Satellite task planning method and device, program product and storage medium

Abstract

The embodiment of the invention provides a satellite task planning method and device, a program product and a storage medium, and the method comprises the steps: extracting basic features of a satellite task and resource state features of satellite resources used for executing the satellite task; generating a task planning scheme based on the basic features, the resource state features and a pre-trained target model; the task planning scheme generation operation comprises the following steps: determining the task priority of the satellite tasks based on the basic characteristics; determining an association relationship between the satellite tasks and the resource states based on the basic features and the resource state features, wherein the association relationship comprises correlation weights between the satellite tasks and between the satellite tasks and the satellite resources; and generating a task planning scheme based on the task priority and the association relationship. By applying the technical scheme of the invention, the technical problem of delay caused by unreasonable resource allocation due to difficulty in dynamically identifying the task priority in the prior art can be effectively solved, and the intelligence and efficiency of satellite task planning are improved.

Description

Technical Field

[0001] This application relates to the field of spacecraft operation control, and more specifically, to a satellite mission planning method and apparatus, program product, and storage medium. Background Technology

[0002] In the field of satellite mission planning, existing technologies exhibit limitations in addressing mission prioritization and resource allocation. Typically, these technologies rely on static rules or expert experience to set mission priorities. However, in dynamically changing environments, such as responding to emergencies or real-time adjustments to resource status, high-priority missions may suffer delays during resource allocation because their importance is not promptly recognized, reducing mission execution efficiency and emergency response capabilities. Furthermore, there are limitations in modeling the relationship between missions and resources; mission priorities are usually determined based on the mission's inherent attributes, failing to adequately consider actual needs.

[0003] Furthermore, mainstream optimization algorithms such as genetic algorithms or greedy algorithms have shortcomings in handling multi-objective optimization problems. They often focus on a single objective, such as maximizing mission benefits or minimizing mission completion time. Such single-objective optimization strategies may lead to unreasonable allocation of resources in satellite mission planning.

[0004] Existing technologies have shortcomings in dynamic mission priority identification, multi-objective optimization strategies, and global correlation modeling. These shortcomings restrict the intelligence and efficiency of satellite mission planning and cannot meet the increasingly complex mission requirements and resource scheduling challenges. Summary of the Invention

[0005] This application provides a satellite mission planning method, apparatus, program product, and storage medium to at least solve the technical problem in the prior art that it is difficult to dynamically identify mission priorities, leading to unreasonable resource allocation and delays.

[0006] According to one embodiment of this application, a satellite mission planning method is provided, comprising: extracting basic characteristics of a satellite mission and resource status characteristics of satellite resources used to perform the satellite mission;

[0007] Based on the aforementioned basic features, resource status features, and pre-trained target model, a task planning scheme is generated;

[0008] The task planning scheme generation operation includes: determining the task priority of the satellite task based on the basic features; determining the correlation between the satellite task and the resource status based on the basic features and the resource status features, wherein the correlation includes the correlation weights between satellite tasks and between satellite tasks and satellite resources; and generating a task planning scheme based on the task priority and the correlation.

[0009] In an exemplary embodiment, the target model is a model obtained by iteratively training an initial model using sample information. The iterative training includes: acquiring sample information, wherein the sample information includes basic characteristics of a satellite mission and resource status characteristics of satellite resources used to execute the satellite mission; constructing an optimized reward function based on mission priority, resource constraints, and execution efficiency; inputting the sample information into the initial model, and having the initial model generate a sample mission planning scheme based on the sample information; calculating the reward value of the sample mission planning scheme using the optimized reward function; and updating the parameters of the initial model based on the reward value to obtain the target model, wherein the optimized reward function of the target model converges.

[0010] In an exemplary embodiment, constructing an optimization reward function based on task priority, resource constraints, and execution efficiency includes: quantifying task priority, resource constraints, and execution efficiency and assigning them preset weight coefficients; and summing the quantified indicators by multiplying them by their respective weight coefficients as the optimization reward function.

[0011] In an exemplary embodiment, the target model includes a neural network module, an association module, and a scheme generation module. Correspondingly, generating a task planning scheme based on the basic features, the resource status features, and the pre-trained target model includes: inputting the basic features into the neural network module to output the task priority of the satellite mission; inputting the basic features and the resource status features into the association module to output the association relationship between the satellite mission and the resource status, wherein the association relationship includes the correlation weights between satellite missions and between satellite missions and satellite resources; and generating a task planning scheme based on the input task priority and the association relationship.

[0012] In one exemplary embodiment, the structure of the neural network module includes: a multilayer perceptron or a convolutional neural network, a plurality of hidden layers, and an activation function introduced between the hidden layers.

[0013] In one exemplary embodiment, when there is a new satellite mission or a status update of satellite resources, the neural network module is trained based on online learning or incremental learning techniques.

[0014] In one exemplary embodiment, the association module includes either a Transformer self-attention mechanism or a graph neural network.

[0015] In an exemplary embodiment, generating a mission planning scheme based on the mission priority and the correlation includes: obtaining constraints, the constraints including a total limit on satellite resources and a minimum requirement for satellite resources for a satellite mission; and, under the constraints, sorting and allocating satellite resources for multiple satellite missions according to the mission priority and the correlation.

