Projeto e otimização de controladores para rotores com mancais magnéticos: Projeto Robusto

Abstract

The present work is dedicated to synthesis of controllers used in supercritical rotors supported by magnetic bearings. First, a numerical/computational model was built in a MATLAB ®/Simulink environment as based on the specifications provided by the bench manufacturer. Then in conjunction with the experimental validation process, a representative model of dynamic behavior of the real physical system resulted. Next step was the design of three controllers for the system, namelly, the PID controller based on the characteristics provided by the manufacturer, the adaptive PID which is based on the update of its reference, and the PID Fuzzy, which permits the adjustment of the controller’s gains. The three controllers were evaluated based on their transfer functions in closed loop and from the point of view of their unbalance responce, which was conducted according to the guidelines of the API 684 standard and classified according to the criteria defined in the ISO 14839-2 standard . Considering the obteined results from the three previous controllers, it was possible to analyze the design difficulties and create the so-called Neuro Fuzzy Robust controller, which is designed so that the uncertainties found in the system are taken into account. It is worth mentioning that these uncertainties were not considered in the previous controllers. The proposed new controller was evaluated by using the same criteria applied to the previous controllers. Based on the obteined results, it was possible to point out that the main contribution of this research work is the systematization of the use of an optimal-robust Neuro Fuzzy controller for the study and design of supercritical rotors supported by magnetic bearings