Análisis comparativo de modelos difusos para la estimación de la dinámica del consumo de oxígeno (VO2) en ciclismo deportivo

Abstract

The tuning of normalized Fuzzy Systems Type I and II are described with the aim of estimating the dynamics of oxygen consumption (VO2) in sport cycling from 3 variables measured in a laboratory: Heart Rate, Respiratory Rate, and Power. The process started by preparing the database from the data collected in 11 Cardiopulmonary Exercise Tests (CPET) applied to male cyclists between 18 and 22 years old from Liga de Ciclismo de Bogotá. The test were carried out in the Unidad de Ciencias Aplicadas al Deporte (UCAD) in Bogotá Colombia following the international protocols. The Type I and II fuzzy estimators were evaluated from 5 performance indexes: RMSE, R2, NSE, MAPE, and VAF, which allowed a comparative analysis by contrasting them with a third system called 'averager' which obtained the output from the averaged-addition of the normalized inputs. The results obtained show that both tuned systems present improvements in all the performance indexes with respect to the 'averager', being the Type I Fuzzy System the best since it presented higher values in 4 of 5 indexes.