[0016] In an exemplary embodiment, the extraction of the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission includes: acquiring raw satellite mission data; wherein the raw satellite mission data includes satellite mission-related data and satellite resources, the satellite mission-related data includes mission type, urgency level, load occupancy, bandwidth consumption, energy demand, expected benefits, time characteristics, and available time window, and the satellite resources include payload status, energy reserves, communication bandwidth, and orbital parameters; preprocessing the raw satellite mission data to eliminate the influence of different units and data ranges, the preprocessing including data cleaning and normalization; and extracting the basic characteristics and resource status characteristics of the preprocessed raw satellite mission data.

[0017] In an exemplary embodiment, extracting the basic features and resource status features of the preprocessed original satellite mission data includes: vectorizing the extracted basic features and resource status features; arranging the feature vectors of the basic features by rows to form a basic feature matrix; arranging the resource status features according to the resource feature dimensions to form an arranged resource status feature matrix; and fusing the basic feature matrix and the arranged resource status features as input to the target model.

[0018] According to another embodiment of this application, a satellite mission planning apparatus is provided, including a first memory, a first processor, and a first computer program stored in the first memory and executable on the first processor. When the first processor executes the first computer program, it performs the following operations: extracting basic features of a satellite mission and resource status features of satellite resources used to perform the satellite mission; generating a mission planning scheme based on the basic features, the resource status features, and a pre-trained target model; wherein the mission planning scheme generation operation includes: determining the mission priority of the satellite mission based on the basic features; determining the correlation between the satellite mission and the resource status based on the basic features and the resource status features, the correlation including correlation weights between satellite missions and between satellite missions and satellite resources; and generating the mission planning scheme based on the mission priority and the correlation.

[0019] According to yet another embodiment of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0020] According to yet another embodiment of this application, a computer-readable storage medium is also provided, wherein a computer program is stored therein, and the computer program is configured to perform the steps in any of the above method embodiments when it is run.

[0021] According to yet another embodiment of this application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0022] The embodiments described above in this application extract the basic characteristics of satellite missions and the resource status characteristics of satellite resources. Using a pre-trained target model, the application determines the mission priority of satellite missions based on the basic characteristics and the correlation between satellite missions and resource status based on the basic characteristics and resource status characteristics. Finally, based on the above-mentioned mission priority and correlation, the application generates and outputs a mission planning scheme. This application effectively solves the technical problem in the prior art that it is difficult to dynamically identify mission priorities, leading to unreasonable resource allocation and delays. Furthermore, by analyzing the correlation between satellite missions and resource status, the application more accurately identifies the matching degree between satellite missions and resource status, achieving efficient utilization of limited satellite resources, improving the intelligence level and execution efficiency of satellite mission planning, and significantly improving the planning quality and resource utilization efficiency of satellite missions. Attached Figure Description

[0023] Figure 1This is a schematic diagram of the hardware environment of a satellite mission planning method according to an embodiment of this application;

[0024] Figure 2 This is a flowchart of a satellite mission planning method according to an embodiment of this application;

[0025] Figure 3 This is a schematic flowchart illustrating the training process of a target model according to an embodiment of this application;

[0026] Figure 4 This is a flowchart of a target satellite mission planning method according to an embodiment of this application;

[0027] Figure 5 This is a structural block diagram of a satellite mission planning device according to an embodiment of this application. Detailed Implementation

[0028] The embodiments of this application will be described in detail below with reference to the accompanying drawings and examples.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0030] The methods and embodiments provided in this application can be executed on a server device or a similar computing device. Taking running on a server device as an example, Figure 1 This is a schematic diagram of the hardware environment for a satellite mission planning method according to an embodiment of this application. Figure 1 As shown, the server device may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The server device may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the server equipment described above. For example, the server equipment may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0031] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to a satellite mission planning method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thus implementing the aforementioned method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to server devices via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0032] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider for the server device. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0033] This embodiment provides a satellite mission planning method. Figure 2 This is a flowchart of a satellite mission planning method according to an embodiment of this application, such as... Figure 2 As shown, the process includes the following steps:

[0034] Step S202: Extract the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission.

[0035] Optionally, a satellite mission can be a specific task or operation performed by the satellite, such as imaging and data transmission missions, involving activities such as satellite observation of Earth or other celestial bodies and data transmission. Basic characteristics can be the fundamental attributes of a satellite mission, including mission type, urgency, payload utilization, bandwidth consumption, energy requirements, and expected benefits, used to describe the specific needs of the mission. Satellite resources can be the available hardware and system resources of the satellite, including payload, energy, and communication bandwidth, used to perform the satellite mission. Resource status characteristics can be indicators used to describe the current state of satellite resources, such as payload operational status, remaining lifetime percentage, energy reserves, communication bandwidth, and orbital parameters, used to assess resource availability and limitations.

[0036] For example, in the on-orbit test mission planning scenario, basic features such as mission type, urgency, load occupancy, bandwidth consumption, energy demand and expected benefits were extracted, while resource status features such as satellite payload status, energy reserves, communication bandwidth and orbital parameters were also obtained.

[0037] Optionally, the extraction of the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission includes: acquiring raw satellite mission data; wherein the raw satellite mission data includes satellite mission-related data and satellite resources, the satellite mission-related data includes mission type, urgency, load occupancy, bandwidth consumption, energy demand, expected benefits, time characteristics, and available time window, and the satellite resources include payload status, energy reserves, communication bandwidth, and orbital parameters; preprocessing the raw satellite mission data to eliminate the influence of different units and data ranges, the preprocessing including data cleaning and normalization; and extracting the basic characteristics and resource status characteristics of the preprocessed raw satellite mission data.

