Condition monitoring of active magnetic bearing systems

Thesis (Ph.D. (Electrical and Electronic Engineering))--North-West University, Potchefstroom Campus, 2007In this thesis, the author contextualises condition monitoring of active magnetic bearing (AMB) systems and proposes the real-time condition monitoring of AMB systems. Three real-time fault detection, diagnosis, correction and identification schemes for vibration forces on the rotor of a rotational AMB system are proposed. Two AMB systems were used to conduct this research. The one was a fully suspended 250 kW water cooling AMB pump from which historical fault data was obtained and the other was an experimental double radial AMB test rack on which the three real-time schemes were implemented. The historical fault data obtained from the AMB pump were categorised into subsynchronous, rotor synchronous and supersynchronous vibration forces. It was decided to use the historical fault data to induce the same fault conditions that occurred on the AMB pump on the AMB test rack. Various fault detection techniques were implemented on the historical fault data. This was done to obtain the best technique for a specific fault condition. These techniques together with the historical fault data were used in the design of the three real-time schemes. The focus of this research is on external faults, since the research of internal faults to the magnetic bearing control system is well established. Various schemes exist where off-line diagnosis and identification of internal and external faults are performed on AMB systems, but the area of real-time identification and correction of external faults still needs to be explored. This study proposes three real-time detection, diagnosis, correction and identification schemes for external faults to the magnetic bearing control system of rotational AMB systems. These schemes use the available AMB sensors and actuators to perform condition monitoring and correction, since machine components in industrial applications operate in harsh environments and are not always easily accessible to install condition monitoring equipment. The last scheme focuses on real-time multiple frequency fault detection, diagnosis, correction and identification of external faults on rotational AMB systems, since more than one fault can occur simultaneously on the AMB system. Analysis of a multiple frequency fault by a single frequency fault scheme causes incorrect identification and correction of the fault. Four articles were submitted for publication on the three real-time schemes. The three real-time schemes perform three main tasks: 1) fault detection, 2) fault diagnosis and error correction and 3) fault identification. Displacement and current masking were performed during the fault detection stage. The vibratory amplitudes and frequencies were extracted by means of the Wigner-Ville distribution. Discrete sampling of the displacement and current signals of the physical AMB system was performed by the dSPACE 1104 controller board. Pattern recognition techniques, statistical diagnosis, cascaded fuzzy logic and pattern construction were used to calculate fault features during the fault diagnosis and error correction stages. Error correction algorithms were used to correct the fault during the error correction stage. During the fault identification stage, data fitting, fuzzy logic, IS0 standards and pattern recognition were used to calculate the type, parameters, vibratory level and zone of the vibration force. A historical fault database was used during the identification process. Correction factors were calculated to observe the improvement each scheme has over the optimised PID controller when vibration forces are applied. Scheme 3 provided the best result for improving subsynchronous vibration forces, scheme 1 the best for rotor synchronous vibration forces and scheme 2 the best for supersynchronous vibration forces. Scheme 3 provided the best overall result with the best average improvement. All three real-time schemes were able to correct and minimize vibration forces to a stable operating condition. The three real-time schemes only stabilize the rotor with respect to the stator and do not remove the vibration force. When correction forces are applied to the AMB system, to correct the effect of the vibration forces, it may increases the stress of other critical components e.g. the power amplifiers and system base, which may cause components to be damaged or break down. These stressed components need to be identified by the user as critical or non-critical and the necessary steps must be taken to operate the AMB system under the fault condition or to shut down the system and repair the fault. The maximum current capability and bandwidth of the power amplifiers, run-time of the DSP processors and bandwidth of the sensors were the main factors limiting the applicability of the real-time schemes. Since the power amplifiers and sensors were designed and specified for normal operational bandwidth, it was not possible to fully evaluate supersynchronous vibration forces.Doctora