ANN based sensorless rotor position estimation for the switched reluctance motor

Abstract

The phase excitation pulse of the Switched Reluctance Motor (SRM) must be synchronized with the angular rotor position to ensure the continuous torque and rotation of the rotor and also to obtain the optimum performance of the SRM drive. In this paper Artificial Neural Network (ANN) based sensorless rotor position estimation technique is presented to fulfill the requirement of the position feedback for the SRM. MATLAB simulink environment is used to design a neural network and to simulate the proposed sensorless method which shows satisfactory result. An idea is presented to reduce the number of neuron for mapping the magnetic characteristics of the neural network which can reduce the complexity and computation burden without much affecting the performance of the SRM. Region of interest of the magnetic characteristics is described & discussed first time in this paper which helps to analyse a region of the magnetic characteristics where the significance of accuracy of the rotor position estimation is more compared to exterior region. © 2011 IEEE.