Fault diagnosis for magnetic bearing systems

A full fault diagnosis for active magnetic bearing (AMB) and rotor systems to monitor the closed-loop operation and analyze fault patterns on-line in case any malfunction occurs is proposed in this paper. Most traditional approaches for fault diagnosis are based on actuator or sensor diagnosis individually and can solely detect a single fault at a time. This research combines two diagnosis methodologies by using both state estimators and parameter estimators to detect, identify and analyze actuators and sensors faults in AMB/rotor systems. The proposed fault diagnosis algorithm not only enhances the diagnosis accuracy, but also illustrates the capability to detect multiple sensors faults which occur concurrently. The efficacy of the presented algorithm has been verified by computer simulations and intensive experiments. The test rig for experiments is equipped with AMB, interface module (dSPACE DS1104), data acquisition unit MATLAB/Simulink simulation environment. At last, the fault patterns, such as bias, multiplicative loop gain variation and noise addition, can be identified by the algorithm presented in this work. In other words, the proposed diagnosis algorithm is able to detect faults at the first moment, find which sensors or actuators under failure and identify which fault pattern the found faults belong to.