[0038] In this embodiment, the satellite mission planning system first acquires raw data on satellite missions and related resources, including mission type, urgency, load occupancy, bandwidth consumption, energy requirements, expected benefits, time characteristics associated with the mission, and available time windows. It also covers the status of satellite resources, such as payload operational status, energy reserve status, remaining communication bandwidth, and orbital parameters. Subsequently, preprocessing this raw data eliminates analytical obstacles caused by differences in units and data ranges, ensuring data consistency and comparability. Preprocessing steps include data cleaning to remove invalid or erroneous information and data normalization to scale all feature values ​​to the same range, laying a solid foundation for subsequent analysis. Next, the system extracts basic satellite mission features and resource status features from the preprocessed data. Through vectorization, various types of information are converted into numerical forms, facilitating computer understanding and processing. This series of operations effectively constructs a feature set containing mission and resource information, providing data-level preparation for mission planning based on policy learning and intelligent mission priority recognition, ensuring the accuracy and timeliness of subsequent model training and decision-making.

[0039] For example: If the task type set is {on-orbit testing, application demonstration, technology experiment}, then "on-orbit testing" is represented as [1,0,0]. Urgency: The normalized value is directly used as the feature value, for example, the value range is [0,1], where 0 represents the lowest urgency and 1 represents the highest urgency. Load occupancy: The load type (e.g., optical payload, radar payload) is one-hot encoded, and the load occupancy duration is normalized to a value in the [0,1] interval; the two are concatenated into a vector. Bandwidth consumption: The normalized bandwidth requirement value is combined with the transmission time window (which can be discretized into time periods for one-hot encoding). Expected benefit: The normalized benefit value can be combined with the one-hot encoding of the benefit type (e.g., scientific value, economic value). Time characteristics: Converted to the proportion of remaining time until the current moment (normalized to [0,1]). Available time window: The time window is discretized into multiple time periods, and the available / unavailable status is represented by a binary vector. Load status: The working status (idle / busy) and remaining lifetime percentage of each load (normalized). Energy reserves: The proportion of current energy reserves to total capacity (normalized). Communication bandwidth: The proportion of remaining available bandwidth to total bandwidth (normalized). Orbital parameters: Orbital location (discreteable into orbital regions), illumination conditions.

[0040] Optionally, the basic features and resource status features of the preprocessed original satellite mission data are extracted, including: vectorizing the extracted basic features and resource status features; arranging the feature vectors of the basic features by rows to form a basic feature matrix; arranging the resource status features according to the resource feature dimensions to form an arranged resource status feature matrix; and fusing the basic feature matrix and the arranged resource status features as input to the target model.

[0041] In this embodiment, the preprocessed raw satellite mission data undergoes feature extraction and vectorization, specifically including the quantification of key features such as basic mission information, urgency, and resource requirements. These basic feature vectors are integrated into a matrix, while resource status features, such as payload operating status, energy reserves, and communication bandwidth, are arranged according to resource feature dimensions. Subsequently, the basic feature matrix and resource status feature vectors are combined through concatenation or attention mechanisms to form the input tensor of the target model. This fusion processing method aims to enable deep policy neural networks to comprehensively capture the correlation between tasks and resources, dynamically adjust task priorities through learning, and achieve efficient resource allocation and collaborative task scheduling. In terms of implementation effectiveness, this scheme can effectively improve the flexibility and accuracy of mission planning, meet multi-objective optimization needs, and significantly improve mission execution efficiency and resource utilization, especially in scenarios with limited resources, urgent tasks, or multiple task conflicts.

[0042] For example: Arrange the feature vectors of each task row-wise to form a basic feature matrix, where n is the number of tasks and d is the feature dimension of a single task. Integrate the satellite resource states into a global resource vector, where m is the resource feature dimension. Input tensor construction involves concatenating the task feature matrix T and the resource state vector R or fusing them through an attention mechanism to form the final input tensor, which serves as the input to the network.

[0043] Step S204: Generate a mission planning scheme based on the basic features, the resource status features, and the pre-trained target model; wherein, the mission planning scheme generation operation includes: determining the mission priority of the satellite mission based on the basic features; determining the correlation between the satellite mission and the resource status based on the basic features and the resource status features, wherein the correlation includes the correlation weights between satellite missions and between satellite missions and satellite resources; and generating a mission planning scheme based on the mission priority and the correlation.

[0044] In this embodiment, the pre-trained target model can be a deep policy neural network model trained using historical or simulation data before actual application, used to dynamically model the relationship between task priority and resource allocation. The task planning scheme involves determining the satellite task execution order and resource allocation strategy based on task priority and resource status. Task priority can be a priority order determined based on indicators such as task urgency and expected benefits, used to guide resource allocation and task scheduling. The correlation relationships can be the interdependencies and influences between tasks and between tasks and satellite resources, modeled through a self-attention mechanism. The correlation weights can be calculated within the self-attention mechanism to determine the correlation strength between tasks and between tasks and resources, used to guide resource allocation and multi-task scheduling.

[0045] Optionally, the target model in this embodiment includes a neural network module, an association module, and a scheme generation module. Correspondingly, the step of generating a task planning scheme based on the basic features, the resource status features, and the pre-trained target model includes: inputting the basic features into the neural network module to output the task priority of the satellite mission; inputting the basic features and the resource status features into the association module to output the association relationship between the satellite mission and the resource status, wherein the association relationship includes the correlation weights between satellite missions and between satellite missions and satellite resources; and the scheme generation module generating a task planning scheme based on the input task priority and the association relationship.

