Saliency based sensorless control of induction machines at frequency overlap of signal components

This paper presents sensorless control at the critical point of operation existing at overlap of saliency frequencies which, so far has not been addressed by research. Considering induction machines, the resulting signals from injection based sensorless control methods are composed of saliency components caused by saturation, slotting or anisotropy, as well as inter-modulation effects. The points of operation at the spectral overlap of any two of the resulting signal's components are critical in terms of control performance as then a reliable separation is only possible feed forward. A well designed and trained neural network is applied to sequentially separate the different components. Up to now research focused on investigating zero speed, zero frequency operation neglecting the mentioned critical overlap area. In this paper first experimental results are presented verifying sustained sensorless operation at overlap of saliency signal's components.