Affiliations: Federal University of Uberlândia, School of
Mechanical Engineering, Campus Santa Monica, Uberlândia, Brazil | Université de Lyon, Laboratoire de
Mécanique des Contacts et des Structures, Institut National des Sciences
Appliquées de Lyon, Lyon, France
Note: [] Corresponding author: J. Mahfoud, Université de Lyon,
Laboratoire de Méanique des Contacts et des Structures, LaMCoS, UMR CNRS
5259, Institut National des Sciences Appliqués de Lyon, Bâtiment
Jean d'Alembert, 18-20, rue des Sciences, 69621 Villeurbanne Cedex, France.
Tel.: +33 4 72 43 89 39; Fax: +33 4 72 43 89 30; E-mail:
[email protected]
Abstract: This work deals with the design and the assessment of
electromagnetic actuators (EMAs) for the control of rotating machines. The
system studied has a hybrid bearing that exhibits nonlinear behavior. The
system is composed of a horizontal flexible shaft supported by two ball
bearings at one end and a roller bearing that is located in a squirrel cage at
the other end. Four identical EMAs supplied with constant current are utilized.
The EMAs associated to the squirrel cage constitutes the hybrid bearing. The
aim is to develop a strategy in order to define and to identify a reliable
model necessary for the control of rotating machinery in the presence of
localized non-linearity. The identification strategy consists in modeling the
system with as many sub-models as needed that are identified separately. This
enables obtaining a straightforward modeling of rotating machinery even in the
case in which system components are frequency or time dependent. For the system
studied, two sub-models were necessary. First the EMAs were modeled by using
classical equations of electromagnetism and then identified experimentally.
Then, a linear model of the shaft mounted on its bearings was defined by using
the finite element method and was identified successfully. The model of the
system was adjusted after assembling the different identified sub-models. The
identification is carried out by using a pseudo-random search algorithm. The
model of the system is then assessed for different configurations. The results
obtained demonstrate the effectiveness of the developed strategy.
