Control predictivo en espacio de estados. Aplicación al control del modelo dinámico simulado de un péndulo de furuta

Abstract

This project aims to control a pendulum of Furuta; which presents a complex nature such as its fast and nonlinear dynamics, instability, coupled of variables, sensitivity to perturbations and, in addition, being an underactuated system. The methodology outlined highlights four main phases; starting with a basic CAD design of the system structure in SOLIDWORKS, so that the value of some variables needed for dynamic modeling could be obtained; which was made by the Lagrangian formalism. Then, the theoretical model was simulated in Simulink and the CAD model imported in the SimMechanics-MATLAB environment configured with its respective Denavit-Hartenberg, so that the mathematical model could be validated. Finally, a linearization was done by Taylor series for the design of the monovariable predictive control in state space; which was then extended to the multivariate case; followed by this, a fuzzy controller was designed for the pendulum lifting problem. The main contribution is the multivariate control of the Furuta pendulum in the SimMechanics environment so that it is not necessary to create a 3D platform for visualization of the CAD model.