A flexible resource allocation method and system for massive orbit prediction calculations

Abstract

The present invention discloses a flexible resource allocation method and system for massive orbit prediction calculations, wherein the method includes the following steps: S1. virtualizing all available resources and providing the hardware foundation for service operation in the form of virtual machines through a virtual machine platform; S2. encapsulating the orbit calculation algorithm into a container image and storing it in an image warehouse. After the virtual machine platform is driven by the scheduling computing service to allocate a new virtual machine, the latest copy of the orbit calculation algorithm is obtained from the image warehouse and deployed and registered through a service registration center; S3. External business software submits orbit calculation requirements through an API gateway, and the orbit calculation requirements are distributed to existing computing resources in a load-balanced manner through the scheduling computing service, and finally the orbit calculation is completed and the calculation results are returned. The present invention can significantly improve the efficiency of massive satellite orbit calculations through scalable software and hardware resource scheduling.

Description

Technical Field

[0001] The present invention relates to the technical field of giant satellite constellation operation control, and in particular to a flexible resource allocation method and system for massive orbit prediction calculations. Background Art

[0002] Current space activities are showing many new characteristics. For example, small satellites such as CubeSats have been rapidly developed and deployed. Many commercial companies (SpaceX of the United States, OneWeb of the United Kingdom, Boeing, etc.) have proposed and gradually implemented large-scale low-orbit satellite constellation plans.

[0003] The working mode of large low-orbit constellations can break through the limitations of traditional single-platform satellites, achieve three-dimensional detection of the same target at the same time, improve the temporal and spatial coverage of regions or targets, and provide large apertures and long measurement limits. It is of great significance in communications, earth observation, navigation, stereo imaging, precise positioning, atmospheric astronomy and geophysical observation and other fields.

[0004] These constellations, with thousands or even tens of thousands of satellites, require a correspondingly large number of ground control stations. The operation and maintenance of these ultra-mega constellations presents immense pressure. To effectively monitor the operational status of satellites in orbit, deduce constellation mission performance and satellite control windows, and support the planning and development of ground-based operations and management plans, it is necessary to update the orbital and transit information of all satellites in the constellation within a certain period of time. Furthermore, recalculation is required after satellite maneuvers.

[0005] The control requirements of large low-orbit constellations include: constellation configuration maintenance, satellite attitude deflection, collision warning and avoidance, command uplink and telemetry downlink, etc., which brings about the need for massive satellite orbit calculations, including satellite ephemeris, six orbital numbers, sub-satellite point trajectory, and designated station transit time within a specified time range. The main functions are as follows: 1) Calculation of satellite operating environment, such as perturbation factors such as Earth perturbation, atmospheric drag, and solar pressure; 2) Calculation of orbital dynamics and attitude dynamics of each satellite in the constellation, using multiple algorithms such as J2, J4, SGP4 / SDP4, and HPOP to predict satellite orbits under different accuracy requirements; 3) Calculation of the operating orbits of other spacecraft and space debris, as well as collision probability calculation and warning; 4) Parameter configuration of ground station data.

[0006] In summary, the control and management of large low-Earth orbit constellations relies on precise predictions of the satellite's orbital position over a period of time. Low-Earth orbit satellites orbit the Earth's center under the influence of Earth's gravity and other factors (such as engine thrust and atmospheric drag). Their future orbital positions are completely determined by environmental conditions and initial conditions. The satellite's initial orbit can be described using six orbital elements, position and velocity information, and two-line elements (Two Line Elements, published regularly by NORAD). Environmental conditions must account for perturbations due to Earth's gravity, atmospheric drag, gravitational perturbations from other celestial bodies, and solar radiation pressure.

