Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Surrogate active vibration control of a rotating shaft supported by magnetic bearings
Submission Author:
Maria Carolina Albuquerque de Souza Santos , MG
Co-Authors:
Maria Carolina Albuquerque de Souza Santos, Aldemir Ap Cavalini Jr, Marco Túlio Santana Alves, Marcelo Costa Tanaka, Valder Steffen Jr
Presenter: Maria Carolina Albuquerque de Souza Santos
doi://10.26678/ABCM.COBEM2023.COB2023-1144
Abstract
Active Magnetic Bearings (AMB) have a wide range of applications in industry, from molecular turbo-pumps to compressors and high-speed turbines for power generation. The rotor is levitated and supported by magnetic forces, which are generated by magnetic fields arising along the bearing structure that contains electromagnetics actuators. This allows non-contact operation, which favors the application of AMB in rotating machines, as it eliminates overheating, friction, and wear of basic rotor materials. Additionally, it eliminates lubrication problems present in most mechanical systems. The lack of lubrificating fluid allows its use in specialized applications, like the bio-medical field. Therefore, interventions for equipment maintenance are less necessary, increasing the safe operation time of the machine. The magnetic forces generated for shaft levitation purposes are determined by using an active vibration control approach. The rotor is intrinsically unstable, and because of that, it is necessary to control the generated magnetic field to prevent the shaft from exceeding the safety limits of the bearings. The closed-loop active vibration control system encompasses position sensors, filters, power amplifiers, and electromagnetic bearings. In recent years, several works have been devoted to the development and implementation of controllers aiming to guarantee the robust stability of the system. Among the linear control techniques, the adaptive PID controller (proportional, integrative, and derivative), associated with fuzzy or neuro-fuzzy logic, has shown good results. Regarding the nonlinear control techniques, the sliding mode modes approach has shown interesting results. However, such techniques often lead to high complexity concerning their development and experimental implementation. In this context, the present contribution is devoted to applying a Kriging surrogate to control a rotor supported by two AMBs. The proposed approach is evaluated by considering the rotor operating at different rotation speeds and unbalanced conditions. Additionally, various strategies concerning the vibration responses used to train the surrogate models are considered in this work.
Keywords
Active magnetic bearing, Kriging surrogate model, Vibration control, Kriging metamodeling
