Sensor less Control of BLDC Motor Using Anfis

Abstract

This study presents a sensor less control strategy for brushless DC (bldc) motors using an adaptive neuro fuzzy inference system (anfis) to generate virtual hall effect signals from the measured three-phase back-emfs. The proposed approach eliminates the need for physical position sensors by producing digital hall equivalent signals that directly interface with conventional commutation logic circuits, providing a seamless drop-in replacement for the hall sensors.

 The controller was modelled and simulated in MATLAB/Simulink with a closed-loop PI speed controller (K­_p = 0.013, K_i = 16.61). The anfis network was trained on data collected at a reference speed of 3000 rpm from a sensor equipped bldc test bench in Simulink. The trained model was validated under both loaded and unloaded operating conditions. A unit step load torque was applied at t=0.1s to evaluate dynamic performance. The results demonstrate that the anfis based system exhibits a dynamic response comparable to the conventional hall sensor-based control.