[0007] Because each satellite is independent and their orbit calculations do not interfere with each other, there are currently two main approaches to addressing the time-consuming orbit calculation problem for massive satellites. One approach involves employing a heterogeneous computing architecture combining graphics processing units (GPUs) and central processing units (CPUs), leveraging the GPU's thousands of cores for parallel computing and reducing computation time. The other approach involves deploying orbit calculations across multiple computers or virtual machines. When business software generates computational needs, the tasks are evenly distributed across the orbit calculation services. Both approaches require extensive hardware dedicated to orbit calculations for massive satellites. However, orbit predictions for these massive satellites are tidal, requiring only rapid updates during high tide within a certain period. Hardware resources are idle and wasted after low tide. Summary of the Invention

[0008] In order to solve the above problems, the present invention proposes a flexible resource allocation method and system for massive orbit prediction calculations, aiming to significantly improve the efficiency of massive satellite orbit calculations through scalable software and hardware resource scheduling.

[0009] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0010] A flexible resource allocation method for massive orbit prediction calculations includes the following steps:

[0011] S1. Virtualize all available resources and provide the hardware foundation for external services in the form of virtual machines through a virtual machine platform;

[0012] S2. Encapsulate the orbital calculation algorithm into a container image and store it in the image repository. After the virtual machine platform allocates a new virtual machine through the scheduling computing service, obtain the latest copy of the orbital calculation algorithm from the image repository and complete the deployment registration through the service registration center;

[0013] S3. External business software submits orbital calculation requirements through the API gateway, and the scheduling computing service distributes the orbital calculation requirements to existing computing resources in a load-balanced manner, ultimately completing the orbital calculation and returning the calculation results.

[0014] Furthermore, after the scheduling computing service receives the computing requirements of the external business software through the API gateway, it decomposes and obtains the historical computing time according to different output types, and estimates the total computing time. When the existing resource load is too large, it drives the virtual machine platform to complete the expansion, and finally allocates computing tasks to each service according to load balancing.

[0015] Furthermore, the scheduling computing service can continuously collect the virtual machine resource occupancy, computing time consumption, and output efficiency fluctuations under various computing requirements, and adjust internal scheduling parameters.

[0016] Furthermore, the scheduling algorithm for scheduling computing services includes virtual machine driving, track computing replica acquisition, service deployment registration and computing demand allocation.

[0017] Furthermore, the orbit calculation includes orbit element prediction, ephemeris calculation, sub-satellite point trajectory drawing and designated site transit prediction.

[0018] A flexible resource allocation system for massive orbit prediction calculations, comprising:

[0019] A virtual machine platform is used to virtualize all available resources and provide external services in the form of virtual machines as the hardware foundation for operation;

[0020] Image repository, used to encapsulate orbital computing algorithms into container images and store them;

[0021] API gateway, used to input orbit calculation requirements submitted by external business software;

[0022] The scheduling computing service module is used to drive the virtual machine platform to allocate a new virtual machine, obtain the latest copy of the orbit calculation algorithm from the image repository, and complete the deployment registration through the service registration center; and is used to distribute the orbit calculation requirements to existing computing resources in a load-balanced manner, and finally complete the orbit calculation and return the calculation results.

[0023] Furthermore, after the scheduling computing service module receives the computing requirements of the external business software through the API gateway, it decomposes and obtains the historical computing time according to different output types, and estimates the total computing time. When the existing resource load is too large, it drives the virtual machine platform to complete the expansion, and finally allocates computing tasks to each service according to load balancing.

[0024] Furthermore, the scheduling computing service module can continuously collect the virtual machine resource occupancy, computing time consumption, and output efficiency fluctuations under various computing requirements, and adjust internal scheduling parameters.

[0025] Furthermore, the scheduling content of the scheduling computing service module includes virtual machine driving, track computing replica acquisition, service deployment registration and computing demand allocation.

[0026] Furthermore, the orbit calculation includes orbit element prediction, ephemeris calculation, sub-satellite point trajectory drawing and designated site transit prediction.

