Performance Comparison and Analysis of Four- and Six-Switch Inverter Controls of BLDC Motor Employing Deep Learning

Abstract

Robust and effective control of brushless dc (BLDC) motors is paramount in modern-day motion control. The BLDC motor is known for its high speed, high torque, small size, low noise, and equally low maintenance requirements compared to the brushed DC motor. Nowadays, it can be found in the areas of robotics, aerospace, military, and industrial machines, among others. In this paper, two inverter topologies, the threephase six-switch driver circuit (TPSSDC) and the three-phase four-switch driver circuit (TPFSDC), are used to drive and control the speed of the BLDC motor. For the control technique, however, a fitting neural network from the deep learning (DL) toolbox is employed to train and improve the speed performance of the motor. Both TPSSDC and TPFSDC are simulated and tested in MATLAB Simulink, and the resultant output is analyzed graphically and analytically. Graphical observation shows that the TPSSDC control approach is superior in terms of reference tracking and has less ripple, and better rise and settling times when compared to the TPFSDC control approach, however, the TPFDC is considered for its low cost and simple circuitry. Numerically, the TPSSDC also outperforms at 1000 rpm with a 1.685 ms rise time and a 1.420 ms settling time compared to the TPFSDC, which rises at 6.526 ms and settles at 5.237 ms. At 3000 rpm motor speed, the TPSSDC is better, having a rise time of 5.815 ms and settling time of 4.048 ms, when compared to the TPFSDC, which rises at 11.277 ms and settles at 10.067 ms.