Speed sensorless flux and position control of induction machines based on pulse injection and multiple saliency extraction

Similar to other sensorless control methods based on signal injection, the resulting signal of the transient voltage excitation is influenced, for example, by the slotting saliency, the anisotropy saliency, as well as by the saturation based saliency and its load and flux dependencies. A separation algorithm to extract the slotting saliency and saturation saliency from the resulting signal is presented. It includes an artificial neural network (ANN) to identify, compensate disturbing influences and interactions of the saliencies. The slotting saliency component is used for calculating the rotor position in combination with the current model (rotor-equation). The saturation saliency component is first corrected using an ANN to get the estimated reference flux angle and then combined with a stabilized voltage model (stator-equation) for flux position calculation. This allows a calculation of the flux- and rotor position especially around zero fundamental frequency. The approach given in this paper attempts to combine both methods to provide an excellent performance of the sensorless control scheme in the whole frequency range. The transient signal is used to stabilize the voltage model what results in smooth output with the advantage that the dynamic performance is not reduced. At low frequencies it is assisted by the current model which provides additional stability.