PERFORMANCE MODELLING AND ANALYSIS OF OLYMPIC CLASS SAILING BOATS AND CREWS USING NEURAL NETWORKS

摘要

This paper addresses the problem of modelling performance accurately in small high performance boats. The approach taken, which uses neural network modelling to produce a dynamic model of the performance of the boat and crew, is compared to the more conventional quasi static approach. The neural network based model is able to function in not only a single point mode but can also be used for time series comparison in tuning runs. The paper describes the development of the initial neural network model to include further parameters, such as standing rigging loads, allowing further understanding of boat performance. The method, which was used in the training of the Olympic Sailing Team in the build up to the 2008 Olympic Games, allows a novel approach to performance analysis that is more accurate than methods previously used. The paper also explores the combination of neural networks and VPPs to produce a global measure of performance.