveness of the mission and data quality. In addition, for tasks involving data processing and transmission, the acquired data needs to be processed and transmitted to ensure that the data can be transmitted back to the ground monitoring center in time for analysis and utilization. In addition, when performing payload scheduling, the satellite also needs to dynamically adjust the payload scheduling according to changes in actual conditions, such as unexpected situations or new mission requirements, to cope with changes in actual conditions.

[0076] In some embodiments, payload scheduling can be performed by the satellite's control system, that is, when payload scheduling needs to be performed, the current operating status is obtained through the onboard computer, and mission planning, resource management, time scheduling, data processing and transmission, dynamic adjustment, etc. are performed according to the current operating status.

[0077] In some embodiments, payload scheduling can be performed by a ground monitoring center, that is, when payload scheduling is required, the ground monitoring center sends a control instruction to the satellite to control the satellite to feed back the current operating status related data to the ground monitoring center. The ground monitoring center then performs mission planning, resource management, time scheduling, data processing and transmission, dynamic adjustment, etc. based on the current operating status of the satellite. The scheduling control command is then generated, and the scheduling control command is sent to the satellite control system, so that the satellite performs payload scheduling according to the scheduling control command.

[0078] In some embodiments, the payload scheduling can also be jointly performed by the satellite control system and the ground monitoring center. That is, when the payload scheduling needs to be performed, it can be determined which scheduling links are performed by the satellite control system and which scheduling links are performed by the ground monitoring center according to the complexity of data processing, data processing volume, data transmission mode and hardware configuration during the scheduling process, so as to make full use of the data processing capabilities of the satellite control system and the ground monitoring center and improve the payload scheduling efficiency.

[0079] It should be noted that, based on the payload scheduling execution process shown in the above embodiments, the payload scheduling process can be executed by a satellite control system and / or a ground monitoring center. Accordingly, in the following embodiments, the AI-based satellite payload automation scheduling method can also be executed by a satellite control system and / or a ground monitoring center. For ease of description, in some embodiments, the execution subject of the AI-based satellite payload automation scheduling method is collectively referred to as a scheduling system.

[0080] It should be understood that the scheduling system can be implemented by running relevant applications through the satellite control system, or by running relevant applications through the ground monitoring center, or by the satellite control system cooperating with the ground monitoring center to run relevant applications. Therefore, from the perspective of hardware equipment, the steps corresponding to the AI-based satellite payload automatic scheduling method can be executed by the satellite control system, or by the ground monitoring center, or some steps can be executed by the satellite control system and some steps can be executed by the ground monitoring center.

[0081] In order to achieve effective scheduling of satellite payloads, the present application embodiment provides an AI-based satellite payload automatic scheduling method, such as Figure 2 , Figure 3 As shown, the method includes:

[0082] S101. Obtain satellite payload data.

[0083] When payload scheduling is required, the scheduling system can obtain satellite payload data. Satellite payload data can be collected by sensors carried by the satellite or by a monitoring module inside the satellite control system. The scheduling system can be configured with a payload resource management module, which is used to collect satellite payload data.

[0084] Satellite payload data may include real-time status information, mission requirement information, and resource constraint information. Among them, real-time status information is used to characterize the real-time operating status of the satellite. Real-time status information may include payload status and operating status. The payload status refers to the working status of the payload (such as sensors, cameras, etc.), data collection status, current load status of the control system, etc. The payload status can be used to determine whether the current payload is working normally, the frequency and quality of data collection, data processing capabilities, etc. The operating status may include position and orbit information, speed and motion status, health status, communication status, environmental monitoring data, power status, fault and abnormality reports, etc. The payload status can be used to determine the current operating status of the satellite, thereby realizing satellite operation management, mission execution, and troubleshooting.

[0085] Task requirement information refers to information used to characterize the load requirements of the tasks involved in load scheduling when performing load scheduling. That is, various information and data required in a specific decision-making process. Task requirement information can help the decision-making module understand the goals, scope and specific requirements of the task. In some embodiments, the task requirement information may include goal definition, functional requirements, constraints, environmental information, data requirements, priority and urgency.