[0027] The beneficial effects of the present invention are:

[0028] 1. The present invention builds a scheduling computing service based on the traditional orbit calculation algorithm. At the hardware level, all available resources are virtualized to provide the hardware foundation for service operation in the form of virtual machines. At the platform level, the orbit calculation algorithm is encapsulated into a container image and stored in the image warehouse. After the scheduling computing service drives the virtual machine platform to allocate a new virtual machine, it can obtain the latest orbit calculation copy from the image warehouse and complete the deployment registration through the service registration center to ensure the parallel expansion of the orbit calculation service. At the call level, the scheduling computing service distributes the computing requirements to the existing computing resources in a load-balancing manner. When the existing resource load is too large, it drives the virtual machine platform to complete the expansion. External business software can submit orbit calculation requirements through the API gateway without feeling. The scheduling computing service automatically completes the calculation of massive satellite orbits through a three-layer structure, providing external business software with orbital root number forecasts, ephemeris calculations, sub-satellite point trajectory drawing, designated site transit forecasts and other calculation results.

[0029] 2. The scheduling computing service of the present invention is based on a three-layer structure of hardware, platform, and call. It has functions such as seamless call, elastic allocation of hardware resources, scalable software resources, computing load balancing, output of multiple calculation results, and scheduling parameter updates, forming a scheduling computing service that can be called on demand by external business software.

[0030] 3. In view of the characteristics of large constellations with many satellites, many ground control stations, large amount of orbit attitude calculations, and tidal calculation requirements, the present invention can output a variety of orbit calculation products according to the requirements of external business software, greatly improving the efficiency of massive satellite orbit calculations. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The figure is a flow chart of a flexible resource allocation method for massive orbital computing according to the present invention.

[0032] Figure 2 This is the call flow chart for external business software when it has track calculation requirements.

[0033] Figure 3 A flowchart for building and iteratively optimizing scheduling computing services. DETAILED DESCRIPTION

[0034] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, the specific embodiments of the present invention are now described. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. That is, the embodiments described are only part of the embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0035] Example 1

[0036] This embodiment provides a flexible resource allocation method for massive orbit prediction calculations, including the following steps:

[0037] S1. Virtualize all available resources and provide the hardware foundation for external services in the form of virtual machines through a virtual machine platform;

[0038] S2. Encapsulate the orbital calculation algorithm into a container image and store it in an image repository. After the computing service scheduler drives the virtual machine platform to allocate a new virtual machine, it obtains the latest copy of the orbital calculation algorithm from the image repository and completes the deployment registration with the service registration center.

[0039] S3. External business software submits orbit calculation requirements through the API gateway, and the scheduling computing service distributes the orbit calculation requirements to existing computing resources in a load-balanced manner, ultimately completing the orbit calculation and returning the calculation results.

[0040] Among them, the virtual machine is a complete computer system simulated by software and has complete hardware system functions, providing a more efficient way to use the underlying hardware. The scheduling computing service can run the track computing image pulled from the image warehouse on the virtual machine. The virtual machine platform divides the hardware resources into multiple virtual machines for management, allowing users to flexibly allocate computing resources.

[0041] The traditional orbit calculation algorithm is encapsulated as a standard TAR file, i.e., an image file, and uploaded to a unified web service for retrieval, i.e., an image warehouse. When the current computing resource load is too large, the scheduling computing service can pull the orbit calculation image from the image warehouse to achieve rapid deployment.

[0042] The service registration center is a middleware that provides service registration for service instances and obtains a list of service instances. When a service is deployed in a multi-instance cluster, the application instance reports service information to the service registration center, which then integrates and generates a service list to achieve unified dynamic management of all services, thus providing feasibility for the horizontal expansion of orbital computing services.

[0043] The API gateway is used to implement communication between two services. It is responsible for service routing requests, load balancing, protocol conversion, authentication filtering and other programs. The service can run independently and freely, provide access addresses to the outside world, and process service requests forwarded by the gateway. Through the API gateway, the orbit computing service, scheduling computing service and other business software can be decoupled. Other business software can submit computing requests to the scheduling computing service through the API gateway without any sense of sensation.