[0046] In this embodiment, the target model integrates three main parts: a neural network module, an association module, and a scheme generation module, aiming to generate optimized task planning schemes. Specifically, the basic features are processed by the neural network module to output a dynamic priority score for satellite tasks. This scoring mechanism can adjust the priority order of tasks in real time according to task characteristics and environmental changes. Simultaneously, the basic features and resource status features are fed into the association module, which uses self-attention mechanisms or other similar techniques to calculate the correlation weights between tasks and between tasks and resources, thereby establishing a global perspective on task and resource dependencies. Finally, the scheme generation module generates a task planning scheme that considers both task urgency and balanced resource allocation based on the task priority scores and the correlation weights between resources. Throughout the entire scheme operation, the modules complement each other, jointly promoting the intelligence and efficiency of task planning, ensuring that high-priority tasks are executed first while resource utilization reaches its optimal state, significantly improving the flexibility of task planning and the scientific nature of resource allocation.

[0047] Optionally, the structure of the neural network module includes: a multilayer perceptron or a convolutional neural network, multiple hidden layers, and activation functions introduced between the hidden layers.

[0048] In this embodiment, the neural network module's structural design includes a multilayer perceptron or convolutional neural network. By stacking multiple hidden layers and introducing activation functions between layers, the model's ability to express complex nonlinear relationships is enhanced. This structure enables the neural network to deeply understand and learn multi-level information from task feature vectors, effectively handle nonlinear relationships, and provide strong support for dynamic task priority modeling. The use of activation functions such as ReLU or Leaky ReLU injects nonlinear factors into the network, further improving the model's ability to fit complex data patterns in the real world, ensuring more accurate and reliable task priority scoring. Of course, in other embodiments, the structure of the neural network module may differ, such as using long short-term memory networks or more advanced network architectures, but overall, they are all aimed at enhancing the model's learning ability and adaptability to respond in real time to changing task environments, optimize resource allocation strategies, and ultimately achieve efficient and scientific task planning. These variations do not deviate from the core technical framework of this embodiment and are all designed to improve the model's understanding and response to dynamic changes in task priorities.

[0049] The design and implementation of the aforementioned neural network module significantly improves the flexibility and resource utilization of mission planning, enabling intelligent prioritization of tasks and dynamic resource allocation, thereby enhancing the overall system efficiency and mission success rate. In specific scenarios, such as those with highly limited resources or fluctuating mission urgency, the efficiency and adaptability of this neural network module are particularly prominent, effectively ensuring the timely execution of high-priority tasks while maximizing the rational utilization of resources. In summary, the neural network module design in this embodiment brings a novel AI-based optimization solution to the field of satellite mission planning, effectively addressing the limitations of traditional planning methods and promoting the intelligentization of mission planning.

[0050] Optionally, the associated module may include: a Transformer self-attention mechanism or a graph neural network.

[0051] In this embodiment, the association module employs a Transformer self-attention mechanism or a graph neural network to capture and analyze the global dependencies between tasks and resources. By introducing these advanced network structures, the model can efficiently calculate the correlation weights between tasks and between tasks and resources, automatically learning long-distance dependencies. This mechanism enables the model to deeply understand the intensity of each task's demand for different resources and the degree of influence of resource status on task execution, thereby providing more accurate guidance in resource allocation and task priority decisions. Implementation results show that both the Transformer self-attention mechanism and the graph neural network significantly improve the global perspective of task planning and the rationality of resource allocation, avoiding resource mismatch and inefficiency problems that may be caused by local optima, thus optimizing the overall efficiency and quality of task execution. In other embodiments not shown, by selecting different network structures or parameters, the performance of the task planning algorithm can be further explored and optimized to adapt to more diverse business scenarios and resource constraints.

[0052] Optionally, generating a mission planning scheme based on the mission priority and the correlation includes: obtaining constraints, including a total limit on satellite resources and a minimum requirement for satellite resources for each satellite mission; and, under the constraints, sorting and allocating satellite resources for multiple satellite missions according to the mission priority and the correlation.

[0053] In this embodiment, constraints are used as the basis for generating the mission planning scheme. These constraints include the upper limit of the total satellite resources and the basic resource requirements of the satellite missions. Under these constraints, satellite missions are prioritized and satellite resources are allocated by dynamically calculating mission priority scores through a deep learning model and revealing the inter-mission relationships through a self-attention mechanism. This technical solution ensures that high-priority missions can be executed first when resources are sufficient, while also considering the rational use of resources and time coordination between missions, avoiding resource waste and mission delays.

[0054] By comprehensively considering multiple dimensions and their weights, this method can effectively balance task priority and resource utilization, improve overall task execution efficiency and the scientific nature of resource allocation, and thus achieve better task planning and execution while meeting resource constraints.

[0055] In this embodiment, the entities executing the above steps can be widely distributed intelligent processing nodes, not limited to ground facilities, but also including devices in the air and even space. These can be intelligent processing devices installed in satellites, edge computing devices, ground control centers or data centers, or cloud computing platforms.