[0086] The task requirement information may specify specific content through a task instruction, that is, in some embodiments, the scheduling system may receive a task instruction and, in response to the task instruction, parse the task requirement information from the task instruction. In order to facilitate the acquisition of the task requirement information, after parsing the task requirement information, a task requirement report may be generated according to the task requirement information.

[0087] Resource constraint information refers to information used to characterize the payload conditions in the satellites that can be scheduled by the scheduling task when performing payload scheduling. That is, the limited resource restrictions that need to be considered in the decision-making process, and resource constraint information will affect the implementation and results of the decision. In some embodiments, resource constraint information may include payload working time, payload power consumption, storage space, etc.

[0088] Satellite payload data can be obtained by receiving sensor signals, collecting operation reports, parsing mission instructions, etc. For satellite payload data from different sources, the specific data representation is also different. For example, for sensor signals, it can be analog or digital signals; for satellite payload data extracted from operation reports, it can be text data; for satellite payload data parsed from mission instructions, it can be natural language text, etc.

[0089] Different types of satellite payload data need to be converted into the same form when performing payload scheduling, that is, in some embodiments, after obtaining the satellite payload data, the scheduling system also needs to pre-process the satellite payload data to ensure data quality. After obtaining the satellite payload data, the scheduling system can perform data cleaning on the satellite payload data according to preset cleaning items. The cleaning items include one or more combinations of data deduplication, missing value supplementation, outlier processing, and standardization.

[0090] For example, after acquiring satellite payload data, the satellite payload data can be converted into text form, and the errors and inconsistencies in the data can be checked and then corrected, including processing missing values, outliers, duplicate records, etc.

[0091] After data cleaning, duplicate records in satellite payload data can be removed, missing data can be supplemented, and abnormal data can be corrected to obtain satellite payload data without missing values, abnormal values, and duplicate records. The satellite payload data after data cleaning is then converted into a target format. The target format is a data format set according to the input neural network node of the intelligent decision model. For example, through format conversion, the range or distribution of satellite payload data can be adjusted to a specified range or form to better meet the input requirements of the model and improve the performance of the model, such as minimum-maximum normalization, Z-score normalization, logarithmic transformation, etc.

[0092] S102: Input the satellite payload data into an intelligent decision-making model to obtain mission payload allocation data output by the intelligent decision-making model.

[0093] After acquiring the satellite payload data, the scheduling system can input the satellite payload data into the intelligent decision model to make intelligent scheduling decisions. The intelligent decision model can integrate advanced AI algorithms (such as deep learning, reinforcement learning, etc.) to make intelligent scheduling decisions based on mission requirements and payload resource evaluation results. That is, the intelligent decision model is configured to perform payload matching based on the satellite payload data.

[0094] The intelligent decision model is a machine learning model obtained by training with payload sample data. Before inputting satellite payload data into the intelligent decision model, model training is required to obtain an intelligent decision model that meets the output accuracy requirements. Model training can be completed by the onboard computer in the satellite control system, or by the data processing equipment of the ground monitoring center. It can also be first performed by the ground monitoring center for general model training and then sent to the onboard computer for personalized training. In the embodiment of the present application, there is no specific restriction on the executor of model training.

[0095] In order to train and obtain the intelligent decision-making model, Figure 4 As shown, in some embodiments, the training device can obtain load sample data and call the trained model. The load sample data includes a load sample and a load distribution label marked for the load sample. For example, in order to obtain load sample data, the training device can first collect data before executing model training, and collect data related to load distribution as training resources. The load sample data can come from the load distribution process corresponding to each mission instruction during the operation of the satellite. That is, when executing the mission instruction, the mission requirement information, real-time status information and resource constraint information can be recorded as a load sample. And monitor the satellite's payload distribution, and record the distribution data with a good payload distribution status as a load distribution label to obtain the load sample data.

