1 / 33

ASEN 5070: Statistical Orbit Determination I Fall 2013 Professor Brandon A. Jones

ASEN 5070: Statistical Orbit Determination I Fall 2013 Professor Brandon A. Jones Professor George H. Born Lecture 32: SNC Example and Dynamic Model Compensation. Announcements. Homework 10 due next week Lecture quiz due by 5pm on Wednesday Posted over the weekend.

toyah
Download Presentation

ASEN 5070: Statistical Orbit Determination I Fall 2013 Professor Brandon A. Jones

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. ASEN 5070: Statistical Orbit Determination I Fall 2013 Professor Brandon A. Jones Professor George H. Born Lecture 32: SNC Example and Dynamic Model Compensation

  2. Announcements • Homework 10 due next week • Lecture quiz due by 5pm on Wednesday • Posted over the weekend

  3. Application of SNC to Ballistic Trajectory

  4. Example – Problem Statement • Ballistic trajectory with unknown start/stop • Red band indicates time with available observations Obs. Stations Start of filter

  5. Example – Problem Statement • Object in ballistic trajectory under the influence of drag and gravity • Nonlinear observation model • Two observations stations

  6. Example – Problem Statement

  7. Changes Since Last Time • Now use an EKF • We will vary the filter model to study the benefits of SNC • Look at two cases: • Run each with and without a process noise model • Error in gravity (g = 9.8 m/s vs. 9.9 m/s) • Error in drag (b = 1e-4 vs. 1.1e-4)

  8. Filter Residuals over Time, Gravity Error Station 1 Station 2 • Blue – Range • Green – Range-Rate Edited Points (ignore them)

  9. State Error and Uncertainty, Gravity Error

  10. SNC Design for Gravity Error • Added SNC to the filter: • Why is the term for x-acceleration smaller?

  11. Filter Residuals over Time, Gravity-SNC Station 1 Station 2 • Blue – Range • Green – Range-Rate

  12. State Error and Uncertainty, Gravity-SNC • 178.3 vs. 0.8 meters RMS

  13. Filter Residuals over Time, Drag Error Station 1 Station 2 • Blue – Range • Green – Range-Rate Edited Points (ignore them)

  14. State Error and Uncertainty, Drag Error

  15. SNC Design for Drag Error • Added SNC to the filter:

  16. Filter Residuals over Time, Drag-SNC Station 1 Station 2 • Blue – Range • Green – Range-Rate

  17. State Error and Uncertainty, Gravity-SNC • 27.6 vs. 1.26 meters RMS

  18. Relative Performance of Cases • Mitigation of the gravity acceleration error was easier to mitigate than the drag error. Why?

  19. Dynamic Model Compensation

  20. What is the Markov property? • The Markov property describes a random (stochastic) process where knowledge of the future is only dependent on the present:

  21. Do these processes have the Markov property? • An object under deterministic motion? • A satellite in a chaotic orbit? • An object under stochastic motion?

  22. Gauss-Markov Processes • Introduction of the random, uncorrelated (in time), Gaussian process noise u(t) makes η a Gauss-Markov process • We will use the GMP to develop another form of process noise

  23. GMP Analytic Solution • Stochastic integral cannot be solved analytically, but has a statistical description: Stochastic Deterministic

  24. GMP Analytic Solution • Stochastic integral cannot be solved analytically, but has a statistical description: Stochastic Deterministic

  25. Correlation of GMP in Time • It may be shown that: • In other words: • The process is exponentially correlated in time • Rate of the correlation fade determined by β • For large β, the faster the decay

  26. Equivalent Process • Instead, let’s use the equivalent process • Lk has the same statistical description as the stochastic integral • Hence, it is an equivalent process

  27. Alternate Solution for η • Process behavior varies with the equation parameters Add dependence on time to emphasize different realizations for different times

  28. GMP Behavior • What happens if β  0?

  29. GMP Behavior • What happens if σ 0?

  30. Illustration of GMP Evolution

  31. Adding the DMC to your Filter • Augment the state vector to include the accelerations • Dubbed Dynamic Model Compensation (DMC) • The random portion determines the process noise matrix Q(t) (see Appendix F, p. 507-508)

  32. DMC More Robust than SNC

  33. GMP1 vs. GMP2 Image: Leonard, Nievinski, and Born, 2013

More Related