Controladores regulatórios auto-sintonizados de distúrbios cíclicos baseados em lógica fuzzy

Abstract

Some industrial processes have intrinsic cyclic disturbances. In these cases, it is natural to use Iterative Learning Control (ILC) or Repetitive Control (RC), or even combinations of these with other controllers, such as the Repetitive GPC (R-GPC), which integrates the RC with the Generalized Predictive Controller (GPC). One of the main weaknesses of these controllers is when the frequency of the periodic signal varies, as there is a gradual loss in its efficiency. Thus, in this PhD Thesis, a Fuzzy Logic based technique is proposed to estimate the total number of samples contained in a periodic disturbance subject to small and unknown frequency variations. Therefore, the Adaptive Fuzzy ILC (AF-ILC) controller and the Adaptive Fuzzy Repetitive GPC (AFR-GPC) controller are proposed here. In the AF-ILC, the proposed Fuzzy Estimator is applied to the ILC controller, whereas, in the AFR-GPC the Fuzzy Estimator is applied to a structure composed of predictive controllers, allowing these controllers to minimize periodic disturbances with frequency changes over time. Before testing, the controllers are tuned offline via genetic algorithm (AG). To adapt these controllers for frequency change variation, the number of disturbance samples is estimated with the system operating in closed loop (self-tuning adaptive controller) using Fuzzy Logic. As a case study, these controllers are tested in computer simulations in a continuous casting plant to compensate for the bulging disturbance present in the mold level control. In addition to the simulations, the controllers are tested in a didactic plant consisting of a resistive-capacitive circuit where periodic disturbances are present, which, although not having the same structure as the casting system, allows demonstrating that the proposed controllers work for a real plant.