[0096] After collecting data, the training device can also clean the collected data, delete duplicate, abnormal or irrelevant data, and ensure data quality. Then perform data annotation, that is, add labels to the cleaned data to obtain a load sample data set. Then divide the load sample data set into a training set, a validation set, and a test set. Among them, the training set is used to train the model, the validation set is used to adjust hyperparameters and monitor model performance, and the test set is used to finally evaluate the model effect.

[0097] When performing model training, the training device also needs to select the trained model according to the decision-making needs. The machine learning neural network structure of the trained model is the same as the intelligent decision-making model. For example, according to the task requirements, select a suitable neural network structure, such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), etc. After selecting the trained model, it is also necessary to determine the model parameters and hyperparameters, such as the number of nodes, learning rate, batch size, number of iterations, etc.

[0098] The load sample data is input into the trained model to obtain the training load distribution data output by the trained model. The training loss is calculated according to the training load distribution data and the load distribution label. And the model parameters of the trained model are iteratively optimized based on the training loss. When the training loss reaches the preset loss requirement, the model parameters are output to obtain the intelligent decision model.

[0099] After obtaining the load sample data, the load sample data can be input into the trained model, that is, forward propagation is performed based on the load sample data, so that the input load sample data passes through the calculation of each layer of the neural network, and finally the output result is obtained. Then based on the loss function, the difference between the output result and the true label is calculated, that is, the training loss. Among them, the loss function can calculate the training loss based on the mean square error (MSE), cross entropy (Cross-Entropy), etc. After calculating the training loss, back propagation can be performed based on the training loss, that is, according to the gradient of the loss function, the weights and biases in the network are updated through the chain rule. Then use optimization algorithms such as Gradient Descent and Adam to adjust the network parameters during the training process to minimize the loss function.

[0100] In some embodiments, after the training loss is calculated, the training loss can be compared with a preset loss threshold. If the training loss is greater than the loss threshold, it means that the current trained model has not converged, and the difference between the output result and the true label is large, so back propagation can be performed based on the training loss, and the model parameters of the trained model can be adjusted through iterative training. If the training loss is less than or equal to the loss threshold, it means that the current trained model has converged, and the difference between the output result and the true label is small, so the model parameters of the trained model can be output to obtain an intelligent decision model.

[0101] After training and obtaining the intelligent decision model, you can match the payload based on the intelligent decision model. The intelligent decision model can analyze the task requirements in detail, including task type, priority, execution time, etc. And select the most appropriate payload for matching according to the task requirements. Consider the resolution, coverage, energy consumption and other characteristics of the payload to ensure that the task can be completed efficiently and accurately.

[0102] like Figure 5 As shown, in some embodiments, when performing payload matching, the intelligent decision model can extract task-related features and payload-related features from the satellite payload data. The task-related features are semantic feature vectors extracted from the task requirement information; and the payload-related features are semantic feature vectors extracted from the resource constraint information.

[0103] The intelligent decision-making model built into the scheduling system may include a feature extraction network, which can extract features from satellite payload data. The feature extraction network can extract task-related features from task requirement information. For example, when the task requirement information is text information, the feature extraction network can extract keywords used to describe the task content from the task requirement information based on a semantic recognition algorithm. For example, for the task instruction of "Earth Observation Mission", keywords such as "observation" used to indicate the task type and "ground" used to indicate the task coverage can be extracted. The extracted keywords are then combined into a feature vector to obtain task-related features. Similarly, the feature extraction network can also extract payload-related keywords from resource constraint information to form payload-related features.

[0104] After feature extraction, the scheduling system matches the appropriate load based on the task-related features and the load-related features, and establishes an association relationship between the appropriate load and the task-related features to generate the task load allocation data. In order to match the appropriate load, the scheduling system can extract state-related features from the satellite payload data when matching the appropriate load based on the task-related features and the load-related features. The state-related features are semantic feature vectors extracted from the real-time state information. Based on the state-related features and the load-related features, the optional load set and the non-optional load set are determined. And according to the task-related features, the appropriate load is extracted from the optional load set.

