A method, apparatus, device and storage medium for generating deep space orbital data

Abstract

This application discloses a method, apparatus, device, and storage medium for generating deep space orbit data, relating to the field of deep space orbit design technology. The method includes: acquiring lunar-Earth space periodic orbit data from a publicly available data source; generating new orbit data samples based on the lunar-Earth space periodic orbit data using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling; sequentially determining whether the new orbit data samples meet the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance, and deep space orbit dynamics feature uniqueness; and determining the orbit data sample as compliant deep space orbit data if it meets these requirements. The progressive judgment of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance, and deep space orbit dynamics feature uniqueness filters out invalid samples at each level, improving the usability of the orbit data and the overall quality of the dataset.

Description

Technical Field

[0001] This application relates to the field of deep space orbit design technology, and in particular to a method, apparatus, device and storage medium for generating deep space orbit data. Background Technology

[0002] With the increasing number of deep space exploration missions, the design requirements for Earth-Moon periodic orbits are shifting from the traditional high-cost, low-frequency model to a low-cost, high-frequency, and intelligent approach. These orbits operate within a multi-body gravitational environment formed by the Earth and the Moon, requiring compliance with multi-field coupling constraints, long-period operational stability, and engineering implementation conditions. This places high demands on the quantity, diversity, and availability of orbital data.

[0003] Existing methods for acquiring deep space orbital data are mainly divided into two categories: one relies on traditional numerical integration and optimization methods to generate data. This method depends on manually setting initial values ​​and iterative correction, resulting in a long generation cycle and difficulty in quickly forming a large-scale usable orbital dataset. The other method uses generative models to generate orbital data, but the generated results are prone to defects such as inconsistency with the actual physical environment and distortion of dynamic characteristics.

[0004] Existing methods for selecting orbital samples often rely on a single metric, failing to verify the dynamic rationality of the samples in a multibody gravitational field, their compatibility with the engineering implementation conditions of deep space missions, or the existence of repetitive dynamic characteristics. This results in a large number of generated orbital samples that do not meet the physical constraints of deep space operation, are mismatched with mission engineering conditions, or contain a large amount of redundant data with highly similar dynamic characteristics. These samples are difficult to use directly for deep space orbit design and related model training, leading to low usability of the orbital data and low overall dataset quality.

[0005] Therefore, improving the usability and quality of generated deep space orbital data has become a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0006] This application provides a method, apparatus, device, and storage medium for generating deep space orbit data, which can generate deep space orbit data with high usability and quality.

[0007] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a method for generating deep space orbital data, including: Obtain Earth-Moon spatial periodic orbit data from publicly available data sources; Based on the aforementioned Earth-Moon space periodic orbit data, new orbit data samples are generated using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling. The new orbital data samples are sequentially judged to determine whether they conform to the consistency of deep space multi-field coupled dynamics, the compliance with deep space mission engineering constraints, and the uniqueness of deep space orbital dynamics characteristics. When the orbital data sample meets the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance, and deep space orbital dynamics uniqueness, the orbital data sample is determined to be compliant deep space orbital data.

[0008] Optionally, the step of generating new orbital data samples based on the Earth-Moon space periodic orbital data using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling includes: The orbital family identifier in the Earth-Moon space periodic orbit data is converted into a one-hot code and used as a conditional input. The conditional variational autoencoder constrained by deep space multibody dynamics is trained in combination with the Earth-Moon space periodic orbit data to obtain a family-specific conditional variational autoencoder. In the latent space of the family-specific conditional variational autoencoder, initial orbital parameters are generated by sampling using latent space noise, latent space interpolation, and prior sampling. Using a deep-space orbit adaptive step-size integrator, the initial orbit parameters are numerically integrated over the entire period according to the corresponding gravitational field model and perturbation term configuration to obtain a complete orbit point sequence and form a new orbit data sample.

[0009] Optionally, the process of determining whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics includes: Using the multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, the position and velocity components in the new orbital data sample are integrated over a complete orbital period using an adaptive step-size integrator to verify orbital closure, Jacobi integral conservation, long-duration perturbation stability, and integration error accumulation suppression characteristics. When the orbital closure, Jacobi integral conservation, long-endurance perturbation stability, and integral error accumulation suppression characteristics all meet the preset requirements, the new orbital data sample is determined to conform to the deep space multi-field coupled dynamics consistency.

[0010] Optionally, the process of determining whether the new orbital data sample meets the engineering constraints of the deep space mission includes: Using a multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, the position and velocity components in the new orbital data sample are integrated over a complete orbital period using an adaptive step-size integrator to verify the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics. When the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics all meet the preset engineering requirements of the deep space mission, the new orbital data sample is determined to conform to the engineering constraints of the deep space mission.

[0011] Optionally, the process of determining whether the new orbital data sample conforms to the uniqueness of deep space orbital dynamics characteristics includes: Four dynamic features—orbit period, Jacobian constant, stability index, and characteristic energy—are extracted from the new orbital data samples and the Earth-Moon space periodic orbital data. Compare the orbital period, Jacobian constant, stability index, and characteristic energy in the new orbital data sample and each Earth-Moon space periodic orbital data; When the differences in orbital period, Jacobian constant, stability index, and characteristic energy between the new orbital data sample and the Earth-Moon space periodic orbital data are all greater than a preset allowable threshold, the new orbital data sample is determined to meet the uniqueness of deep space orbital dynamics characteristics.

[0012] Optionally, the Earth-Moon space periodic orbit data includes: X-axis position component, Y-axis position component, Z-axis position component, X-axis velocity component, Y-axis velocity component, Z-axis velocity component, orbital period, Jacobian constant, orbital stability index, characteristic energy, total velocity increment, and orbital family identifier in the rotating center-of-mass coordinate system.

