Abstract
This paper proposes a new approach for separation of saliency signals of the induction machine at very low speed and heavy loads. The change of current responses due to the transient voltage pulse excitation is employed to obtain the inherent impedance saliency information. The proposed separation approach uses the discrete wavelet transform (DWT) in combination with Taylor's expansion and least square error (LSE) methods. The DWT removes the higher and lower harmonic orders to extract the operating frequency band based on the reference speed. Then, Taylor's expansion modeling processed by LSE algorithm is applied to extract the slotting saliency signal from the nearby frequency of the inter-modulation signal. The extracted slotting saliency signal is employed to estimate the rotor/flux position using the phase locked loop observer and current model. Experimental results are provided to prove the effectiveness of the proposed new separation approach at very low speed including zero electrical speed operations. It is found that the proposed new approach is accurate at very low speed and heavy load operations with less noise compared to previous methods.