Parameters tuning of multivariable controllers based on memetic algorithm: fundamentals and application

Abstract

This paper presents the design and tuning of multivariable controllers through a memetic algorithm based on Lamarckian theory. Proposed memetic algorithm consists of a fast evolutionary programming combined with a simulated annealing algorithm. Multivariable controllers with memetic tuning are Implemented under the following conceptions: (i) multiloop PD (proportional-derivative) and (ii) multiloop PID (proportional-integral-derivative) configurations. The simulations are assessed on a robot manipulator of two degree-of-freedom that constitute a MIMO nonlinear dynamic system. Simulation results of multivariable control design based on memetic algorithm are presented and discussed. Results show that the application of these control strategies effectively Improve the trajectory tracking precision of position and velocity variables and that the proposed memetic algorithm is a powerful tool for parameter optimization in multivariable control applications.