[0056] By applying the technical solution of this embodiment, a task planning method based on policy learning and intelligent task priority recognition is realized, effectively solving the problem in existing satellite mission planning where dynamic identification of task priorities is difficult, leading to delays in high-priority tasks due to unreasonable resource allocation. This method extracts the basic features of satellite missions and the resource status features of satellite resources using deep learning technology, and dynamically establishes a task priority model using a deep policy neural network, automatically adjusting task priorities based on factors such as urgency and resource requirements. Simultaneously, it analyzes the global dependencies between tasks and between tasks and resources through a self-attention mechanism, calculating correlation weights to more accurately identify the matching degree between tasks and resources. Based on dynamic priorities and global correlations, this technology generates an optimal task planning scheme, ensuring that high-priority tasks receive priority scheduling and resource support, while efficiently utilizing limited resources, avoiding resource mismatch and low utilization, and improving the intelligence level and execution efficiency of the entire satellite mission planning system. This method overcomes the shortcomings of traditional task planning algorithms, especially its limitations in dealing with emergencies and dynamic resource changes, significantly improving the planning quality and resource utilization efficiency of satellite missions through intelligent decision-making.

[0057] In an exemplary embodiment, the target model described above is a model obtained by iteratively training an initial model using a set of sample information. This iterative training includes:

[0058] Step S302: Obtain sample information, wherein the sample information includes the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission.

[0059] Step S304: Construct an optimized reward function based on task priority, resource constraints, and execution efficiency;

[0060] Step S306: The sample information is input into the initial model, and the initial model generates a sample task planning scheme based on the sample information;

[0061] Step S308: Calculate the reward value of the sample task planning scheme using the optimized reward function;

[0062] Step S310 updates the parameters of the initial model based on the reward value to obtain the target model, wherein the optimized reward function of the target model converges.

[0063] In this embodiment, the target model is obtained through iterative training on sample information, which covers the basic characteristics of satellite missions and the status characteristics of satellite resources required to execute these missions. After the initial model generates a mission planning scheme based on the input sample information, it uses a designed optimization reward function that comprehensively considers mission priority, resource constraints, and execution efficiency to calculate the reward value of the mission planning scheme. By feeding back the reward value, the parameters of the initial model are updated, gradually approaching the optimal state until the optimization reward function of the target model converges, ensuring that the model can produce the optimal mission planning result. This iterative training process continuously strengthens the model's learning ability, enabling it to balance various factors when handling satellite missions, making more reasonable resource allocation and mission priority arrangements, and improving mission execution efficiency and resource utilization. Of course, in embodiments not detailed above, the specific steps of model iterative training may differ, such as using different optimization algorithms or adjusting the calculation method of the optimization reward function, but the core objective remains the same: to improve the intelligence level and execution effect of mission planning.

[0064] Optionally, the target model is obtained through iterative training. The training process includes acquiring basic features and resource status features of the satellite mission as sample information, and constructing an optimized reward function that integrates mission priority, resource constraints, and execution efficiency. The sample information is input into the initial model, which generates a mission planning scheme. The optimized reward function is then used to calculate the reward value of this scheme. Based on the reward value, the model parameters are updated. Finally, the optimized reward function converges, and the generated target model can more accurately evaluate and adjust mission priorities, achieving effective resource allocation. This process ensures that the model can quickly adapt to changes in mission urgency and resource status in a dynamic environment, providing the optimal mission planning strategy, thereby improving the performance and response speed of the entire mission planning system. Of course, in other embodiments, the specific methods of iterative training may differ, such as employing different learning strategies or optimization algorithms, but the aim remains the same: to achieve higher mission planning accuracy and resource utilization efficiency through continuous learning and adjustment.

[0065] Optionally, although this embodiment employs a specific iterative training process and reward function design, in other embodiments not shown, horizontal shifting or a completely different model architecture, such as LSTM or CNN replacing Transformer, may be used to achieve progressive optimization and dynamic adjustment of task priorities. These variations also aim to improve the intelligence level of task planning, ensuring efficient and reasonable resource allocation and task scheduling even under complex conditions of limited resources and fluctuating task urgency, thereby maximizing the successful execution and resource utilization of satellite missions.

[0066] Optionally, an optimization reward function based on task priority, resource constraints, and execution efficiency can be constructed, including: quantifying task priority, resource constraints, and execution efficiency and assigning them preset weight coefficients respectively; and using the sum of the quantified indicators multiplied by their respective weight coefficients as the optimization reward function.

[0067] In this embodiment, the constructed optimized reward function further enhances the intelligence level of task planning. By quantifying task priority, resource constraints, and execution efficiency, and assigning preset weight coefficients to each quantified indicator, a comprehensive evaluation system is formed. The core of this strategy lies in its ability to convert the above three dimensions of indicators into numerical form. Through weight adjustment, the optimized reward function can reflect the multi-objective requirements of task planning. Specifically, the quantification of task priority considers the urgency and type of the task, the resource constraint indicator focuses on resource availability and task requirements, and execution efficiency measures the timeliness and success rate of task completion. In practical applications, the quantified indicators are multiplied by their corresponding weight coefficients and summed. The resulting sum is used as the optimized reward function value to guide the model in learning the optimal task prioritization strategy. This mechanism ensures that in a dynamically changing environment, the model can comprehensively consider the importance of tasks and the utilization rate of resources, achieving more scientific task scheduling. By dynamically adjusting the weight coefficients, the model can prioritize urgent or important tasks based on real-time task requirements and resource status, while avoiding resource waste and improving the overall efficiency and success rate of task execution. In other embodiments not shown, the construction of the optimized reward function may differ, for example, by using a nonlinear combination or other mathematical models. However, the core remains the same: quantifying multiple indicators and adjusting their weights according to the specific needs of task planning in order to achieve more efficient task priority decisions and resource allocation.