[0105] For example, after feature extraction, load matching can be performed based on task-related features, load-related features, and state-related features, so that each task is assigned a suitable load and the effective load is utilized to the maximum extent.

[0106] When performing payload matching, the intelligent decision model can use the feature extraction and pattern recognition capabilities of deep learning to optimize scheduling decisions. For example, the intelligent decision model can model the problem based on the satellite payload allocation decision process, and can use modeling methods such as Markov Decision Process (MDP) to model the scheduling problem as a sequential decision problem. Then, a state representation method is designed based on the satellite payload allocation decision process to capture real-time state information. A reward function is designed to guide the agent to learn an effective scheduling strategy, where the reward function can be based on machine idle time or other performance indicators.

[0107] like Figure 6 As shown, in some embodiments, when performing payload matching, the constraint conditions and objective function of the intelligent decision model can be set according to the satellite payload data. That is, the scheduling system can read the real-time status information, the task requirement information and the resource constraint information from the satellite payload data. And generate task requirement items according to the task requirement information, generate current status items according to the real-time status information, and generate resource constraint items according to the resource constraint information.

[0108] The task requirement item includes at least one of the task type, task priority and task execution time. The current state includes at least one of the resolution, coverage and energy consumption. The resource constraint item includes at least one of the load working time, load power consumption and storage space.

[0109] Then, the task analysis item and the load status item are set to the intelligent decision model as constraints of the intelligent decision model. And an objective function is generated according to the resource constraint item. The dependent variable of the objective function can be the utilization rate of the resource constraint item, which is used to maximize the utilization of resources, so that the objective function is set to the intelligent decision model.

[0110] For example, set the task analysis item and load status item as the constraints of the intelligent decision-making model, and set the objective function of the intelligent decision-making model according to the resource constraint item. Then select the appropriate deep reinforcement learning algorithm, such as Proximal Policy Optimization (PPO), Deep Q-Network (DQN), etc., to select the optimal action for scheduling according to the current state. In the scheduling model, multiple objectives need to be optimized at the same time, such as execution time, cost, fairness and continuity.

[0111] S103: Generate a scheduling command set according to the task load distribution data.

[0112] After the mission load allocation data is output through the intelligent scheduling model, the scheduling system can perform command conversion based on the mission load allocation data, that is, generate scheduling commands based on the mission load allocation data. Since the satellite needs to perform multi-objective optimization based on multiple tasks when performing tasks, the mission load allocation data can include multiple tasks and the loads allocated to multiple tasks. For each task in the scheduling process, a scheduling command can be generated, and for multiple tasks, multiple scheduling commands can be generated to form a scheduling command set.

[0113] For example, the task load allocation data may include: Task 1, Earth observation task; Task 2, data transmission task. The corresponding allocated payload is: Task 1 uses camera equipment, communication equipment, observation data processing process, and data transmission process; Task 2 uses communication equipment and data transmission process. Then at least two scheduling commands can be generated, namely, Command 1, which is used to call the camera equipment, communication equipment, observation data processing process, and data transmission process; Command 2, which is used to schedule the communication equipment and data transmission process.

[0114] S104: Send the scheduling command set to a satellite control system, so that the satellite control system schedules the satellite payload according to the scheduling command set.

[0115] After generating the scheduling command set, the scheduling system can automatically generate a scheduling plan based on the output of the intelligent decision model. The scheduling plan is sent to the satellite control system to realize the automatic scheduling and execution of the payload. After obtaining the scheduling command set, the satellite control system can parse the scheduling commands from the scheduling command set, and then execute the scheduling commands separately to realize the scheduling of the payload.

[0116] It can be seen that by applying the technical solutions provided in the above embodiments, the scheduling system can achieve efficient configuration of satellite payload resources and improve task execution efficiency through intelligent decision-making processes. And through accurate task requirement analysis and payload matching, it is ensured that tasks can be completed efficiently and accurately, and the reliability of the system is improved. In addition, the system design is flexible and can be expanded and upgraded according to actual needs to adapt to more complex and diverse task requirements in the future.

