Novel flux linkage control of switched reluctance motor drives using observer and neural network-based correction methods

From the perspective of control of switched reluctance motor (SRM) drives, current control is typically used as an inner loop control method. In this paper, observer and artificial neural network (ANN)-based novel flux linkage control of SRM drives is presented and examined as an alternate approach to current control. The main advantage of flux linkage control is computational simplicity due to the insensitivity of controller gains to machine operation conditions, while current control depends on controller gains which are very sensitive to self-inductance of SRMs. Flux linkage control needs a reliable flux linkage estimator for desirable control of SRMs. Integration method to estimate flux linkage from measured phase voltages, currents and resistances is commonly used, but it is sensitive to measurement error and white noise. Another way to measure the flux linkage is to use a look-up table which is very sensitive to input currents because it is current- and position-based data. In this paper, a simple observer-based voltage and ANN-based current correction method is proposed to overcome the measurement error. Furthermore, ANNs with two layers and five neurons are applied to produce an acceptable flux linkage estimate at each corrected current and measured position, instead of a look-up table. Finally, simulation results are presented to validate its performance.