[0068] For example, key indicators such as task priority, resource constraints, and execution efficiency are quantified and assigned different weights to reflect their importance in task planning. Task priority is quantified based on priority scores output by the policy neural network; resource constraints consider satellite resource availability (e.g., remaining energy, payload usage time) and the task's resource requirements; execution efficiency measures the timeliness and success rate of task execution. A multi-objective optimization reward function is constructed that comprehensively considers task priority, resource constraints, and execution efficiency. The reward function can be a weighted sum, where each quantified indicator is multiplied by its corresponding weight and then summed to obtain a comprehensive reward value.

[0069] The reward function can be expressed as:

[0070] R=w1 P+w2 C+w3 E

[0071] Where R is the overall reward value, P is the task priority indicator, C is the resource constraint indicator, E is the execution efficiency indicator, and w1, w2 and w3 are the corresponding weight coefficients.

[0072] Optionally, the weight coefficients of each indicator can be dynamically adjusted based on the specific needs and characteristics of the task planning scenario. When resources are scarce, the weight of resource constraint indicators is increased, making the model more focused on the rational use of resources. When the task urgency is high, the weight of task priority indicators is increased, prioritizing the scheduling of high-priority tasks. A performance feedback-based weight adjustment mechanism is adopted, dynamically adjusting the weight coefficients according to the model's performance in actual task planning (such as task delay rate, resource utilization rate, etc.) to achieve a dynamic balance among multiple objectives.

[0073] Optionally, the initial model includes a neural network module, an association module, and a scheme generation module. Correspondingly, sample information is input into the initial model, and the initial model generates sample task planning schemes based on the sample information, including: inputting basic features into the neural network module to output the task priority of the satellite mission; inputting basic features and resource status features into the association module to output the association relationship between the satellite mission and the resource status, including the correlation weights between satellite missions and between satellite missions and satellite resources; and generating a task planning scheme based on the task priority and the association relationship.

[0074] In this embodiment, the initial model integrates three main parts: a neural network module, an association module, and a scheme generation module, aiming to generate optimized task planning schemes. Specifically, the basic features are processed by the neural network module to output a dynamic priority score for satellite tasks. This scoring mechanism can adjust the priority order of tasks in real time based on task characteristics and environmental changes. Simultaneously, the basic features and resource status features are fed into the association module, which uses self-attention mechanisms or other similar techniques to calculate the correlation weights between tasks and between tasks and resources, thereby establishing a global perspective on task and resource dependencies. Finally, the scheme generation module generates a task planning scheme that considers both task urgency and balanced resource allocation based on the task priority scores and the correlation weights between resources. Throughout the entire scheme operation, the modules complement each other, jointly promoting the intelligence and efficiency of task planning, ensuring that high-priority tasks are executed first while resource utilization reaches its optimal state, significantly improving the flexibility of task planning and the scientific nature of resource allocation. In variant embodiments not explicitly shown, the neural network module may be replaced by alternative structures such as LSTM and CNN, the correlation module may use graph neural network technology to analyze the interaction between tasks and resources in more detail, and the solution generation module may also adopt strategies such as priority queues for multi-task scheduling. These variations are all designed to flexibly adjust the technical solutions to achieve the best results according to different application scenarios and specific needs.

[0075] Optionally, when there are new satellite missions or updates to the status of satellite resources, the neural network module can be trained based on online learning or incremental learning techniques.

[0076] In this embodiment, when a new satellite mission emerges or the satellite resource status changes, the system updates the training of the deep policy neural network in real time based on online learning or incremental learning techniques. This means that the policy network can quickly absorb new mission information and the latest resource status without having to train the entire model from scratch, thereby significantly shortening the model update cycle and maintaining the real-time performance and effectiveness of mission planning decisions. Through online learning or incremental learning, the system can continuously optimize model parameters to better fit the actual situation of the current mission environment, improving the model's adaptability to dynamic scenarios and prediction accuracy. This mechanism ensures that even under constantly changing mission requirements and resource constraints, mission priority scores can be dynamically adjusted to achieve efficient resource allocation and flexible mission scheduling, thereby improving the overall performance and intelligent decision-making level of satellite mission planning. Of course, in other implementation variations of the technology, different learning frameworks or optimization algorithms can be explored to achieve similar effects. For example, online learning can be performed using Q-learning or Policy Gradients methods of reinforcement learning, or transfer learning techniques can be used to achieve cross-scenario knowledge transfer. These alternatives also aim to enhance the dynamic adaptability of the system and the intelligence of resource allocation.

[0077] Optionally, to achieve collaborative scheduling of multiple tasks, multiple tasks are sorted and scheduled according to task priority and resource allocation results to ensure that high-priority tasks can be executed first, while taking into account the execution efficiency and resource utilization of other tasks.

[0078] The above method will be illustrated with specific examples below:

[0079] A satellite mission planning system aims to dynamically identify mission priorities in real time to allocate resources rationally. This system includes a satellite mission planning model and an event information generation model comprising a neural network module, an association module, and a solution generation module. Figure 4 This is a flowchart of a target satellite mission planning method according to an embodiment of this application. The system utilizes an event information generation model through... Figure 4 The following steps are shown to generate target event information:

[0080] Step S402: Extract the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission.

[0081] Step S404: Input the basic features into the neural network module and output the mission priority of the satellite mission.

