An Advance Angle Control-based SRM Drive for SWPS using Artificial Neural Networks for Rotor Position Estimation

This paper presents a low-cost solution of switched reluctance (SR) drive employed for water source application using an artificial neural network for rotor position estimation. The drive eliminates the use of expensive mechanical position sensor, and instead uses a sensorless scheme to obtain a rotor position information. Because of its highly non-linear magnetic characteristics, SR motor has a challenging task in its mathematical modelling for the control of drive. The neural network modelled in this paper uses a pre-collected magnetic characteristics and is trained offline using a Levenberg–Marquardt (LM) algorithm in a MATLAB/Simulink environment. Since, the number of neurons plays a prominent role in training the network and the fitting level, the model is optimized in using a number of neurons in the hidden layers of the network. Besides using a sensorless scheme, the drive employs an advance angle control for the speed control of SR drive which eliminates the switching losses induced by the conventional hysteresis controller. Overall performance of the drive is evaluated and presented for accounting its suitableness for the solar photovoltaic water pumping system (SWPS).