Power estimation in wheelchair basketball

The right training and equipment become increasingly more important in professional wheelchair sports such as wheelchair basketball. Individual performance characteristics need to be measured to be able to optimise for example wheelchair settings. Recently, a method using shimmer sensors has been developed and validated to measure speed, acceleration and rotation speed of a wheelchair. This thesis explains how these data can be used in wheelchair basketball to estimate a different performance measure: the external power. Models for the external power are based on the power balance equation, and include the rolling resistance, which is the main resistance, and possibly air resistance or bending forward of the athlete. Resistance parameters can be estimated based on known coasting periods, or based on estimated coasting points or periods. A simple model is proposed to identify coasting points and more sophisticated Hidden Markov Models (HMMs) are proposed to estimate resistance parameters and coasting periods simultaneously in a Bayesian framework. Next to three available data sets, a pilot study is carried out in which data are acquired of both shimmer sensors and a smartwheel, which measures external power. The models are analysed and applied to multiple data sets. The pilot data are used to compare estimated with measured external power. Using the simple model to identify coasting points results in an overestimation of the rolling resistance, and in an estimation of the external power that is highly sensitive for the subjective choice of parameters in the simple model. The proposed HMMs show promising results for the use of HMMs in automatically tuning resistance parameters. Best results in power estimation in the pilot data are obtained when the rolling resistance is based on a known coasting period in bent forward position.

To reference this document use:

http://resolver.tudelft.nl/uuid:e63b6564-214d-4b1c-b721-8195ce3ddecf

Part of collection

Student theses

Document type

master thesis

Rights

Files

PDF

Masterthesis Mirrelijn van Nee.pdf

10.19 MB

Close viewer