Intelligent and adaptive on-line direct electromagnetic torque estimator for switched reluctance motors based on artificial neural networks

Torque estimation is an important task in the implementation of controllers for electric motor drives. In the case of switched reluctance motors (SRM), the computation technique should account for the nonlinearity of the magnetic material and the variations of the flux-linkage characteristics with rotor position and current level. Also, it is essential to adapt to the characteristics of the individual SRM (due to manufacturing deviations) when requiring high accuracy in this task. This paper presents a technique based on artificial neural networks (ANN) that estimates the electromagnetic torque developed by learning the characteristics of the SRM drive system using online measurements. The technique is then illustrated by applying it in simulations for predicting the electromagnetic torque.