[0044] The scheduling algorithm for scheduling computing services includes virtual machine driving, orbit calculation replica acquisition, service deployment registration, and computing demand allocation, while the orbit calculation content includes orbital element prediction, ephemeris calculation, sub-satellite point trajectory drawing, and designated site transit prediction.

[0045] Preferably, the scheduling computing service receives computing requirements from external business software through an API gateway, breaks down and obtains historical computing time by output type, and estimates the total computing time. When the existing resource load is excessive, it drives the virtual machine platform to complete capacity expansion and finally distributes computing tasks to each service based on load balancing. Even more preferably, the scheduling computing service can continuously collect information on virtual machine resource usage, computing time, and output efficiency fluctuations under various computing requirements, and adjust internal scheduling parameters.

[0046] Preferably, if Figures 1 to 3 As shown, the flexible resource allocation method of this embodiment specifically includes the following steps:

[0047] (1) In the pre-stage, the orbit calculation algorithm is encapsulated as a separate container image;

[0048] (2) Upload the orbit calculation image to the established private image repository;

[0049] (3) Virtualize all available computing resources and build a virtual machine platform for unified management;

[0050] (4) Entering the application phase, the scheduling computing service receives the total computing requirements submitted by external business software through the API gateway;

[0051] (5) Split the satellite into different subtasks and estimate the resources and computing time required for each subtask based on historical mission statistics;

[0052] (6) The scheduling computing service queries the virtual machine platform for the number of idle resources and applies for new resources for deploying orbit computing services;

[0053] (7) The scheduling computing service obtains the orbit computing image from the image warehouse and deploys it to each virtual machine. After successful startup, the started orbit computing service is registered as a callable service through the service registration center;

[0054] (8) The scheduling computing service sends each subtask to the existing orbit computing service in a load-balanced manner;

[0055] (9) The orbit calculation service on each virtual machine performs orbit extrapolation, sub-satellite point, designated station transit and other forecast calculations according to mission requirements;

[0056] (10) After each subtask is calculated, the calculation results are returned to the external business software;

[0057] (11) At the same time, the computing resources and computing time consumed by the subtask are collected to provide a basis for estimating computing resources and time in advance for scheduling computing services;

[0058] (12) Release phase: After completing all computing requirements, the scheduling computing service deregisters and uninstalls each track computing service, and notifies the virtual machine platform to recycle resources.

[0059] Specifically, the calculation formula involved in this embodiment is as follows:

[0060] A satellite is affected by the gravitational field at the center of the Earth and orbits the Earth in an elliptical orbit. Since the mass of the satellite is much smaller than that of the Earth, the distance vector r from the Earth to the satellite can be determined using the following basic equation of the restricted two-body problem, where t is time, G is the gravitational constant, and M is the mass of the Earth.

[0061]

[0062] In fact, in addition to the influence of gravity, satellites are also affected by other tiny forces, which cause the actual orbit to deviate slightly from the offline orbit. This deviation is called orbital perturbation, which is mainly caused by the non-spherical perturbation of the earth, atmospheric resistance, the gravity of third celestial bodies such as the moon and the sun, and solar radiation pressure.

[0063] For satellites, the largest perturbation factor is the perturbation caused by the second-order band harmonic coefficient J2 related to the Earth's oblateness. At this time, the Earth's gravitational potential can be rewritten as the following equation, where μ is the Earth's gravitational constant, R E is the Earth's equatorial radius.

[0064]

[0065] The average orbit element method is applicable to the prediction of six orbit elements in the future when the initial orbit information of the satellite is in the form of six orbit elements. m0 (semi-major axis), e m0 (eccentricity), Ω m0 (right ascension of the ascending node), i m0 (inclination angle), ω m0 (argument of perigee), M m0(Mean Anomaly) Considering only the second-order band harmonic coefficient J2, the average orbital elements at any time t are calculated using the following equation, where W Ω is the average rate of change of the right ascension of the ascending node, W ω is the average rate of change of the argument of perigee, ΔW M is the average rate of change of the mean anomaly.

