文摘
This thesis presents a Neural Network-based performance analysis and optimization of sailing configurations of an America's Cup class yacht. In the approach, the analysis is based on experimental data from the sailing records provided by sensors to train the neural networks. The focus of the paper is on the performance analysis, with the objective of maximizing boat speed (objective function) by varying the sailing setups (design variables). The neural network based on the experimental data is coupled with a genetic algorithm to determine the maximum boat speed and corresponding yacht settings at various wind speeds. The solutions for the design variables, however, show some differences suggesting that further analysis is warranted to analyze the network accuracy as it transitions from the dense regions with experimental data to the empty regions which have been artificially populated. The thesis demonstrates that multiple yacht configurations can be analyzed instantaneously on a regular PC.