Fuzzy-based nonlinear PID controller and its application to CSTR

Abstract

This study presents a new design method for a nonlinear variable-gain PID controller, the gains of which are described by a set of fuzzy rules. User-defined parameters are tuned using a genetic algorithm by minimizing the integral of absolute error and the weighted control input deviation index. It was observed in the experimental results on a continuous stirred tank reactor (CSTR) that the proposed controller provided performances: overshoot M p ≤1.25%, 2% settling time t s ≤1.71 s and IAE≤1.26 for set-point tracking, perturbance peak M peak ≤0.05%, 2% recovery time t rcy ≤3.97 s and IAE≤0.10 for disturbance rejection, and M peak ≤0.04%, t rcy ≤2.74 s and IAE≤0.04 for parameter changes. Comparison with those of two other methods revealed that the proposed controller not only led to less overshoot and shorter settling time for set-point tracking and less perturbance peak and shorter recovery time for disturbance rejection, but also showed less sensitivity to parameter changes.