An application of fuzzy modeling to rowing motion analysis
Kanta Tachibana, Takeshi Furuhashi, M. Shimoda, Yasuo Kawakami, T. Fukunaga · 2003
Fuzzy modeling has distinct features, which are applicable to nonlinear systems and has an ability to extract knowledge. Fuzzy neural networks (FNNs) enable automatic acquisition of knowledge. The authors have proposed an uneven division of input space for an FNN which reduces the number of fuzzy rules without sacrificing the precision of the model. In many sports, nonlinear factors affect the performance. In rowing competitions, the performance criterion is the boat speed. In the paper, fuzzy modeling is applied to reveal the relationships between the supplied power and the boat speed. The forces and the angles of on-water rowing are measured. The subjects are candidate Japanese national team members. The total propulsive work, consistency and uniformity of the propulsive power were calculated from the force and the angle data. The relationships between these factors and the performance were identified with fuzzy modeling. Compared to linear regression, a more precise and more comprehensive model was obtained.