| |
Amphibious-Vehicle Control Method Based on Neural Network Estimation of Underwater Running-Resistance
| [View]
[Download]
|
| Author(s) |
K. Yoshida; A. Nafa |
| Abstract |
When running underwater a linear synchronous motor train will face a large running-resistance
difficult to evaluate which is mainly due to much stronger hydrodynamic effect than the mechanical
friction between guide-rollers and guideway. To evaluate this running-resistance a neural network is
designed and trained off-line using experimental data. The trained neural network can evaluate the
running-resistance for any demand position and speed patterns within the range of speed limited by the
training data. Combining both the robustness of a servo controller and the neural network fast
calculation capability, the trained neural network is used as a compensator in a feed-forward control
loop to increase the efficiency of the decoupled controller used so far to control marine express 02 for
levitation, propulsion and guidance. |
| Download |
| Filename: | EPE-PEMC2002 - T9-042 - Yoshida.pdf |
| Filesize: | 393.8 KB |
|
| Type |
Members Only |
| Date |
Last modified 2004-05-24 by System |
| |
 |