[0066] a m =a m0

[0067] e m =e m0

[0068] i m =i m0

[0069] Ω m =Ω m0 +W Ω (t-t0)

[0070] ω m =ω m0 +W ω (t-t0)

[0071]

[0072] The Cowell method is applicable to the prediction of the position and velocity vectors in the future when the initial orbit information of the satellite is in the form of position and velocity vectors. i ,y i ,z i ,v xi ,v yi ,v zi ], using the Runge-Kutta method to integrate the following differential equation of motion, the position and velocity vectors at any time can be obtained.

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079] The SGP4 / SDP4 model applies the two-row root form published by the North American Aerospace Defense Command and predicts position and velocity vectors for a period of time into the future. Proposed by Ken Cranford in 1970, SGP4 is suitable for low-Earth orbit predictions. Its gravitational perturbation correction model employs simplified results from a more widely applicable analytical theory and utilizes a density function containing quadratic terms in the atmospheric perturbation correction model. SDP4, on the other hand, is used for deep-space orbit predictions with orbital periods greater than 225 minutes. It incorporates solar and lunar gravitational terms and certain sectoral and field-shaped harmonic terms in the Earth's gravitational potential function.

[0080] The sub-satellite point trajectory is the ground trace of the satellite, which is very important for practical applications of satellites, such as communication, navigation, and earth observation. The calculation method of the geocentric latitude φ and longitude λ of the sub-satellite point is as follows, where u is the latitude argument, α is the latitude angle, and G is the Greenwich right ascension.

[0081] φ=arcsin(sin u sin i)

[0082]

[0083] The forecast of the satellite's transit time relative to a designated site can provide ground constellation control with a window of satellite-to-ground communication information, allowing the ground control center to predict in advance the time and direction of communication between various ground devices and the satellite, thereby scheduling various ground devices to complete the constellation's tasks and transmit status data and control commands.

[0084] The longitude λ of the known station P P and latitude φ P , and the satellite's altitude h and sub-satellite longitude λ at a certain moment S and latitude φ S The following equations can be used to calculate the distance q, elevation angle ε, and azimuth angle μ of the satellite relative to the station. R is the mean radius of the Earth. The satellite's elevation angle relative to the station is calculated at each moment during the required time period and compared with the station's minimum tracking elevation angle. If the elevation angle is greater than the minimum, the satellite is considered a transit.

[0085] cosψ=sinφ P sinφ S +cosφ P cosφ S cos(λ S -λ P )

[0086]

[0087]

[0088] q 2 =R2 +(R+h) 2 -2R(R+h)cosψ

[0089]

[0090]

[0091] Example 2

[0092] This embodiment provides a flexible resource allocation system for massive orbit prediction calculations, including a virtual machine platform, an image repository API gateway, a service registration center, and a scheduling and computing service module. The virtual machine platform is used to virtualize all available resources and provide the hardware foundation for service operation in the form of virtual machines; the image repository is used to encapsulate orbit calculation algorithms into container images and store them; the API gateway is used to input orbit calculation requirements submitted by external business software; the scheduling and computing service module is used to drive the virtual machine platform to allocate new virtual machines, obtain the latest copy of the orbit calculation algorithm from the image repository, and complete the deployment registration through the service registration center; and is used to distribute orbit calculation requirements to existing computing resources in a load-balancing manner, ultimately completing the orbit calculation and returning the calculation results. Specific instructions are as follows:

[0093] A virtual machine is a complete computer system simulated by software and has complete hardware system functions, providing a more efficient way to use the underlying hardware. The scheduling computing service module can run the track computing image pulled from the image warehouse on the virtual machine. The virtual machine platform divides hardware resources into multiple virtual machines for management, allowing users to flexibly allocate computing resources.