[0082] Step S406: Input the basic features and the resource status features into the association module, and output the association relationship between satellite missions and resource status. The association relationship includes the correlation weights between satellite missions and between satellite missions and satellite resources.

[0083] Step S408: Input the task priority and the relationship into the solution generation module. The solution generation module generates a task planning solution based on the input task priority and the relationship.

[0084] The execution order of steps S404 and S406 can be interchanged; that is, step S404 can be executed first and then step S406, or step S406 can be executed first and then step S404.

[0085] The feature extraction process begins with data preprocessing. The initial goal is to collect satellite mission-related data, including mission type, urgency, required resources (such as workload, bandwidth consumption, and energy demand), and expected benefits. This raw data needs to be cleaned and normalized to eliminate the influence of different units and data ranges, ensuring the reliability and comparability of subsequent inputs. Based on the preprocessed data, key features are extracted, focusing on dimensions closely related to priority assessment, such as mission urgency indicators, resource utilization (such as workload duration and energy consumption ratio), and expected benefits. These key features are then converted into vectors using one-hot encoding.

[0086] Through the steps described above, the system extracts the fundamental features of satellite missions and the resource status characteristics of satellite resources using deep learning technology. It then dynamically establishes a mission priority model using a deep policy neural network, automatically adjusting mission priorities based on factors such as urgency and resource requirements. Simultaneously, it analyzes global dependencies between tasks and between tasks and resources using a self-attention mechanism, calculating relevance weights to more accurately identify the matching degree between tasks and resources. Based on dynamic priorities and global correlations, this technology generates optimal mission planning schemes, ensuring that high-priority tasks receive priority scheduling and resource support, while efficiently utilizing limited resources, avoiding resource mismatch and low utilization rates, and improving the intelligence level and execution efficiency of the entire satellite mission planning system. This method overcomes the shortcomings of traditional mission planning algorithms, especially its limitations in dealing with emergencies and dynamic resource changes. Through intelligent decision-making, it significantly improves the planning quality and resource utilization efficiency of satellite missions.

[0087] It should be noted that, through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described above in the various embodiments of this application.

[0088] This embodiment also provides a satellite mission planning apparatus for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0089] Figure 5 This is a structural block diagram of a satellite mission planning device according to an embodiment of this application, such as... Figure 5 As shown, the device includes a first memory 52, a first processor 54, and a first computer program 5202 stored in the first memory 52 and executable on the first processor 54. When the first processor 54 executes the first computer program 5202, it performs the following operations: extracting basic features of the satellite mission and resource status features of the satellite resources used to perform the satellite mission; generating a mission planning scheme based on the basic features, the resource status features, and a pre-trained target model; wherein, the mission planning scheme generation operation includes: determining the mission priority of the satellite mission based on the basic features; determining the correlation between the satellite mission and the resource status based on the basic features and the resource status features, wherein the correlation includes the correlation weights between satellite missions and between satellite missions and satellite resources; and generating a mission planning scheme based on the mission priority and the correlation.

[0090] The target model is obtained by iteratively learning and training the initial model using a set of sample information. When the first processor 54 executes the first computer program 5202, it also performs the following operations: acquiring sample information, wherein the sample information includes the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission; constructing an optimized reward function based on mission priority, resource constraints, and execution efficiency; inputting the sample information into the initial model, which generates a sample mission planning scheme based on the sample information; calculating the reward value of the sample mission planning scheme using the optimized reward function; updating the parameters of the initial model based on the reward value to obtain the target model, wherein the optimized reward function of the target model converges.

[0091] When the first processor 54 executes the first computer program 5202, it also performs the following operations: quantifies the task priority, resource constraints, and execution efficiency and assigns them preset weight coefficients respectively; and uses the sum of the quantified indicators multiplied by their corresponding weight coefficients as the optimization reward function.

[0092] When the first processor 54 executes the first computer program 5202, it also performs the following operations: inputting the basic features into the neural network module and outputting the mission priority of the satellite mission; inputting the basic features and the resource status features into the association module and outputting the association relationship between the satellite mission and the resource status, wherein the association relationship includes the correlation weights between satellite missions and between satellite missions and satellite resources; and generating a mission planning scheme based on the input mission priority and the association relationship.

[0093] When the first processor 54 executes the first computer program 5202, it also performs the following operations: the structure of the neural network module includes: a multilayer perceptron or convolutional neural network, multiple hidden layers, and an activation function introduced between the hidden layers.

[0094] When the first processor 54 executes the first computer program 5202, it also performs the following operation: when there is a new satellite mission or the status of satellite resources is updated, the neural network module is trained based on online learning or incremental learning technology.

[0095] When the first processor 54 executes the first computer program 5202, it also performs the following operations: the associated module includes a Transformer self-attention mechanism or a graph neural network.

[0096] When the first processor 54 executes the first computer program 5202, it also performs the following operations: obtaining constraints, including a total limit on satellite resources and a minimum requirement for satellite resources for satellite missions; and, under the constraints, sorting and allocating satellite resources for multiple satellite missions according to the mission priorities and the aforementioned relationships.

[0097] When the first processor 54 executes the first computer program 5202, it also performs the following operations: retrieving raw satellite mission data; wherein the raw satellite mission data includes satellite mission-related data and satellite resources; the satellite mission-related data includes mission type, urgency level, load occupancy, bandwidth consumption, energy demand, expected benefits, time characteristics, and available time window; the satellite resources include payload status, energy reserves, communication bandwidth, and orbital parameters; preprocessing the raw satellite mission data to eliminate the influence of different dimensions and data ranges; the preprocessing includes data cleaning and normalization; and extracting the basic characteristics and resource status characteristics of the preprocessed raw satellite mission data.