[0013] Optionally, after determining that the orbital data sample is deep-space orbital data that meets the requirements, the method further includes: The qualified deep space orbit data are classified and stored according to orbit family, gravitational field model and mission type, and supplemented into the benchmark dataset corresponding to the Earth-Moon space periodic orbit data to form an expanded deep space orbit dedicated dataset.

[0014] Secondly, this application provides a deep space orbital data generation apparatus, comprising: The orbital data acquisition module is used to acquire lunar spatial periodic orbital data from public data sources; The sample generation module is used to generate new orbit data samples based on the Earth-Moon space periodic orbit data by using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling. The judgment module is used to sequentially judge whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics, the compliance of deep space mission engineering constraints, and the uniqueness of deep space orbital dynamics characteristics. The data determination module is used to determine that the orbital data sample is a qualified deep space orbital data when the orbital data sample meets the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance and deep space orbital dynamics uniqueness.

[0015] Thirdly, this application provides a computing device, including a memory and a processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.

[0016] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.

[0017] As can be seen from the above technical solution, this application has at least the following beneficial effects: This application obtains lunar orbital data from publicly available data sources; based on the lunar orbital data, it generates new orbital data samples using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling; it then sequentially determines whether the new orbital data samples meet the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraints compliance, and deep space orbital dynamics uniqueness; when the orbital data samples meet the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraints compliance, and deep space orbital dynamics uniqueness, the orbital data samples are determined to be qualified deep space orbital data. Orbital data samples are generated using a conditional variational autoencoder constrained by deep-space multibody dynamics. The physical laws of the multibody gravitational environment in the Earth-Moon space are incorporated into the data generation stage, reducing or eliminating samples that do not conform to deep-space dynamics characteristics and improving the physical rationality of the generated orbital samples. A progressive judgment is then performed on the consistency of deep-space multi-field coupled dynamics, compliance with deep-space mission engineering constraints, and the uniqueness of deep-space orbital dynamic characteristics. First, the physical existence of the sample is verified; then, its engineering practicality is verified; and finally, its data value is verified. Invalid samples are filtered out layer by layer to ensure that the final retained orbital data simultaneously meets the physical constraints of deep-space operation and the engineering implementation requirements of deep-space missions. The generated, compliant deep-space orbital data can be directly used for Earth-Moon space periodic orbit design and related intelligent model training, reducing the workload of manual screening and correction, improving the usability of the orbital data and the overall quality of the dataset, and adapting to the low-cost, high-frequency orbital design requirements of Earth-Moon space exploration missions.

[0018] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a method for generating deep space orbital data provided in this application embodiment; Figure 2 A schematic diagram of a deep space orbit data generation device provided in an embodiment of this application; Figure 3 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation

[0020] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0021] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0022] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first: Deep space orbital data is a structured collection of data describing the trajectory and characteristics of spacecraft operating in the deep space environment, outside the Earth's atmosphere and outside the dominant near-Earth gravitational pull. Lunar periodic orbital data consists of orbital data operating within the multi-body gravitational field formed by the Earth and the Moon, exhibiting periodic closed-loop characteristics. It serves as fundamental data for orbital design, intelligent model training, and mission simulation for missions such as Lunar relay communication, deep space observation, and lunar exploration.

[0023] Existing methods for acquiring deep-space orbital data mainly fall into two categories: one relies on traditional numerical integration and optimization methods, which depend on manually setting initial values ​​and iterative corrections, resulting in a long generation cycle and making it difficult to quickly form a large-scale usable orbital dataset; the other uses generative models to generate orbital data, but the generated results are prone to discrepancies with the actual physical environment and distortion of dynamic characteristics. Deep-space orbital sample selection methods often employ single-index judgments, which cannot verify the dynamic rationality of samples in multi-body gravitational fields, whether they match the engineering implementation conditions of deep-space missions, or whether there is any duplication of dynamic characteristics. Consequently, many generated orbital samples do not meet the physical constraints of deep-space operation, do not match mission engineering conditions, or contain a large amount of redundant data with highly similar dynamic characteristics, making them difficult to directly use for deep-space orbital design and related model training. This results in low usability of the orbital data and low overall dataset quality.

[0024] In view of this, embodiments of this application provide a method for generating deep space orbit data, which can be executed by a processing device. This processing device can be a terminal or a server. Terminals include, but are not limited to, smartphones, tablets, laptops, personal digital assistants, or smart wearable devices. The server can be a cloud server, such as a central server in a central cloud computing cluster or an edge server in an edge cloud computing cluster. Alternatively, the server can be a server in a local data center. A local data center refers to a data center directly controlled by the user.

[0025] In view of this, this application provides a method for generating deep space orbit data. This method addresses the problems of dynamic distortion and low quality in generating Earth-Moon space orbit data using existing generative models. It embeds physical constraints during generation and employs a progressive, hierarchical filtering approach during verification. Deep space multibody dynamics constraints are embedded into the training process of a conditional variational autoencoder, enabling the model to learn the basic physical laws of the Earth-Moon multibody gravitational field while learning the orbit data distribution, thus reducing or eliminating the generation of samples that do not conform to dynamic characteristics. Combined with latent space multi-strategy sampling, the coverage of orbit data is expanded. During data verification, a three-level progressive judgment is used: first, the physical existence of the samples is verified, eliminating invalid samples that violate multibody gravitational laws; second, the engineering practicality of the samples is verified, selecting usable samples that match the boundary conditions of deep space missions; finally, the uniqueness of the samples is verified, eliminating redundant samples with repeated dynamic characteristics. The generated orbit data can be directly used for Earth-Moon space orbit design and intelligent model training, solving the problems of low availability and quality of existing orbit datasets and their difficulty in adapting to the needs of high-frequency deep space exploration missions.

