Abstract
A novel separation procedure applied to overlap condition of saliencies in induction machine is presented. The machine's current reaction to voltage step excitation is used to obtain saliency signals. The proposed procedure is based on a digital signal processing implementation of the discrete sine and cosine transforms for extracting the slotting saliency signal at close and equal frequencies of the slotting and saturation saliencies under heavy load operation. The proposed algorithms are processed based on an online adaptive threshold. Using the proposed separation algorithm guarantees precise estimation of rotor/flux angle and efficient sensorless control of the induction machine. A sensorless induction motor drive is implemented in the laboratory. Extensive experimental results are presented. A performance comparison between the proposed algorithm and previous works is provided. The robustness and effectiveness of the applied signal processing algorithms for sensorless torque control are verified at low speed and heavy load.