Robust Granular Feedback Linearization

Abstract

This paper addresses a novel granular feedback linearization approach to robustly and efficiently control imprecise nonlinear systems. This approach employs an evolving participatory learning algorithm to estimate the nonlinearities and to cancel their effects in the control loop. Simulation experiments with a surge tank are used to evaluate and to compare the performance of the robust granular feedback linearization with an adaptive controller based on a bacterial foraging technique, and an alternative granular evolving fuzzy robust feedback linearization mechanism. The results suggest that the robust granular feedback linearization performs better than the other approaches from the point of view the performance indexes commonly adopted in the process control practice like integral of the absolute error, the integral of the time-weighted absolute error, and the integral of the variability of the error.