Abstract
A model based experimental method of tracking the 'hidden' parameter drift in nonlinear dynamical systems is described. Local linear models constructed using delay coordinate embedding are used to analyze data collected from systems with parameters that are drifting on a slow time scale according to some hidden rate law. Data is sampled over a fast time scale. The method is applied to a numerical study of a forced nonlinear duffing oscillator. A forcing amplitude and a first natural frequency of a corresponding linear system served as the drifting parameters. The mean-square model single step prediction error, tracking parameter, is shown to follow closely the assigned variations in drifting parameters. Ramped and sinusoidal parameter variations were used. It is concluded that, at least in the cases studied, the tracking model based on delay coordinate embedding provides one to one correspondence between the drifting and tracking parameters.
Original language | English (US) |
---|---|
Pages (from-to) | 525-530 |
Number of pages | 6 |
Journal | Proceedings - National Conference on Noise Control Engineering |
Volume | 1 |
State | Published - 1997 |
Event | Proceedings of the 1997 National Conference on Noise Control Engineering , NOISE-CON. Part 2 (of 2) - University Park, PA, USA Duration: Jun 15 1997 → Jun 17 1997 |
All Science Journal Classification (ASJC) codes
- Acoustics and Ultrasonics