[0098] When the first processor 54 executes the first computer program 5202, it also performs the following operations: extracting the basic features and resource status features of the preprocessed original satellite mission data, including: vectorizing the extracted basic features and resource status features; arranging the feature vectors of the basic features by rows to form a basic feature matrix; arranging the resource status features according to the resource feature dimensions to form an arranged resource status feature; and fusing the basic feature matrix and the arranged resource status features as input to the target model.

[0099] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above method embodiments when run.

[0100] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0101] Embodiments of this application also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0102] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0103] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0104] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the above method embodiments.

[0105] The embodiments described herein also provide a computer program that includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in any of the above method embodiments.

[0106] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.

[0107] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0108] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A satellite mission planning method, characterized in that, The method includes: Extract the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission; Based on the aforementioned basic features, resource status features, and pre-trained target model, a task planning scheme is generated; The task planning scheme generation operation includes: determining the task priority of the satellite task based on the basic features; determining the correlation between the satellite task and the resource status based on the basic features and the resource status features, wherein the correlation includes the correlation weights between satellite tasks and between satellite tasks and satellite resources; and generating a task planning scheme based on the task priority and the correlation.

2. The method according to claim 1, characterized in that, The target model is obtained by iteratively training an initial model using sample information. The iterative training includes: Obtain sample information, wherein the sample information includes basic characteristics of the satellite mission and resource status characteristics of the satellite resources used to perform the satellite mission; Construct an optimized reward function based on task priority, resource constraints, and execution efficiency; The sample information is input into the initial model, and the initial model generates a sample task planning scheme based on the sample information; The reward value of the sample task planning scheme is calculated using the optimized reward function; The parameters of the initial model are updated based on the reward value to obtain the target model, wherein the optimized reward function of the target model converges.

3. The method according to claim 2, characterized in that, The construction of an optimized reward function based on task priority, resource constraints, and execution efficiency includes: Task priority, resource constraints, and execution efficiency are quantified and assigned preset weight coefficients respectively; The sum of the quantified indicators multiplied by their respective weighting coefficients is used as the optimized reward function.

4. The method according to claim 1, characterized in that, The target model includes a neural network module, an association module, and a scheme generation module. Correspondingly, the generation of a task planning scheme based on the basic features, the resource state features, and the pre-trained target model includes: The basic features are input into the neural network module, which outputs the mission priority of the satellite mission. The basic features and the resource status features are input into the association module, and the association relationship between satellite missions and resource status is output. The association relationship includes the correlation weights between satellite missions and between satellite missions and satellite resources. The scheme generation module generates a task planning scheme based on the input task priority and the correlation relationship.

5. The method according to claim 4, characterized in that, The structure of the neural network module includes: a multilayer perceptron or a convolutional neural network, multiple hidden layers, and activation functions introduced between the hidden layers.

6. The method according to claim 4, characterized in that, Also includes: When there are new satellite missions or updates to the status of satellite resources, the neural network module is trained based on online learning or incremental learning techniques.

7. The method according to claim 4, characterized in that, The associated module includes either a Transformer self-attention mechanism or a graph neural network.

8. The method according to claim 4, characterized in that, The process of generating a task planning scheme based on the task priority and the correlation includes: Obtain the constraints, which include the total limit on satellite resources and the minimum requirement of satellite resources for satellite missions; Under the constraints, multiple satellite missions are sorted and satellite resources are allocated according to the mission priorities and the correlation relationships.

9. The method according to claim 1, characterized in that, The extraction of the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission includes: Acquire raw satellite mission data; wherein, the raw satellite mission data includes satellite mission-related data and satellite resources, the satellite mission-related data includes mission type, urgency level, load occupancy, bandwidth consumption, energy demand, expected benefits, time characteristics and available time window, and the satellite resources include payload status, energy reserves, communication bandwidth and orbital parameters; The raw data of the satellite mission is preprocessed to eliminate the influence of different units and data ranges. The preprocessing includes data cleaning and normalization. Extract the basic features and resource status features of the preprocessed original data of the satellite mission.

10. The method according to claim 9, characterized in that, Extract the basic features and resource status features of the preprocessed raw satellite mission data, including: The extracted basic features and resource status features are vectorized. The feature vectors of the basic features are arranged in rows to form a basic feature matrix. The resource status features are arranged according to the resource feature dimension to form an arranged resource status feature matrix. The basic feature matrix and the arranged resource state features are fused together and used as input to the target model.

11. A satellite mission planning device, characterized in that, It includes a first memory, a first processor, and a first computer program stored in the first memory and executable on the first processor. When the first processor executes the first computer program, it performs the following operations: Extract the basic characteristics of the satellite mission and the resource status characteristics of the satellite resources used to perform the satellite mission; Based on the aforementioned basic features, resource status features, and pre-trained target model, a task planning scheme is generated; The task planning scheme generation operation includes: determining the task priority of the satellite task based on the basic features; determining the correlation between the satellite task and the resource status based on the basic features and the resource status features, wherein the correlation includes the correlation weights between satellite tasks and between satellite tasks and satellite resources; and generating a task planning scheme based on the task priority and the correlation.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of the method described in any one of claims 1 to 10.

14. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 10.

Images