The phase current method was used in [3]. In this method, rotor position is detected when the overlap of the rotor pole and stator pole begins and the phase inductance starts to increase. Phase advance excitation is used in which the torque production phase is turned on at its minimum inductance region. The phase current then increases and when the stator and rotor poles start to overlap, the increment in the phase inductance causes the slope of the phase current to change. An inequality of the rate of change of phase current at this position is defined where the slope of the phase current before the overlapping is always larger than after the overlapping. Only one rotor position is detected over a rotor pole pitch since overlapping occurs only once from unalignment towards alignment for motoring operation. By detecting the overlapping angle, a constraint is imposed on the turn-on angle which must occur before the rotor pole and stator pole start to overlap.
The neural network method was used in [4] and [5]. In both the methods, neural networks were used to learn the non-linear electrical characteristics of a SR motor. Since the electrical characteristics are dependent on the rotor position, the neural networks which were trained to learn these characteristics, can provide useful information on the rotor position.