[0026] To make the technical solution of this application clearer and easier to understand, a method for generating deep space orbital data provided by an embodiment of this application is described below with reference to the accompanying drawings. Figure 1 As shown, this figure is a flowchart of a deep space orbit data generation method provided in an embodiment of this application. The method includes: S101 obtains lunar orbital data from publicly available data sources.

[0027] A lunar periodic orbit refers to a closed orbit in which a spacecraft's position and velocity return to their initial state after a fixed time interval within the multibody gravitational field formed by the Earth and the Moon. Publicly available data sources include the official website of NASA's Jet Propulsion Laboratory, the European Space Agency's SPICE deep space exploration core library, and publicly available data sources from domestic deep space exploration projects. The acquired lunar periodic orbit data includes the X-axis position components, Y-axis position components, and Z-axis position components in a rotating center-of-mass coordinate system, as well as the X-axis velocity components, Y-axis velocity components, and Z-axis velocity components, orbital period, Jacobian constant, orbital stability index, characteristic energy, total velocity increment, and orbital family identifier. The rotating center-of-mass coordinate system is the standard coordinate system for calculating deep space translational point orbits. Its origin is located at the common center of mass of the Earth and the Moon. The X-axis points towards the Moon along the line connecting the Earth and the Moon, the Z-axis is perpendicular to the lunar orbital plane, and the Y-axis is determined by the right-hand rule. The Jacobian constant is the only conserved quantity of motion in the restricted three-body problem, and its value determines the range of space a spacecraft can reach without additional fuel. The orbital stability index characterizes the orbit's resistance to initial small disturbances; the smaller the index, the more stable the orbit. Characteristic energy characterizes the energy required for a spacecraft to escape Earth's gravitational field, directly corresponding to the launch vehicle's maximum payload capacity. Total velocity increment is the sum of all changes in maneuver velocity required by the spacecraft throughout the mission, directly determining its fuel consumption. Orbital family identifiers are used to distinguish different types of Earth-Moon periodic orbits. For example, this embodiment obtained 32 types of Earth-Moon periodic orbits, totaling 183,695 samples, from the official website of NASA's Jet Propulsion Laboratory, covering Halo orbits, Lyapunov orbits, vertical orbits, and pseudo-halo orbits at the Earth-Moon system's L1, L2, L4, and L5 translation points.

[0028] It should be noted that the acquisition of the above data was all permitted by the data source providers. By obtaining comprehensive lunar spatial periodic orbit data from multiple authoritative and publicly available data sources, high-quality sample data can be provided for the model training process, ensuring that the generated orbit data is physically feasible.

[0029] S102, Based on the Earth-Moon space periodic orbit data, new orbit data samples are generated by combining a conditional variational autoencoder constrained by deep space multibody dynamics with latent space sampling.

[0030] A Conditional Variational Autoencoder (CVA) is a generative deep learning model consisting of an encoder and a decoder. It maps high-dimensional input data to a low-dimensional latent space and samples from this latent space to generate new samples that conform to the original data distribution. Deep space multibody dynamics constraints refer to embedding the physical laws of the Earth-Moon multibody gravitational field into the training process of the CVA, ensuring that the samples generated by the model conform to fundamental celestial mechanics laws. The latent space is a low-dimensional feature space obtained by compressing the high-dimensional original input data by the encoder; points that are close together in this space correspond to similar data samples.

[0031] Specifically, the orbital family identifiers in the Earth-Moon space-time periodic orbit data are first converted into one-hot codes and used as conditional inputs. These codes are then combined with the Earth-Moon space-time periodic orbit data to train a conditional variational autoencoder constrained by deep-space multibody dynamics, resulting in a family-specific conditional variational autoencoder. One-hot coding is a method of converting discrete categorical variables into binary vectors, enabling the model to better handle categorical features.

[0032] During training, the model's total loss function consists of reconstruction loss, annealed KL divergence, and deep space multibody dynamics loss. KL divergence measures the difference between two probability distributions; annealing the KL divergence helps prevent posterior collapse during training. Deep space multibody dynamics loss constrains the dynamic characteristics of generated samples, ensuring they conform to the fundamental laws of multibody gravitational fields. By using orbital family identifiers as input and embedding them into deep space multibody dynamics constraints, the model can generate specific samples for different orbital families, reducing or eliminating samples that do not conform to dynamic characteristics and improving the physical plausibility of the generated samples.

[0033] Then, in the latent space of the family-specific conditional variational autoencoder, initial orbital parameters are generated by sampling using latent space noise addition, latent space interpolation, and prior sampling. Latent space noise addition involves adding small amounts of random noise to points in the latent space, generating new samples similar to but different from the original samples. Latent space interpolation involves linear interpolation between two known points in the latent space, generating new samples located between the two samples. Prior sampling involves directly sampling from the prior distribution of the latent space to generate samples from areas not covered by the existing dataset. In this embodiment, as an example, latent space noise addition can account for 60% to generate translational point periodic orbits with different amplitudes within the same family; latent space interpolation can account for 30% to generate Earth-Moon transfer orbits with different transfer durations; and prior sampling can account for 10% to supplement the orbital feasible solution space not covered by the existing dataset. By employing three different latent space sampling strategies, the coverage of orbital data can be effectively expanded, generating more diverse orbital samples.

