Sensorless control of switched reluctance generator drive based on neural networks

In this paper, the analysis, design, and implement- tation of a novel rotor position estimator for the control of switched reluctance generator(SRG) are presented. The rotor position is obtained by the improved minimal neural networks (NNs) whose inputs are the average phase current and the flux linkage. The flux linkage is calculated by flux integrator. Compared with the traditional NNs, it is demonstrated that a minimal NNs is easy to operate and attainable on a low-cost DSP. Experimental verification of the proposed control system applied to a wind energy conversion system is provided to demonstrate that the motor drive system has a small error of location observation and a good performance in generating operation. The control system of sensorless switched reluctance generator is simple, reliable and applicable to some harsh environments such as wind energy.