Neuro-Fuzzy Modeling and Fuzzy Control of a Fast Ferry - Fuzzy Modeling and Control: Theory and Applications - page 184

Fig. 10.2 Pitch excitationmodel obtained by fuzzy inference using real and simulated data, 30 knots
The fuzzy model fits very well at SSN of 4, 5 and 6. For lower and higher sea state
the experimental data are more scattered and that makes the fitness more difficult.
For SSN lower than 4 there is no problem of stabilization, and for SSN larger than
6, the ferry could hardly travel. The results are then very encouraging in the range
of interest (Santos et al. 2006 ).
We have also applied fuzzy inference systems to obtain a model of the ship
using experimental data (Fig. 10.2 ). These data were provided by CEHIPAR, Madrid
(CEHIPAR 1998 ). There were only 9 experiments with real (irregular) waves that
we used for the generalization. The training was done using 45 experiments with
regular waves (obtained by the previous model). When it was tested with irregular
waves the results were not completely satisfactory mainly due to the phase.
10.2.2 Adaptive Neuro-Fuzzy Predictive Model
A control oriented model using neuro-fuzzy techniques was also developed (Santos
et al. 2005b ). The model is general, for any sea state and ship velocity. The data come
from real experiments, not simulated ones, provided by the CEHIPAR towing tank
(CEHIPAR 2013 ). They have been obtained by carrying out some experiments with
a 1/25 scale replica of the ferry with series of real waves. It is based on the following