An Approach for the Holistic Integration of Fault Detection and Diagnosis into Active Magnetic Bearing Systems

Present research on fault detection and diagnosis on magnetic bearing systems aims on the diagnosis of the state of single components, such as sensor and actuator faults or fault states of the plant. Although these methods often have the potential to detect more than one fault state, there is no fault detection method allowing the detection of all possible faults . This paper describes aconceptional approach to combine different fault detection methods aiming on a holistic online diagnosis of the active magnetic bearing system consisting of the functional components controller, actuator, plant and sensor. On the example of a centrifugal pump in active magnetic bearings this approach gives the possibility to combine fault detection methods detecting internal faults of the bearing system as well as the determination of the overall system status, e.g., the duty point of the pump, based on the signals of the active magnetic bearing system. This approach allows integrating new functions into the active magnetic bearing system via an enhancement of the software without changing the widely developed hardware of active magnetic bearings.