Superiority of q-Chlodowsky operators versus fuzzy systems and neural networks: Application to adaptive impedance control of electrical manipulators

Abstract

This paper introduces a novel application of q-Chlodowsky operators in the approximation of unknown nonlinear functions including uncertainties, un-modeled dynamics, and external disturbances. In fact, q-Chlodowsky operators play the role of basis functions with unknown coefficients. Furthermore, an effective model-free observer is designed for estimation of the task-space velocity signals of the end-effector. The results illustrate that the performances of both radial basis functions neural networks (RBFNN) and the q-Chlodowsky-based approach are nearly the same due to the universal approximation property of both estimators, while the adaptive fuzzy controller needs optimal tuning which is time consuming. Therefore, compared with fuzzy systems and neural networks, the proposed scheme is superior in terms of simplicity and is less computational due to the state-free basis functions in the regressor vector. Simulation results on a 2-DOF electrical manipulator effectively verify the efficiency of the proposed strategy.