[0034] Finally, using a deep-space orbit adaptive step-size integrator, the initial orbital parameters are numerically integrated over the entire period according to the corresponding gravitational field model and perturbation term configuration to obtain a complete orbital point sequence, forming a new orbital data sample. The deep-space orbit adaptive step-size integrator is a numerical integration method that can automatically adjust the integration step size based on integration error. This embodiment uses the RK45 adaptive step-size integrator, which can improve computational efficiency while ensuring integration accuracy. The gravitational field model is used to describe the gravitational environment of the spacecraft. The gravitational field models used in this embodiment include the circular restricted three-body model, the elliptical restricted three-body model, and the restricted four-body model. Perturbation terms refer to all forces acting on the spacecraft other than the central gravity of the primary and secondary celestial bodies, mainly including non-spherical gravitational perturbations, solar gravitational perturbations, and solar radiation pressure perturbations. For example, for a 12-day period Halo orbit at the L1 point of the Earth-Moon system, the corresponding gravitational field model is the circular restricted three-body model, and the perturbation configuration is the essential perturbation term plus the solar gravitational perturbation.

[0035] By employing a gravitational field model and perturbation configuration corresponding to the orbit type for numerical integration, accurate orbital trajectories can be obtained, making the dynamic properties of the generated orbital data more reliable.

[0036] S103, sequentially determine whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics, the compliance of deep space mission engineering constraints, and the uniqueness of deep space orbital dynamic characteristics.

[0037] The reason for prioritizing deep-space multi-field coupled dynamics consistency assessment before deep-space mission engineering constraint compliance assessment is that deep-space orbital data is significant because it conforms to fundamental laws of celestial mechanics and can guide actual spacecraft operations. Deep-space multi-field coupled dynamics consistency assessment determines the physical legitimacy of the sample, a prerequisite for the existence of deep-space orbital data; while deep-space mission engineering constraint compliance assessment determines the engineering usability of the sample, a secondary screening based on physical rationality. If a sample fails the deep-space multi-field coupled dynamics consistency assessment, it indicates that it does not meet the basic dynamic constraints of the circular restricted three-body problem, cannot form a stable closed trajectory, and is essentially not valid periodic orbit data. At this point, conducting an engineering applicability assessment yields fuel consumption, field-of-view coverage, and other indicators based on incorrect dynamic properties, rendering them meaningless. Furthermore, approximately 30%-40% of the orbital samples generated by generative models exhibit dynamic defects, such as non-closed trajectories, energy drift, and instability anomalies. If these samples are not filtered through the deep space multi-field coupled dynamics consistency judgment first, physical errors will be transmitted to the deep space mission engineering constraint compliance judgment stage, resulting in distorted engineering evaluation results. Thus, it will be impossible to distinguish whether the abnormal indicators caused by physical errors are due to differences in engineering performance itself. Furthermore, the model trained based on this dataset will learn both the erroneous dynamic laws and the engineering evaluation logic.

[0038] Specifically, the first step is to determine whether the new orbital data sample conforms to the consistency of deep-space multi-field coupled dynamics. The consistency of deep-space multi-field coupled dynamics refers to the orbital sample conforming to the fundamental physical laws of the complex multi-body gravitational field and perturbation environment in deep space, enabling it to form a stable closed trajectory. The specific process involves using a multi-body gravitational field model and perturbation configuration corresponding to the new orbital data sample. An adaptive step-size integrator is used to integrate the position and velocity components of the new orbital data sample over a complete orbital period, verifying orbital closure, Jacobi integral conservation, long-duration perturbation stability, and integration error accumulation suppression characteristics. Orbital closure refers to the degree to which the spacecraft's position and velocity coincide with its initial state after a complete orbital period. Jacobi integral conservation refers to the characteristic that the Jacobi constant remains unchanged during orbital operation. Long-duration perturbation stability refers to the orbit's ability to resist various perturbation effects during long-term operation. Integration error accumulation suppression characteristics refer to the rate at which the error increases with integration time during numerical integration. When the orbital closure, Jacobi integral conservation, long-endurance perturbation stability, and integral error accumulation suppression characteristics all meet the preset requirements, the new orbital data sample is determined to conform to the deep space multi-field coupled dynamics consistency.

[0039] In the embodiments provided in this application, the deep space multi-field coupled dynamics consistency score of the new orbital data sample can be calculated first. ,exist ≥ Deep Space Multi-Field Coupling Dynamics Consistency Threshold At that time, it was determined that the new orbital data samples conformed to the consistency of deep space multi-field coupled dynamics. < If the new orbital data sample does not conform to the consistency of deep space multi-field coupled dynamics, the sample will be discarded.

[0040] Consistency Score of Deep Space Multi-Field Coupled Dynamics The calculation formula is:

[0041] in, , , and Let be the weighting coefficient for each sub-item score, and + + =1, which can be adjusted according to the type of deep space orbit; The formula for calculating the orbital closure score is as follows:

[0042] in, The 6-dimensional state vector is obtained by integrating one orbital period T using the RK45 integrator with the corresponding gravitational field model. The six-dimensional state vector at the initial moment of the orbit consists of the X-axis position component, the Y-axis position component, the Z-axis position component, the X-axis velocity component, the Y-axis velocity component, and the Z-axis velocity component in the rotating centroid coordinate system. The 2-norm of a vector; The Jacobi integral conservation score is calculated using the following formula:

[0043] in, Let be the Jacobi constant obtained by integrating the orbit to time T. is the Jacobi constant at the initial moment of the orbit, and the Jacobi constant is the only conserved quantity for deep space orbits in the restricted three-body problem; The formula for calculating the long-endurance perturbation stability score is as follows:

