Speed control of induction motor without rotational transducers

This paper describes a newly developed speed sensorless drive based on neural networks. A backpropagation neural network is used to provide real-lime adaptive identification of the motor speed, The estimation objective is the sum of squared errors between a target trajectory and the neural network model output. A backpropagation algorithm is used to adjust the motor speed, so that the neural model output follows the target trajectory. Backpropagation forces the estimated speed to follow precisely the actual motor speed. The zero-speed crossing phenomena is also described, and experimental results are presented and analyzed.