Sintonia ótima de controlador PID descentralizado para processos MIMO.

Abstract

The most widely used controller inindustrial units is the Proportional-Integral-Derivative PID. Despite theextensive use, formost of these controllers, the performance is below the desired result, since they are poorly tuned. Thus, due to the need to improve the performance of industrial control loops, this thesis develops three methods for SISO and MIMO processes which consider the conflict between robustness and performance by solving optimization problems. The first contribution is a tuning method for PI controllers for SISO processes, whose integral gain is computed by a nonlinear optimization problem. The second contribution is related to the design of decentralized PID controllers for weakly coupled MIMO processes. This method is formulated as a linear programming problem and the loop interactions are considered by Gershgorin bands. The third contribution is related to the iterative procedure of tuning the decentralized PID controller for MIMO processes and the loops interactions are considered by the Effective Open-Loop Process Transfer Functions. For all proposed methods, minimum bounds on robustness index are design parameters and the closed-loop stability is guaranteed. The tuning results of the proposed methods are applied in both simulated processes as industrial processes and compared to results of tuning techniques already considered renowned in the literature.