[0044] in, The stability index for generating samples is calculated by integrating a single-valued matrix, representing the sensitivity of the orbit to initial perturbations; The average stability index corresponding to the deep space orbit family is calculated from the Earth-Moon space periodic orbit data obtained in step S101; The formula for calculating the integral error cumulative suppression score is as follows:

[0045] Where N is the number of discrete integration moments throughout the entire orbital period; Let be the Jacobi constant at the k-th discrete time. is the Jacobi constant at the initial moment of the orbit; It should be noted that the consistency threshold of deep space multi-field coupled dynamics The preset thresholds for each component of the deep space multi-field coupled dynamics consistency can be set. The preset threshold for orbit closure, based on the convergence accuracy requirements of the traditional multiple-target method, is set to the relative deviation between the initial and final states being less than 1×10 after one complete orbital integration cycle. -6 This accuracy meets the basic requirements of deep space exploration missions for orbital position and velocity; the preset threshold for the conservation of the Jacobian integral, based on the accuracy control requirements of long-duration numerical integration, is set to a relative deviation of the Jacobian constant of less than 1 × 10⁻⁶ over the entire period. -5 To ensure the stability of orbital energy characteristics during operation, a preset threshold for long-duration perturbation stability is set based on the statistical distribution of stability indices in a benchmark dataset of the same orbital family. This threshold is set to no more than twice the average stability index of that orbital family; orbits exceeding this range will not meet mission requirements for long-term operational stability. A preset threshold for suppressing cumulative integral errors is set to ensure that the average relative deviation of the Jacobian constant at all discrete moments throughout the entire cycle is less than 1 × 10⁻⁶. -6 This is used to control the degree of error accumulation during numerical integration. For example, for the Halo orbit with a 12-day period at the L1 point of the Earth-Moon system, its orbital closure threshold is 1 × 10⁻⁶. -6 The Jacobi integral conservation threshold is 1 × 10⁻⁶. -5 The long-endurance perturbation stability threshold is 2360.8, and the integral error accumulation suppression threshold is 1×10. -6 By combining the thresholds of these sub-items with their corresponding weights, the consistency threshold of deep space multi-field coupled dynamics is obtained. .

[0046] By verifying the dynamic consistency of orbits from multiple dimensions, invalid samples that violate the laws of multibody gravity can be eliminated, thus ensuring that the retained samples are real physical orbits.

[0047] For samples that conform to the consistency of deep-space multi-field coupled dynamics, the following assessment is made regarding their compliance with deep-space mission engineering constraints. Deep-space mission engineering constraint compliance refers to the ability of the orbital sample to adapt to the engineering boundary conditions of the deep-space exploration mission. Specifically, the process involves using a multi-body gravitational field model and perturbation configuration corresponding to the new orbital data sample, and integrating the position and velocity components of the new orbital data sample over a complete orbital period using an adaptive step-size integrator. This verifies the deep-space transfer characteristics, fuel consumption characteristics, deep-space telemetry and communication characteristics, and flyby / capture constraint characteristics. Deep-space transfer characteristics refer to the degree of matching between the orbit and the launch vehicle's launch capability. Fuel consumption characteristics refer to the degree of matching between the total fuel consumption required for the orbit and the spacecraft's fuel carrying capacity. Deep-space telemetry and communication characteristics refer to the continuity of communication with ground-based deep-space stations during orbital operation. Flyby / capture constraint characteristics refer to the degree of matching between the orbit and the rendezvous conditions of the target celestial body. When the deep-space transfer characteristics, fuel consumption characteristics, deep-space telemetry and communication characteristics, and flyby / capture constraint characteristics all meet the preset engineering requirements of the deep-space mission, the new orbital data sample is determined to meet the deep-space mission engineering constraint compliance.

[0048] In the embodiments provided in this application, the deep space mission engineering constraint compliance score can be calculated first. ,exist ≥Deep Space Mission Engineering Constraint Compliance Threshold At that time, it was determined that the new orbital data samples met the engineering constraints of the deep space mission. < If the new orbital data sample does not meet the deep space mission engineering constraints, the sample will be removed.

[0049] Deep Space Mission Engineering Constraint Compliance Score The calculation formula is:

[0050] in, , , and Let be the weighting coefficient for each sub-item score, and + + =1, which can be adjusted according to the type of deep space mission; The formula for calculating the deep space transfer characteristic compliance score is as follows:

[0051] Wherein, C3 is the feature energy of the generated sample, representing the carrying capacity requirement of the deep space launch mission; The characteristic energy required for the corresponding deep space mission type can be obtained statistically from the Earth-Moon space periodic orbit data acquired in step S101. The formula for calculating fuel consumption compliance score is as follows:

[0052] in, This refers to the total velocity increment required for orbit maintenance, perturbation compensation, and maneuver correction throughout the entire deep space orbit cycle. This represents the maximum speed increment allowed for a deep space mission. The formula for calculating the deep space tracking, telemetry, and command (STDM) compliance score is as follows:

[0053] in, This refers to the duration of continuous tracking and control that meets the elevation angle constraints of the ground-based deep space station and is unobstructed by celestial bodies within the orbital operation cycle. This refers to the total duration of the entire orbital cycle; The formula for calculating the flyby / capture constraint compliance score is as follows:

[0054] in, This refers to the altitude of the pericenter of a planet's flyby orbit, or the altitude of the entry point of a target object's capture orbit. This represents the median value of the altitude range required for deep space missions.

