Method and apparatus for rapid discovery of satellite behavior

Patent metadata

Field	Value
Publication number	US-11574223-B2
Application number	US-201916595107-A
Country	US
Kind code	B2
Filing date	Oct 7, 2019
Priority date	Oct 7, 2019
Publication date	Feb 7, 2023
Grant date	Feb 7, 2023

Abstract

A method for rapid discovery of satellite behavior, applied to a pursuit-evasion system including at least one satellite and a plurality of space sensing assets. The method includes performing transfer learning and zero-shot learning to obtain a semantic layer using space data information. The space data information includes simulated space data based on a physical model. The method further includes obtaining measured space-activity data of the satellite from the space sensing assets; performing manifold learning on the measured space-activity data to obtain measured state-related parameters of the satellite; modeling the state uncertainty and the uncertainty propagation of the satellite based on the measured state-related parameters; and performing game reasoning based on a Markov game model to predict satellite behavior and management of the plurality of space sensing assets according to the semantic layer and the modeled state uncertainty and uncertainty propagation.

First claim

What is claimed is:

1. A method for rapid discovery of satellite behavior, applied to a pursuit-evasion (PE) system including at least one satellite and a plurality of space sensing assets, comprising:
performing transfer learning and zero-shot learning to obtain a semantic layer using space data information, wherein the space data information includes simulated space data based on a physical model and data collected by multiple sensors;
obtaining measured space-activity data of the at least one satellite from the plurality of space sensing assets;
performing manifold learning on the measured space-activity data to obtain measured state-related parameters of the at least one satellite;
modeling state uncertainty and uncertainty propagation of the at least one satellite based on the measured state-related parameters; and
performing game reasoning based on a Markov game model to predict satellite behavior of the at least one satellite and management of the plurality of space sensing assets according to the semantic layer and the modeled state uncertainty and uncertainty propagation, the satellite behavior including one or more of: a position of the at least one satellite relative to the observing station, a velocity, an orbital energy, or an angular momentum of the at least one satellite, wherein performing game reasoning based on the Markov game model to predict the satellite behavior of the at least one satellite and the management of the plurality of space sensing assets according to the semantic layer and the modeled state uncertainty and uncertainty propagation includes using a classifier model to detect satellite behavior patterns and identify abnormal satellite behavior patterns.

2. The method according to claim 1 , wherein:
the space data information used for transfer learning and zero-shot learning further includes the predicted satellite behavior of the at least one satellite.

3. The method according to claim 1 , wherein:
the semantic layer depicts relationships between features, attributes, and classes extracted from the space data information, and includes space pattern dictionary and/or semantic rules.

4. The method according to claim 1 , further including:
in response to the predicted management of the plurality of space sensing assets different from current management of the plurality of space sensing assets, updating the management of the plurality of space sensing assets.

5. The method according to claim 1 , wherein:
the state uncertainty is represented as a product of a Gaussian distribution of measurement noise and a Von-Mises distribution of initial condition uncertainty, wherein both distributions are defined on a cylindrical manifold R 5 ×S, where is a space of reals and S denotes a circular space.

6. The method according to claim 1 , wherein:
the game reasoning performed based on the Markov game model is a Markov decision process, wherein:
the at least one satellite and the plurality of space sensing assets in the PE system are players;
a payoff function is defined for each action based on a predicted state of the PE system prior to the action and a measured state of the PE system after the action;
an action-value function is defined as a sum of weighted payoff functions of actions taken in a preset period, wherein for every two consecutive actions, a discount factor ranging between 0 and 1 is used to weight a former action between the two consecutive actions; and
performing game reasoning based on the Markov game model includes examining the action-value function for the actions taken in the preset period.

7. The method according to claim 1 , wherein:
performing transfer learning and zero-shot learning to obtain the semantic layer includes modeling satellite behavior patterns of the at least one satellite, including extracting features from the space data information, and generating the classifier model through training using the extracted features.

8. The method according to claim 7 , when an abnormal satellite behavior pattern is identified, further including:
triggering a warning message.

9. An apparatus for rapid discovery of satellite behavior in a PE system that further includes at least one satellite and a plurality of space sensing assets, comprising:
a processor, and a memory, wherein:
the memory is configured to store computer-executable instructions; and
when the processor executes the computer-executable instructions stored in the memory, a method for rapid discovery of satellite behavior is implemented, wherein the method includes:
performing transfer learning and zero-shot learning to obtain a semantic layer using space data information, wherein the space data information includes simulated space data based on a physical model and data collected by multiple sensors;
receiving measured space-activity data of the at least one satellite from the plurality of space sensing assets, and performing manifold learning on the measured space-activity data to obtain measured state-related parameters of the at least one satellite;
modeling state uncertainty and uncertainty propagation of the at least one satellite based on the measured state-related parameters; and
performing game reasoning based on a Markov game model to predict satellite behavior of the at least one satellite and management of the plurality of space sensing assets according to the semantic layer and the modeled state uncertainty and uncertainty propagation, the satellite behavior including one or more of: a position of the at least one satellite relative to the observing station, a velocity, an orbital energy, or an angular momentum of the at least one satellite, wherein performing game reasoning based on the Markov game model to predict the satellite behavior of the at least one satellite and the management of the plurality of space sensing assets according to the semantic layer and the modeled state uncertainty and uncertainty propagation includes using a classifier model to detect satellite behavior patterns and identify abnormal satellite behavior patterns.

10. The apparatus according to claim 9 , wherein:
the space data information used for transfer learning and zero-shot learning further includes the predicted satellite behavior of the at least one satellite.

11. The apparatus according to claim 9 , wherein:
the semantic layer depicts relationships between features, attributes, and classes extracted from the space data information, and includes space pattern dictionary and/or semantic rules.

12. The apparatus according to claim 9 , wherein:
in response to the predicted management of the plurality of space sensing assets different from current management of the plurality of space sensing assets, the method implemented by the apparatus further includes updating the management of the plurality of space sensing assets.

13. The apparatus according to claim 9 , wherein:
the state uncertainty is represented as a product of a Gaussian distribution of measurement noise and a Von-Mises distribution of initial condition uncertainty, wherein both distributions are defined on a cylindrical manifold R 5 ×S, where is a space of reals and S denotes a circular space.

14. The apparatus according to claim 9 , wherein:
the game reasoning performed by the apparatus based on the Markov game model is a Markov decision process, wherein:
the at least one satellite and the plurality of space sensing assets in the PE system are players;
a payoff function is defined for each action based on a predicted state of the PE system prior to the action and a measured state of the PE system after the action;
an action-value function

Assignees

• Intelligent Fusion Tech Inc

Inventors

• Shen Dan
• Sheaff Carolyn
• Lu Jingyang
• Chen Genshe
• Blasch Erik
• Pham Khanh

Classifications

• G06N3/096

  Transfer learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N7/01 Primary

  Probabilistic graphical models, e.g. probabilistic networks · CPC title