The objective of this research is to develop and demonstrate a commercially viable technique for self-sensing in active magnetic bearings by realizing a virtual position probe using measured amplifier voltage and current. Two potential approaches will be pursued. The first treats the system comprised of amplifier, magnetic actuator, and rotor journal as a linear plant parameterized by rotor position: a parameter observer takes advantage of persistency of excitation provided by switching amplifiers to determine this parameter (rotor position). The second approach exploits a linear periodic model of the full rotor in combination with the actuator and constructs a state observer for this system. Virtual probes (observers) based on both approaches will be implemented either in hardwired electronic form or, ideally, as a digital field programmable gate array. In either case, the performance of the resulting estimation (virtual probe) hardware will be examined using an existing magnetic bearing supported rotor and compared to conventional eddy current type proximitors.
The principal benefit of this research is to reduce the complexity of hardware in active magnetic bearings. By removing components from the harsh environment to which the bearings are exposed, the cost and reliability of these devices can be improved substantially. This greatly enhances the application of magnetic bearings to a number of important technologies ranging from micro gas turbines for distributed electric power generation to implantable artificial hearts for treatment of congestive heart failure and other similar cardiac ailments. Hardware complexity and reliability is a major obstacle to large-scale commercial use of magnetic bearings in these and many other applications: self-sensing is a pivotal step in reducing these problems. A tantalizing long term benefit and central goal of continuation of this research is development of truly "off-the-shelf" magnetic bearings which can be marketed in the same manner as conventional fluid film or rolling element bearings. By eliminating the position sensor, a significant hurdle is removed in approaching a critical "passivity" property which ensures that the bearing will stably support any stable rotor. This would mean that active magnetic bearings could be sold without tailoring them to each specific application, avoiding very costly engineering effort and stimulating the economies of scale critical to unlocking the nascent commercial potential of active magnetic bearing technology.