[0055] It should be noted that the deep space mission engineering constraint compliance threshold These thresholds can be set based on the pre-defined thresholds for each component of the deep space mission's engineering constraint compliance. These pre-defined thresholds are based on the actual engineering boundary conditions of different types of deep space exploration missions. The pre-defined threshold for deep space transfer characteristics, based on the Earth-Moon transfer launch capabilities of existing mainstream launch vehicles, is set at a characteristic energy C3 between -2.0 and 0 km² / s². Orbits exceeding this range cannot be launched into orbit using existing launch vehicles. The pre-defined threshold for fuel consumption characteristics, based on the spacecraft's fuel carrying capacity and mission lifespan requirements, is set at no more than 100 m / s for the total velocity increment during a 10-year mission lifespan for Earth-Moon relay missions, and no more than 200 m / s for lunar exploration missions within a 5-year mission lifespan. The pre-defined threshold for deep space telemetry and communication characteristics, based on the coverage capability of the global deep space telemetry and communication network, is set at a minimum of 80% of the continuous telemetry and communication time meeting ground station elevation angle constraints during the orbital operating cycle, ensuring continuous status monitoring and command transmission from the ground. The preset thresholds for flyby / capture constraint characteristics, based on the safety requirements of rendezvous with the target celestial body, are set between 100 and 500 km for lunar capture missions. Too low a threshold would lead to a collision with the lunar surface, while too high a threshold would prevent effective capture by lunar gravity. Combining these sub-thresholds with their corresponding weights yields the deep space mission engineering constraint compliance thresholds. .

[0056] By verifying the compliance of orbits with engineering constraints, we can select orbital samples that can be practically applied to deep space exploration missions, thus avoiding the generation of samples that are physically feasible but not engineering-feasible.

[0057] For samples that meet the engineering constraints for deep space missions, the next step is to determine whether they meet the uniqueness requirement for deep space orbital dynamics characteristics. Uniqueness of deep space orbital dynamics characteristics means that the main dynamic properties of the orbital sample differ significantly from all existing orbital samples. Specifically, this involves extracting four dynamic characteristics—orbital period, Jacobian constant, stability index, and characteristic energy—from the new orbital data sample and the Earth-Moon space-period orbital data. The orbital period, Jacobian constant, stability index, and characteristic energy of the new orbital data sample are then compared with those of each Earth-Moon space-period orbital data set. If the differences in orbital period, Jacobian constant, stability index, and characteristic energy between the new orbital data sample and the Earth-Moon space-period orbital data all exceed a preset allowable threshold, the new orbital data sample is determined to meet the uniqueness requirement for deep space orbital dynamics characteristics.

[0058] In the embodiments provided in this application, the dynamic characteristic distance D between the new orbital data sample and all samples in the Earth-Moon space periodic orbital data can be calculated. If D > the preset allowable threshold for the uniqueness of deep space orbital dynamic characteristics... If the new orbital data sample conforms to the uniqueness of deep space orbital dynamics characteristics, then D ≤ the preset allowable threshold for the uniqueness of deep space orbital dynamics characteristics. If the sample is not found to be a duplicate, it will be discarded. The formula for calculating the dynamic feature distance D is:

[0059] in, , , and Let be the weight coefficients of each feature, and + + =1; For the orbital period of the new orbital data sample, The orbital period in the Earth-Moon space periodic orbit data; The Jacobi constant for the new orbital data samples; The Jacobi constant is used in the periodic orbital data of the Earth and Moon. A stability index for new orbital data samples; A stability index in the periodic orbital data of the Earth and Moon in space; The characteristic energy of the new orbital data sample; This refers to the characteristic energy in the Earth-Moon space periodic orbit data.

[0060] It should be noted that the uniqueness of deep space orbital dynamics characteristics is subject to a preset allowable threshold. The settings can be based on the statistical characteristics of the Earth-Moon spatial periodic orbit data obtained in step S101. First, for all samples within the same orbital family, the distances between each pair of samples are calculated for four dynamic characteristics: orbital period, Jacobian constant, stability index, and characteristic energy. This yields the minimum characteristic distance between different orbits within the same family. 0.8 times this minimum characteristic distance is then used as the preset threshold for the third-level judgment. For example, statistical calculations on the baseline dataset of the Halo orbit at the Earth-Moon L1 point show that the minimum characteristic distance between different orbits within the same family is 0.05; therefore, the preset allowable threshold is 0.04. This setting method can remove duplicate samples with highly similar dynamic characteristics while retaining all valid samples with significantly different dynamic characteristics. This avoids redundant samples in the dataset and prevents the loss of valid sample data.

[0061] The four dynamic characteristics of an orbit—orbital period, Jacobian constant, stability index, and characteristic energy difference—together determine the essential dynamic behavior of an orbit. By making uniqueness judgments based on these four characteristics, duplicate samples with highly similar dynamic characteristics can be accurately eliminated, thereby improving the information density of the dataset.

[0062] S104, when the orbital data sample meets the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance and deep space orbital dynamics uniqueness, the orbital data sample is determined to be deep space orbital data that meets the requirements.

[0063] In the embodiments provided in this application, after determining that the orbital data samples are qualified deep-space orbital data, the qualified deep-space orbital data can be classified and stored according to orbital family, gravitational field model, and mission type, and supplemented into the benchmark dataset corresponding to the Earth-Moon space periodic orbital data to form an expanded deep-space orbital-specific dataset. By classifying and storing qualified orbital data and supplementing it into the benchmark dataset, the deep-space orbital-specific dataset can be continuously enriched and improved, providing more high-quality data samples for orbital design and model training.

