Implementación de un algoritmo de aprendizaje de máquina para la sintonización de un controlador aplicado a un péndulo invertido.

Abstract

This degree project presents the design and implementation of a tuning of an angular position controller for an inverted rotary pendulum, using a Bio-inspired clonal selection algorithm for adjusting the parameters of a PID controller. This details the process of creating the PID controller, the simulated model in MATLAB of the inverted rotary pendulum, and the programming clonal selection algorithm for tuning the gains required by the controller for its control action. Likewise, comparisons of the performance of controllers in their transitive process are made, where they measure how much improves the overshoot,time setting and steady-state error of the base controllers and the controller tuned to the clonal algorithm. Finally, the tuning performance of the clonal selection algorithm on an FSF (Full State Feedback) controller applied to a QUANSER Rotary Inverted Pendulum physical plant is verified.