Projeto de um controlador preditivo baseado em modelo aplicado ao pêndulo invertido

Abstract

In the area of process control, the most widespread controller in the industry is the so-called PID controller. Its simplicity and good results made it the most implemented controller since its conception. However, advanced control strategies are necessary in more complex cases or cases that must take other criteria into account, in addition to seeking reference and rejecting disturbances. A family of controllers that use optimization criteria is the MPC controller family (Model-Based Predictive Controllers) whose characteristics and advantages have been standing out within the process control area. This work explores the performance differences between a classic PID controller and an MPC controller, the latter considered to be an advanced control strategy, applied to the system called inverted pendulum, a system commonly used to test controller performance due to its unstable characteristic. The results were obtained in a simulated environment and show that both controllers have similar performance. However, when the system is subject to disturbances and limitations in the control action, it was noticed that the MPC controller copes with these situations with better results, including the fact that the PID controller needs two separate controllers to control the pendulum angle and the position 𝑥 of the car, while the MPC is a controller that can easily handle multi-objectives, including them directly in the cost function to be minimized by the optimizer.