[0064] Based on the above description, this application has the following beneficial effects: This application provides a method for generating deep space orbit data. The method involves acquiring periodic Earth-Moon orbit data from a publicly available data source; generating new orbit data samples based on the Earth-Moon periodic orbit data using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling; sequentially determining whether the new orbit data samples meet the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraints compliance, and deep space orbit dynamics uniqueness; and determining the orbit data samples as compliant deep space orbit data when they meet these requirements. Orbital data samples are generated using a conditional variational autoencoder constrained by deep-space multibody dynamics. The physical laws of the multibody gravitational environment in the Earth-Moon space are incorporated into the data generation stage, reducing or eliminating samples that do not conform to deep-space dynamics characteristics and improving the physical rationality of the generated orbital samples. A progressive judgment is performed on the consistency of deep-space multi-field coupled dynamics, compliance with deep-space mission engineering constraints, and the uniqueness of deep-space orbital dynamic characteristics. First, the physical existence of the sample is verified; then, its engineering practicality is verified; and finally, its data value is verified. Invalid samples are filtered out layer by layer to ensure that the final retained orbital data simultaneously meets the physical constraints of deep-space operation and the engineering implementation requirements of deep-space missions. The generated, compliant deep-space orbital data can be directly used for Earth-Moon space periodic orbit design and related intelligent model training, reducing the workload of manual screening and correction, improving the usability of orbital data and the overall quality of the dataset, and adapting to the low-cost, high-frequency orbital design requirements of Earth-Moon space exploration missions.

[0065] The above text combined Figure 1 The method for generating deep space orbit data provided in the embodiments of this application has been described in detail. The apparatus and equipment provided in the embodiments of this application will be described below with reference to the accompanying drawings.

[0066] like Figure 2 As shown in the figure, this is a schematic diagram of a deep space orbit data generation device provided in an embodiment of this application. The deep space orbit data generation device 200 includes: The orbital data acquisition module 201 is used to acquire lunar spatial periodic orbital data from public data sources. Sample generation module 202 is used to generate new orbit data samples based on the Earth-Moon space periodic orbit data by using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling. The judgment module 203 is used to sequentially judge whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics, the conformity of deep space mission engineering constraints, and the uniqueness of deep space orbital dynamic characteristics; The data determination module 204 is used to determine that the orbital data sample is a qualified deep space orbital data when the orbital data sample meets the requirements of deep space multi-field coupling dynamics consistency, deep space mission engineering constraint compliance and deep space orbital dynamics uniqueness.

[0067] Optionally, the sample generation module 202 is specifically used for: The orbital family identifier in the Earth-Moon space periodic orbit data is converted into a one-hot code and used as a conditional input. The conditional variational autoencoder constrained by deep space multibody dynamics is trained in combination with the Earth-Moon space periodic orbit data to obtain a family-specific conditional variational autoencoder. In the latent space of the family-specific conditional variational autoencoder, initial orbital parameters are generated by sampling using latent space noise, latent space interpolation, and prior sampling. Using a deep-space orbit adaptive step-size integrator, the initial orbit parameters are numerically integrated over the entire period according to the corresponding gravitational field model and perturbation term configuration to obtain a complete orbit point sequence and form a new orbit data sample.

[0068] Optionally, the judgment module 203 includes: The dynamic consistency judgment unit is used to integrate the position and velocity components in the new orbital data sample for one complete orbital period using a multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, through an adaptive step-size integrator. This verifies orbital closure, Jacobian integral conservation, long-endurance perturbation stability, and integration error accumulation suppression characteristics. When the orbital closure, Jacobian integral conservation, long-endurance perturbation stability, and integration error accumulation suppression characteristics all meet preset requirements, the new orbital data sample is determined to conform to deep space multi-field coupled dynamic consistency.

[0069] Optionally, the judgment module 203 includes: The deep space mission engineering constraint compliance judgment unit is used to integrate the position and velocity components in the new orbital data sample for one complete orbital period using a multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, through an adaptive step-size integrator. This verifies the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics. When the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics all meet the preset engineering requirements of the deep space mission, the new orbital data sample is determined to meet the deep space mission engineering constraint compliance.

[0070] Optionally, the judgment module 203 includes: The deep space orbital dynamics feature uniqueness determination unit is used to extract four dynamic features—orbital period, Jacobian constant, stability index, and characteristic energy—from the new orbital data sample and the Earth-Moon space periodic orbital data; compare the orbital period, Jacobian constant, stability index, and characteristic energy of the new orbital data sample with those of each Earth-Moon space periodic orbital data; when the differences in orbital period, Jacobian constant, stability index, and characteristic energy between the new orbital data sample and the Earth-Moon space periodic orbital data are all greater than a preset allowable threshold, it is determined that the new orbital data sample meets the requirements for deep space orbital dynamics feature uniqueness.

[0071] Optionally, the orbit data acquisition module 201 is specifically used to acquire the X-axis position component, Y-axis position component, Z-axis position component, X-axis velocity component, Y-axis velocity component, Z-axis velocity component, orbit period, Jacobian constant, orbit stability index, characteristic energy, total velocity increment, and orbit family identifier from a public data source.

[0072] Optionally, the deep space orbit data generation device 200 further includes a dataset acquisition module, which is used to classify and store the qualified deep space orbit data according to orbit family, gravitational field model and mission type, and supplement it to the benchmark dataset corresponding to the Earth-Moon space periodic orbit data to form an expanded deep space orbit dedicated dataset.

[0073] The deep space orbit data generation apparatus according to the embodiments of this application can correspondingly execute the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the deep space orbit data generation apparatus are respectively for implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.

[0074] This application also provides a computing device.

[0075] like Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 700 includes a bus 701, a processor 702, a communication interface 703, and a memory 704. The processor 702, the memory 704, and the communication interface 703 communicate with each other via the bus 701.

[0076] The 701 bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0077] The processor 702 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0078] The communication interface 703 is used for communication with external devices.

