Estimation of loss factor and system parameters of active magnetic thrust bearing using RBF neural networks and differential evolution

An active magnetic bearing (AMB) is a mechatronical product which suspends a rotating element called rotor in position without any mechanical contact. This characteristic makes AMB attractive in high speed, high precision and high vacuum applications. Estimation of losses and system parameters of AMB using artificial intelligent techniques is rarely attempted in literature. In the present work using the differential evolution (DE) algorithm in combination with radial basis function neural networks (RBFNN) the loss factor, current and other system parameters such as position and current stiffness have been estimated at any air gap and a specified force.