Calibration of a power-velocity-model for road cycling using real power and height data

Abstract

The relationship between the pedaling power and the velocity with road cycling can be described in the form of a dynamical system, that accounts for the inertia and several kinds of mechanical friction. Six out of twelve involved constant parameters are known or can be measured, whereas the determination of the remaining six parameters is prohibitive and depends on the specific cyclist, the bicycle, and the road surface.The sought six parameters occur in the model in only four terms with distinct dependency on power and velocity. The parameters are combined to yield one coefficient per term. Using real power and velocity measurements on a calibration track and a least squares fitting technique, we estimate these four coefficients.Although the interpretation of the distinct physical phenomena cannot be preserved completely, an experimental evaluation shows that our calibration improves the model velocity estimation on both the calibration track and on other tracks with the same road surface. Moreover, we compare the significance of the parameters and show that the selection of a flat calibration track and precise slope measurements are essential for a successful calibration.