[0079] Memory 704 may include volatile memory, such as random access memory (RAM). Memory 704 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0080] The memory 704 stores executable code, and the processor 702 executes the executable code to perform the aforementioned deep space orbit data generation method.

[0081] Specifically, in achieving Figure 2 In the case of the illustrated embodiment, and Figure 2 When the modules or units of the deep space orbit data generation device described in the embodiments are implemented by software, the execution... Figure 2The software or program code required for the functions of each module / unit can be partially or entirely stored in memory 704. Processor 702 executes the program code corresponding to each unit stored in memory 704 to execute the aforementioned deep space orbit data generation method.

[0082] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium capable of being stored by a computing device, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct a computing device to execute the aforementioned deep space orbit data generation method.

[0083] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0084] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0085] When the computer program product is executed by a computer, the computer performs any of the aforementioned deep space orbit data generation methods. The computer program product can be a software installation package; when any of the aforementioned deep space orbit data generation methods needs to be used, the computer program product can be downloaded and executed on the computer.

[0086] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for generating deep space orbital data, characterized in that, The method includes: Obtain lunar orbital data from publicly available data sources; Based on the aforementioned Earth-Moon space periodic orbit data, new orbit data samples are generated using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling. The new orbital data samples are sequentially judged to determine whether they conform to the consistency of deep space multi-field coupled dynamics, the compliance with deep space mission engineering constraints, and the uniqueness of deep space orbital dynamics characteristics. When the orbital data sample meets the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance, and deep space orbital dynamics uniqueness, the orbital data sample is determined to be compliant deep space orbital data.

2. The method according to claim 1, characterized in that, The method of generating new orbital data samples based on the Earth-Moon space periodic orbital data using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling includes: The orbital family identifier in the Earth-Moon space periodic orbit data is converted into a one-hot code and used as a conditional input. The conditional variational autoencoder constrained by deep space multibody dynamics is trained in combination with the Earth-Moon space periodic orbit data to obtain a family-specific conditional variational autoencoder. In the latent space of the family-specific conditional variational autoencoder, initial orbital parameters are generated by sampling using latent space noise, latent space interpolation, and prior sampling. Using a deep-space orbit adaptive step-size integrator, the initial orbit parameters are numerically integrated over the entire period according to the corresponding gravitational field model and perturbation term configuration to obtain a complete orbit point sequence and form a new orbit data sample.

3. The method according to claim 1, characterized in that, The process of determining whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics includes: Using the multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, the position and velocity components in the new orbital data sample are integrated over a complete orbital period using an adaptive step-size integrator to verify orbital closure, Jacobi integral conservation, long-duration perturbation stability, and integration error accumulation suppression characteristics. When the orbital closure, Jacobi integral conservation, long-endurance perturbation stability, and integral error accumulation suppression characteristics all meet the preset requirements, the new orbital data sample is determined to conform to the deep space multi-field coupled dynamics consistency.

4. The method according to claim 1, characterized in that, The process of determining whether the new orbital data sample meets the engineering constraints of the deep space mission includes: Using a multibody gravitational field model and perturbation configuration corresponding to the new orbital data sample, the position and velocity components in the new orbital data sample are integrated over a complete orbital period using an adaptive step-size integrator to verify the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics. When the deep space transfer characteristics, fuel consumption characteristics, deep space telemetry and communication characteristics, and flyby / capture constraint characteristics all meet the preset engineering requirements of the deep space mission, the new orbital data sample is determined to conform to the engineering constraints of the deep space mission.

5. The method according to claim 1, characterized in that, The process of determining whether the new orbital data sample conforms to the uniqueness of deep space orbital dynamics characteristics includes: Four dynamic features—orbit period, Jacobian constant, stability index, and characteristic energy—are extracted from the new orbital data samples and the Earth-Moon space periodic orbital data. Compare the orbital period, Jacobian constant, stability index, and characteristic energy in the new orbital data sample and each Earth-Moon space periodic orbital data; When the differences in orbital period, Jacobian constant, stability index, and characteristic energy between the new orbital data sample and the Earth-Moon space periodic orbital data are all greater than a preset allowable threshold, the new orbital data sample is determined to meet the uniqueness of deep space orbital dynamics characteristics.

6. The method according to claim 1, characterized in that, The Earth-Moon space periodic orbit data includes: X-axis position component, Y-axis position component, Z-axis position component, X-axis velocity component, Y-axis velocity component, Z-axis velocity component, orbital period, Jacobian constant, orbital stability index, characteristic energy, total velocity increment, and orbital family identifier in the rotating center-of-mass coordinate system.

7. The method according to claim 1, characterized in that, After determining that the orbital data sample is deep space orbital data that meets the requirements, the method further includes: The qualified deep space orbit data are classified and stored according to orbit family, gravitational field model and mission type, and supplemented into the benchmark dataset corresponding to the Earth-Moon space periodic orbit data to form an expanded deep space orbit dedicated dataset.

8. A deep space orbit data generation device, characterized in that, The device includes: The orbital data acquisition module is used to acquire lunar spatial periodic orbital data from public data sources; The sample generation module is used to generate new orbit data samples based on the Earth-Moon space periodic orbit data by using a conditional variational autoencoder constrained by deep space multibody dynamics combined with latent space sampling. The judgment module is used to sequentially judge whether the new orbital data sample conforms to the consistency of deep space multi-field coupled dynamics, the compliance of deep space mission engineering constraints, and the uniqueness of deep space orbital dynamics characteristics. The data determination module is used to determine that the orbital data sample is a qualified deep space orbital data when the orbital data sample meets the requirements of deep space multi-field coupled dynamics consistency, deep space mission engineering constraint compliance and deep space orbital dynamics uniqueness.

9. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 7.

Images