Position sensorless control of a SRM drive using ANN-techniques

Nowadays speed-sensorless electromechanical drives are feasible for many applications. The elimination of the mechanical transducer increases reliability, reducing costs. As far as synchronous machine drives are concerned this issue is critical since the rotor position is fundamental to derive the proper switching sequence. If a SRM is used as actuator, different sensorless techniques have been proposed, essentially based on the rotor position detection by reluctance variation. Being the SRM a highly non-linear machine, it is an ideal candidate for the application of Artificial Neural Networks (ANNs). In this paper, at first the results of a mixed closed loop scheme for the sensorless identification of a SRM drive are presented. A radial basis function ANN has been used, since it is well suited to simulate a non-linear system in a wide operating area. Then the proposed ANN is applied within a standard control scheme as a position sensor. Simulation results show the effectiveness of the proposed method, since a good agreement is reached with experimental results obtained from a SRM drive with a standard position sensor.