[0094] The traditional orbit calculation algorithm is encapsulated as a standard TAR file, i.e., an image file, and uploaded to a unified web service for retrieval, i.e., an image warehouse. When the current computing resource load is too large, the scheduling computing service module can pull the orbit calculation image from the image warehouse to achieve rapid deployment.

[0095] The service registration center is a middleware that provides service registration for service instances and obtains a list of service instances. When a service is deployed in a multi-instance cluster, the application instance reports service information to the service registration center, which then integrates and generates a service list to achieve unified dynamic management of all services, thus providing feasibility for the horizontal expansion of orbital computing services.

[0096] The API gateway is used to implement communication between two services. It is responsible for service routing requests, load balancing, protocol conversion, authentication filtering and other programs. The service can run independently and freely, provide access addresses to the outside world, and process service requests forwarded by the gateway. Through the API gateway, the orbit computing service and scheduling computing service modules can be decoupled from other business software. Other business software can submit computing requests to the scheduling computing service module through the API gateway without any sense of sensation.

[0097] The scheduling content of the scheduling computing service module includes virtual machine driving, orbit calculation copy acquisition, service deployment registration and computing demand allocation, while the orbit calculation content includes orbital root number prediction, ephemeris calculation, sub-satellite point trajectory drawing and designated site transit prediction. Preferably, after the scheduling computing service module receives the computing requirements of the external business software through the API gateway, it decomposes and obtains the historical computing time according to different output types, and estimates the total computing time. When the existing resource load is too large, it drives the virtual machine platform to complete the expansion, and finally allocates computing tasks to each service according to load balancing. More preferably, the scheduling computing service module can continuously collect the virtual machine resource occupancy, computing time, and output efficiency fluctuations under various computing requirements, and adjust the internal scheduling parameters.

[0098] It should be noted that, for the sake of simplicity, the aforementioned method embodiments are described as a series of action combinations. However, those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

Claims

1. A flexible resource allocation method for massive orbit prediction calculations, characterized in that: The following steps are involved: S1. Virtualize all available resources and provide external services in the form of virtual machines through a virtual machine platform to provide the hardware foundation for operation. S2. Encapsulate the orbital calculation algorithm into a container image and store it in an image repository. After the scheduling computing service drives the virtual machine platform to allocate a new virtual machine, it obtains the latest copy of the orbital calculation algorithm from the image repository and completes the deployment registration with the service registration center. S3. External business software submits orbit calculation requests through the API gateway. The scheduling computing service distributes the orbit calculation requests to existing computing resources in a load-balanced manner, ultimately completing the orbit calculation and returning the calculation results. The specific implementation steps of the flexible resource allocation method include: In the pre-production phase, the orbit calculation algorithm is encapsulated into a separate container image; Upload the orbit calculation image to the established private image repository; Virtualize all available computing resources and build a virtual machine platform for unified management; Entering the application phase, the scheduling computing service receives the total computing requirements submitted by external business software through the API gateway; Split each satellite into different subtasks, and estimate the resources and computing time required for each subtask based on historical mission statistics; The scheduling computing service queries the virtual machine platform for the number of idle resources and applies for new resources for deploying orbit computing services; The scheduling computing service obtains the orbit computing image from the image repository and deploys it to each virtual machine. After successful startup, the started orbit computing service is registered as a callable service through the service registration center. The scheduling computing service sends each subtask to the existing track computing service in a load-balancing manner; The orbit calculation service on each virtual machine performs prediction calculations according to mission requirements, including orbit extrapolation, sub-satellite point and designated station transit calculations; After each subtask is calculated, the calculation results are returned to the external business software; Collect the computing resources and computing time consumed by the subtask calculation to provide a basis for pre-estimating computing resources and time for scheduling computing services; In the release phase, after completing all computing needs, the scheduling computing service deregisters and uninstalls each track computing service, and notifies the virtual machine platform to recycle resources.

