Identificação usando método não linear de um sistema de posicionamento

Abstract

Friction is one of the main challenges in robotics, specifically in robot manipulators, affecting the lifetime and performance of the system. The presence of complex and non-linear aspects in this phenomenon makes its modelling challenging to conceive. We present an original experimental bench composed of an motor and a link joined by a rigid joint. We seek to find the friction model that best represents the developed servosystem through experimental measurements. Thus, the Gray Box type identification process is used to optimize the different parameters of the friction models. The parameters of the different friction models are identified using the Shooting Method through the data. Then the different models are compared. We observed that the models that consider the most significant number of friction phenomena perform better.