Tuning of a Modified Model Reference Adaptive Controller using a PSO Algorithm

This article presents the tuning of a Modified Model Reference Adaptive Controller (MMRAC) through the Particle Swarm Optimization algorithm (PSO). The MMRAC is applied to the velocity control of a DC motor and its main idea is to add damping to eliminate the oscillations in its transitory response. In order to tune the parameters of the MMRAC, the PSO algorithm takes into account the restrictions encountered in the Lyapunov stability analysis to establish the set of feasible solutions. A key feature of the proposed tuning method is that the fitness function employed in the PSO algorithm uses an upper bound of the Lyapunov function time derivative to penalize solutions that lead to instability. Only upper and lower bounds on the parameters of a model of the DC motor are known. In this way, the dynamic simulations executed during the optimization performed by the PSO employ the average of these bounds. Finally, real-time experiments are performed using a DC motor driving different inertia disks to verify that the parameters of the MMRAC tuned through the PSO are useful despite the changes in the DC motor parameters.