Metamodelagem Kriging aplicada no controle de mancais magnéticos ativos

Abstract

Active Magnetic Bearings (AMBs) have a wide range of applications in the industry, from turbo-molecular pumps to compressors and high-speed turbines for power generation. The rotor is levitated and supported by magnetic forces generated by the magnetic actuators present in the bearings. This allows for contactless operation, favoring the application of AMBs in rotating Machinery by eliminating friction and material wear. Their use also addresses issues associated with lubrication, common in most other types of bearings, reducing the need for maintenance interventions and thereby increasing operational uptime. The magnetic forces used for shaft levitation exhibit na unstable nature, necessitating the use of a control system for operational safety. The implementation of a closed-loop control system is associated with a position sensor system, filters, controllers, and power amplifiers. In recent years, research has focused on the development and implementation of controllers aiming to ensure system stability and robustness. Among linear control techniques, adaptive PID controllers, coupled with fuzzy or neuro-fuzzy logic, have shown promising results. Within the scope of nonlinear techniques, methodologies like sliding modes have shown excellent results, as their modeling is robust in the face of system uncertainties. However, such techniques often require a high degree of complexity in the development of the control law or experimental implementation. In this context, this work proposes the use of metamodels for the control of a rotor supported by two AMBs. The performance evaluation of the proposed controller is conducted with the rotor operating at different speeds, and various strategies are employed in terms of vibration amplitude for its construction. The results demonstrate that the Kriging metamodel ensures shaft stability across various rotor operating speeds.