Development and Experimental Tests of a Simple Neurofuzzy Learning Sensorless Approach for Switched Reluctance Motors

Abstract

Despite becoming competitive with ac and dc machines, the necessity for a shaft position transducer makes switched reluctance (SR) machines lose their low cost advantage, mainly as low power machines such as fans and pumps. Many techniques have been proposed for indirect rotor position detection for SR machines. However, their characteristics can be summed up as being based on a lookup table plus an interpolation algorithm, making them specific to a particular machine. For economic reasons and also dynamic performance, sensorless algorithms need a learning mechanism to allow them to adapt to a new SR machine or even adapt to changes in the SRM parameters. This paper presents a novel methodology for position sensor elimination for SR machines. Using the voltage from each conducting phase and the reference current signal as inputs, the rotor speed is first obtained as the output of a neurofuzzy learning system used as a “virtual” speed sensor. Then, the rotor position is determined by integrating the estimated value of speed. The effectiveness of the proposed sensorless technique was investigated through a series of real-time experiments on an SR drive system. The experimental results show that the suggested “virtual” speed sensor and corresponding rotor position can operate well in a sensorless SR speed control system.