[0117] The AI-based satellite payload automatic scheduling method provided in the above embodiment can adapt to different application scenarios. For example, the method can be applied to environmental monitoring, that is, to observe and analyze the atmosphere, water resources, soil pollution, etc. through satellite payloads, and provide scientific support for environmental governance and protection. The method can also be applied to urban management, that is, to use satellite payloads to achieve planning, monitoring and evaluation of urban construction and management, and provide accurate information support for urban development. The method can also be applied to agricultural monitoring, through satellite payloads to monitor and analyze the growth status of crops and land use in real time, and provide scientific reference for agricultural production and food security. The method can also be applied to disaster monitoring, using satellite payloads to monitor and warn of natural disasters in real time, improve disaster response capabilities, and reduce disaster losses.

[0118] Furthermore, as a refinement and expansion of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, some embodiments of this application also provide an AI-based satellite payload automatic scheduling method. Based on the scheduling process of the above embodiment, this method can also have autonomous learning and adaptability, and can be dynamically adjusted according to the changing task requirements and resource constraints. Figure 7 As shown, the method also includes:

[0119] S201, obtaining scheduling result data, wherein the scheduling result data includes task execution progress and external environment parameters;

[0120] S202, extracting an optimized feature set from the scheduling result data based on an autonomous learning algorithm, wherein the optimized feature set includes one or more combinations of clustering features, dimensionality reduction features, and association rule features;

[0121] S203, using the optimization feature set to evaluate the task load distribution data to obtain evaluation result data;

[0122] S204: Adjust the model parameters of the intelligent decision-making model according to the evaluation result data.

[0123] In this embodiment, the intelligent scheduling model can perform autonomous learning based on algorithms such as reinforcement learning, supervised learning and unsupervised learning, deep learning, and autonomous domain randomization to improve the adaptability of the intelligent scheduling model. For example, in reinforcement learning, the intelligent decision-making model can continuously try different actions through interaction with the environment, and adjust the strategy based on feedback (reward or punishment) to achieve the optimal goal. The intelligent decision-making model can also achieve autonomous learning through supervised learning and unsupervised learning. Supervised learning is trained with labeled data, and the goal is to enable the model to predict output results based on input data. Unsupervised learning does not require labeled data, and can learn by discovering patterns and structures in the data, such as cluster analysis and dimensionality reduction. Deep learning extracts useful features from complex and high-dimensional data through multi-layer neural networks, performs pattern recognition and decision-making, and realizes autonomous learning.

[0124] Take the unsupervised autonomous learning process as an example. Unsupervised learning is a machine learning training method that can discover potential structures and patterns in unlabeled data. Unsupervised learning tasks include clustering, dimensionality reduction, anomaly detection, and association rule mining.

[0125] When conducting unsupervised autonomous learning, the intelligent scheduling model can collect the scheduling result data generated after load scheduling. And perform data preprocessing, including data cleaning, feature scaling, feature encoding, etc., to remove noise and outliers in the scheduling result data and handle missing values. Normalize or standardize the features in the scheduling result data to make them on the same scale. Use algorithms such as One-Hot Encoding to convert non-numeric data into numerical data. Then perform feature extraction and selection to extract useful features from the scheduling result data and select the feature subset that is most helpful for discovering the data structure. Then, according to the type of load scheduling problem, select a suitable unsupervised learning algorithm, such as clustering, dimensionality reduction, association rule learning, etc. By using indicators such as the Silhouette Coefficient to evaluate the quality of the clustering results, as well as the interpretability and information retention of the data after dimensionality reduction, the model can be optimized and adjusted, that is, according to the result evaluation, the model parameters can be adjusted or different algorithms can be selected to improve the results. According to the application results, the model can be continuously iterated and optimized to improve the output accuracy and adaptability of the model.

