Projeto orientado a dados do controlador PID adaptativo para sistemas não lineares

Abstract

Data-Driven Control (DDC) methods have gained prominence in recent decades because they do not require the parametric process model to tune industrial controllers. In this work, the Virtual Reference Feedback Tuning (VRFT) technique is evaluated in the design of Lookup Table (LUT) and polynomial gain schedulers, which are used to continuously adjust the gains of adaptive Proportional-Integral-Derivative (PID) controllers. The objective of the control is to apply these gain schedulers in the control of nonlinear processes, so that the closed-loop system follows a desired dynamic, defined by a reference model. Thus, it is also the objective of this work to present a methodology to define a linear reference model from data obtained of a nonlinear system. The performance validation of the gain schedulers, according to the dynamic of the proposed reference model, is performed based on the IAE, IAU and IAOE metrics. The results show greater efficiency of the closed-loop control with the adaptive PID controller and gain schedulers compared to the fixed-gain PID controller.