Active magnetic bearings (AMBs) are currently used in a number of commercial and research applications. AMBs allow for higher rotational speeds than rolling element bearings while eliminating the need for lubrication. In addition, they also may be used concurrently as support bearings and non-invasive force sensors by monitoring bearing flux-density through the addition of Hall Effect probes or through measurement of bearing currents applied to magnetic circuit models of the bearing. Many current-based approaches are limited because of model assumptions due to unknowns in the final geometry/ assembly of the bearing itself resulting in unknown flux distribution. This paper addresses a new strategy for AMB current-based force measurement, the Multi-Point Method, which uses a system identification approach in order to improve the accuracy of force measurements in field conditions with no additional hardware. The Multi-Point Method described in this paper is a system identification approach to force measurement where both gap and forces are determined through the use of perturbation current application, recording of resulting states, and the use of an on-line algorithm. The Multi-Point Method has been shown to be capable of providing force measurements that are more accurate and more precise over a range of vertical rotor displacements (mimicking misalignment, assembly variations, and thermal growth effects on field gaps) than conventional single point current-based measurements that rely on assumed gaps in static tests. This paper presents the results of a proof-of-concept static test in a non-rotating two bearing rotor. Initial static experimental results demonstrate that the multi-point force predictions are within 1.03% of the known rotor force on average, with an average standard deviation of 0.83%. Comparatively, the conventional single point force predictions were within 5.76% of the known force with an average standard deviation of 6.17%. Dynamic effects due to eddy currents and hysteresis effects also degrade current-based force model accuracy and will be addressed in future work. The goal of this study is to demonstrate the importance of accounting for geometric parameters through a system identification approach to improve the fundamental current-based model accuracy. (C) 2006 Elsevier Ltd. All rights reserved.
