Saliency Harmonic Induction Motor Speed Estimation Using Artificial Neural Networks

Several methods have been proposed to measure the speed of an induction machine without the need for external speed transducers. These sensorless methods typically use observers, require injected signals, or are dependent on the detecting a specific harmonic. This paper investigates using neural networks to identify the speed of a machine using a range of harmonics without specific knowledge of machine parameters. As a proof of concept currents measured from various machines are analyzed with the proposed method. This analysis showed that such a system has the potential for relatively accurate speed determination.