Differential evolution based identification of the LuGre friction model in the cart motion of an inverted pendulum system

Abstract

This paper discusses the off-line identification of friction encountered in the cart motion of an inverted pendulum system. A LuGre model is chosen as the parametric model to represent this phenomenon. We propose an identification technique based on a differential evolution algorithm. The proposed identification technique distinguishes itself from the previous works by optimizing both of the static and the dynamic parameters simultaneously. This fact is possible through adopting closed loop identification where the velocity is controlled and using a sequence of static and dynamic signals as the reference signal to enhance the effects of the different friction phenomena. The simulation results demonstrate that the proposed differential evolution based identification method is effective in finding the parameters precisely despite the nonlinearity of the LuGre friction model and the coupling between the parameters.