| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 438.12 KB | Adobe PDF |
Advisor(s)
Abstract(s)
Online parameter estimation for time-varying systems is a fundamental part of adaptive control, real-time system monitoring and prediction. A well-known framework for dealing with such a task is the Kalman filtering. Meanwhile Kalman filtering may be cumbersome for some time-critical systems and inappropriate for systems whose stochastic characteristics are not Gaussian. To overcome these shortcomings, a parameter estimation algorithm devised from Sutton’s dynamic learning rate techniques and based on a learning window and forgetting factor criterion has been used. In doing so, the proposed algorithm avoids the need for heuristic choices of the initial conditions and noise covariance matrices required by the Kalman filtering. The performance of the proposed method is demonstrated successfully on a lateral-directional flight dynamics parameter estimation process for an unmanned aerial vehicle through computational simulation.
Description
Keywords
Nonlinear parameter estimation Time-varying systems Dynamic learning rate Flight dynamics
Citation
Coelho, M.F., Bousson, K., Ahmed, K., Nonlinear Time-varying Parameter Estimation from Noisy Measurements, International Conference on Engineering ICEUBI-2017, Paper ID: 16.01, pp. 621-629, Covilhã, Portugal, December 5-7, 2017.
Publisher
Universidade da Beira interior