[0126] It can be seen that by applying the technical solutions provided by the above embodiments, the scheduling system uses advanced algorithms such as deep learning and reinforcement learning to achieve accurate matching and intelligent scheduling of mission requirements and payload resources, and uses big data technology to quickly analyze and process the real-time data of satellite payloads, providing accurate data support for intelligent scheduling. In addition, an optimization and feedback module can also be set, which can dynamically adjust and optimize the scheduling scheme according to the task execution results and external environment changes, and feed back the optimization results to the intelligent scheduling algorithm module for iterative learning. Therefore, the intelligent scheduling model can have autonomous learning and adaptability, can be dynamically adjusted according to the changing mission requirements and resource constraints, and support the expansion and access of new payloads in the future. In addition, the scheduling system can also have high availability and fault tolerance, ensuring that the system can still operate stably in extreme environments or sudden failures, and improving the reliability and safety of the system.

[0127] Further, as a specific implementation of the AI-based satellite payload automatic scheduling method described in the above embodiment, the present application embodiment provides an AI-based satellite payload automatic scheduling system, such as Figure 8 As shown, the system includes:

[0128] A payload resource management module, used to obtain satellite payload data, wherein the satellite payload data includes real-time status information, mission requirement information, and resource constraint information;

[0129] An intelligent scheduling algorithm module is used to input the satellite payload data into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model, wherein the intelligent decision model is a machine learning model trained using payload sample data; the payload sample data includes a payload sample and a payload allocation label annotated for the payload sample; the intelligent decision model is configured to perform payload matching according to the satellite payload data;

[0130] A command generation module, used to generate a scheduling command set according to the task load distribution data;

[0131] The execution and monitoring module is used to send the scheduling command set to the satellite control system so that the satellite control system schedules the satellite payload according to the scheduling command set.

[0132] It should be noted that for other corresponding descriptions of the functional units involved in the AI-based satellite payload automatic scheduling system provided in the embodiment of the present application, reference can be made to the corresponding descriptions in the AI-based satellite payload automatic scheduling method provided in the above embodiment, and will not be repeated here.

[0133] By applying the technical solutions provided in the above embodiments, the AI-based satellite payload automation scheduling system can realize the dynamic scheduling and optimal configuration of satellite payloads (including sensors, communication equipment, data processing units, etc.) by integrating advanced artificial intelligence algorithms and real-time data analysis. The scheduling system can identify the task requirements, identify and parse the specific requirements of the task according to the task instructions received by the satellite, including the task type, target area, time requirements, etc. Then perform a payload resource assessment to monitor and evaluate the current status, performance parameters, power consumption, etc. of all available payloads on the satellite in real time. Then make intelligent scheduling decisions based on the intelligent decision-making model, and use AI algorithms to make intelligent decisions based on the task requirements and payload resource assessment results to determine the optimal payload configuration and scheduling plan. In addition, the scheduling system can also be dynamically adjusted and optimized. During the execution process, the payload scheduling plan is dynamically adjusted and optimized according to the progress of the task and changes in the external environment to solve the problem that satellite payloads are difficult to effectively schedule.

[0134] like Fig. 9 As shown, the embodiment of the present application also provides a computer device, which can be specifically a personal computer, a server, a network device, etc. The computer device includes a bus, a processor, a memory and a communication interface, and can also include an input and output interface and a display device. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store location information. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in each method embodiment are implemented.

[0135] Those skilled in the art will appreciate that the structure of the above-mentioned computer device is only a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components, or combine certain components, or have a different arrangement of components.

[0136] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium may be non-volatile or volatile, and stores a computer program thereon. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0137] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0139] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0140] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An AI-based satellite payload automation scheduling method, characterized in that: The method comprises: Acquiring satellite payload data, wherein the satellite payload data includes real-time status information, mission requirement information, and resource constraint information; The satellite payload data is input into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model, wherein the intelligent decision model is a machine learning model trained using payload sample data; the payload sample data includes a payload sample and a payload allocation label annotated for the payload sample; the intelligent decision model is configured to perform payload matching according to the satellite payload data; generating a scheduling command set according to the task load distribution data; The scheduling command set is sent to a satellite control system, so that the satellite control system schedules the satellite payload according to the scheduling command set.