2. The flexible resource allocation method for massive orbit prediction calculation according to claim 1, characterized in that: After the scheduling computing service receives the computing requirements of the external business software through the API gateway, it decomposes and obtains the historical computing time according to different output types, and estimates the total computing time. When the existing resource load is too large, it drives the virtual machine platform to complete capacity expansion, and finally allocates computing tasks to each service according to load balancing.

3. The flexible resource allocation method for massive orbit prediction calculation according to claim 2, characterized in that: The scheduling computing service can continuously collect the virtual machine resource usage, computing time, and output efficiency fluctuations under various computing requirements, and adjust internal scheduling parameters.

4. The flexible resource allocation method for massive orbit prediction calculation according to claim 1, characterized in that: The scheduling algorithm for scheduling computing services includes virtual machine driving, track computing replica acquisition, service deployment registration and computing demand allocation.

5. The flexible resource allocation method for massive orbit prediction calculation according to claim 1, characterized in that: The orbit calculation includes orbit element prediction, ephemeris calculation, sub-satellite point trajectory drawing and designated station transit prediction.

6. A flexible resource allocation system for massive orbit prediction calculations, characterized by: include: A virtual machine platform is used to virtualize all available resources and provide external services in the form of virtual machines as the hardware foundation for operation; Image repository, used to encapsulate orbital computing algorithms into container images and store them; API gateway, used to input orbit calculation requirements submitted by external business software; The scheduling computing service module is used to drive the virtual machine platform to allocate a new virtual machine, obtain the latest orbit calculation algorithm copy from the image repository, and complete the deployment registration through the service registration center; and for distributing the orbit calculation requirements to existing computing resources in a load-balanced manner, ultimately completing the orbit calculation and returning the calculation results; The elastic resource allocation system can implement the following steps: In the pre-production phase, the orbit calculation algorithm is encapsulated into a separate container image; Upload the orbit calculation image to the established private image repository; Virtualize all available computing resources and build a virtual machine platform for unified management; Entering the application phase, the scheduling computing service receives the total computing requirements submitted by external business software through the API gateway; Split each satellite into different subtasks, and estimate the resources and computing time required for each subtask based on historical mission statistics; The scheduling computing service queries the virtual machine platform for the number of idle resources and applies for new resources for deploying orbit computing services; The scheduling computing service obtains the orbit computing image from the image repository and deploys it to each virtual machine. After successful startup, the started orbit computing service is registered as a callable service through the service registration center. The scheduling computing service sends each subtask to the existing track computing service in a load-balancing manner; The orbit calculation service on each virtual machine performs prediction calculations according to mission requirements, including orbit extrapolation, sub-satellite point and designated station transit calculations; After each subtask is calculated, the calculation results are returned to the external business software; Collect the computing resources and computing time consumed by the subtask calculation to provide a basis for pre-estimating computing resources and time for scheduling computing services; In the release phase, after completing all computing needs, the scheduling computing service deregisters and uninstalls each track computing service, and notifies the virtual machine platform to recycle resources.

7. The flexible resource allocation system for massive orbit prediction calculations according to claim 6, characterized in that: After the scheduling computing service module receives the computing requirements of the external business software through the API gateway, it decomposes and obtains the historical computing time according to different output types, and estimates the total computing time. When the existing resource load is too large, it drives the virtual machine platform to complete the expansion, and finally allocates computing tasks to each service according to load balancing.

8. The flexible resource allocation system for massive orbit prediction calculations according to claim 7, characterized in that: The scheduling computing service module can continuously collect the virtual machine resource occupancy, computing time consumption, and output efficiency fluctuations under various computing requirements, and adjust internal scheduling parameters.

9. The flexible resource allocation system for massive orbit prediction calculations according to claim 6, characterized in that: The scheduling content of the scheduling computing service module includes virtual machine driving, track computing replica acquisition, service deployment registration and computing demand allocation.

10. The flexible resource allocation system for massive orbit prediction calculations according to claim 6, characterized in that: The orbit calculation includes orbit element prediction, ephemeris calculation, sub-satellite point trajectory drawing and designated station transit prediction.

Images