A neuro-fuzzy model of the residuary resistance of sailing yachts

The estimation of residuary resistance in sailing yachts is very important for the initial stage of their design, because of its influence on the yacht performance and the assessment of the required propulsive power. This is why substantial efforts have been made during the past to produce an accurate model for its prediction. However, the models of residuary resistance presented in the literature are exclusively based on regression analysis and represent polynomial formulas, which suffer from great involvement with too many parameters and their coefficients depend heavily on the Froude number. The paper presents an alternative approach to modeling the residuary resistance of sailing yachts based on fuzzy logic and neural network techniques. The neuro-fuzzy model developed for prediction of the residuary resistance in sailing yachts is a fuzzy inference system with a learning mechanism based on statistical learning theory and extended relevance vector machines. To evaluate its performance, the model is compared to other unconventional models based on fuzzy logic and neural network techniques. The simulation results show that the built neuro-fuzzy model solves very elegantly the problem of modeling the residuary resistance of sailing yachts and shows superior performance compared to the conventional regression models as well as the other fuzzy and neural models.