2. The method according to claim 1, characterized in that: After acquiring the satellite payload data, the method further includes: Reading the real-time status information, the mission requirement information and the resource constraint information from the satellite payload data; Generate a task requirement item according to the task requirement information, wherein the task requirement item includes at least one of a task type, a task priority, and a task execution time; Generate a current status item according to the real-time status information, wherein the current status includes at least one of resolution, coverage, and energy consumption; The task analysis item and the load status item are set to the intelligent decision-making model as constraint conditions of the intelligent decision-making model.

3. The method according to claim 2, characterized in that After acquiring the satellite payload data, the method further includes: Generate a resource constraint item according to the resource constraint information, wherein the resource constraint item includes at least one of load working time, load power consumption, and storage space; generating an objective function according to the resource constraint item, wherein the dependent variable of the objective function is the utilization rate of the resource constraint item; The objective function is set to the intelligent decision-making model.

4. The method according to claim 1, characterized in that: After acquiring the satellite payload data, the method further includes: Performing data cleaning on the satellite payload data according to preset cleaning items, wherein the cleaning items include one or more combinations of data deduplication, missing value supplementation, outlier processing, and standardization; The satellite payload data after data cleaning is converted into a target format, wherein the target format is a data format set according to an input neural network node of the intelligent decision-making model.

5. The method according to claim 1, characterized in that: The method further comprises: Receive task instructions; In response to the task instruction, parsing the task requirement information from the task instruction; A task requirement report is generated according to the task requirement information.

6. The method according to claim 1, characterized in that Inputting the satellite payload data into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model includes: Extracting task-related features and load-related features from the satellite payload data, wherein the task-related features are semantic feature vectors extracted from the task requirement information; and the load-related features are semantic feature vectors extracted from the resource constraint information; matching a suitable load based on the task-related characteristics and the load-related characteristics; An association relationship between the suitable load and the task-related feature is established to generate the task load distribution data.

7. The method according to claim 6, characterized in that Matching a suitable load based on the task-related feature and the load-related feature includes: Extracting state-related features from the satellite payload data, wherein the state-related features are semantic feature vectors extracted from the real-time state information; Determine an optional load set and a non-optional load set based on the state association feature and the load association feature; The suitable load is extracted from the optional load set according to the task association feature.

8. The method according to claim 1, characterized in that The method further comprises: Obtaining load sample data and calling a trained model, wherein the machine learning neural network structure of the trained model is the same as that of the intelligent decision-making model; Inputting the load sample data into the trained model to obtain training load distribution data output by the trained model; Calculating training loss according to the training load distribution data and the load distribution label; Iteratively optimizing model parameters of the trained model based on the training loss; When the training loss reaches a preset loss requirement, the model parameters are output to obtain the intelligent decision-making model.

9. The method according to claim 1, characterized in that: The method further comprises: Obtaining scheduling result data, wherein the scheduling result data includes task execution progress and external environment parameters; Extracting an optimized feature set from the scheduling result data based on an autonomous learning algorithm, wherein the optimized feature set includes one or more combinations of clustering features, dimensionality reduction features, and association rule features; Using the optimized feature set to evaluate the task load distribution data to obtain evaluation result data; Adjust the model parameters of the intelligent decision-making model according to the evaluation result data.

10. An AI-based satellite payload automation scheduling system, characterized in that: The system comprises: A payload resource management module, used to obtain satellite payload data, wherein the satellite payload data includes real-time status information, mission requirement information, and resource constraint information; An intelligent scheduling algorithm module is used to input the satellite payload data into an intelligent decision model to obtain mission payload allocation data output by the intelligent decision model, wherein the intelligent decision model is a machine learning model trained using payload sample data; the payload sample data includes a payload sample and a payload allocation label annotated for the payload sample; the intelligent decision model is configured to perform payload matching according to the satellite payload data; A command generation module, used to generate a scheduling command set according to the task load distribution data; The execution and monitoring module is used to send the scheduling command set to the satellite control system so that the satellite control system schedules the satellite payload according to the scheduling command set.

Images