Fault tolerant machine drives are key enabling technologies for safety critical applications such as electric vehicle traction, and aerospace power generation, actuation and propulsion. High performance in healthy conditions and excellent fault tolerance against various faults are required for a fault tolerant drive, however, these two aspects usually conflict with each other. Thus, this PhD study aims to develop a fault tolerant machine drive which exhibits high performance and good fault tolerance, and can be realised in a simple and cost-effective manner.
First, a novel triple redundant 3x3-phase permanent magnet assisted synchronous reluctance machine (PMA SynRM) with segregated windings is proposed. Its performance under healthy conditions and its ability to tolerate various faults with appropriate mitigation measures are investigated and assessed. Based on outcomes of the investigation, a 40kW machine is designed to tolerate all key electrical faults, including the worst single turn short circuit, and is optimised to maximise the efficiency in healthy conditions whilst satisfying the electrical, thermal and mechanical constraints.
To analyse and realise a fault tolerant machine drive, fault modelling and fault detection techniques are essential. Thus, a general model is proposed based on the magneto-motive force (MMF) decomposition. The model is capable of predicting the machine behaviour in various operation modes, including the healthy condition, open circuit, short circuit and inter-turn short circuit fault with different number of turns and different coil locations.
With the aid of the fault modelling technique, a turn fault detection technique is developed using the 2nd harmonics in the instantaneous reactive and active powers as fault indicators for motoring and generating modes, respectively. By cross-reference of the fault indicators of the three 3-phase sets, the technique can detect the turn fault during transient without false alarm.
The optimised machine drive is constructed and the developed fault detection technique together with fault mitigation strategies is implemented in a DSP based controller. The performance of the drive under healthy conditions and its fault tolerant capability are validated by extensive tests. The accuracy of the fault modelling and the effectiveness of the detection technique are also experimentally evaluated. The test results demonstrate that the developed fault tolerant machine drive can be a competitive candidate for safety critical applications.
