An improvement in dynamic behavior of single phase PM brushless DC motor using deep neural network and mixture of experts

Abstract

Brushless DC motors play a vital role as a workhorse in many applications, especially home appliances. In the competitive world of the day, a brushless DC motor is a wise choice for many applications because of its high power density, a simple driving circuit, and high efficiency. Accordingly, demonstrating the feasibility of a new controller on this type of motor has undoubtedly paramount importance. Two methods of speed controllers, namely linear-quadratic regulator, and proportional-integral-derivative, are mixed using a mixture of experts (MoE) for a single-phase PM brushless DC external rotor motor. The dynamic model of the SP PM BLDC ER motor characterizes the behavior of the motor, involving cogging torque and electromotive force (EMF) gained from 2D finite element analyses. The motor is supplied by a pulse width modulation inverter with a constant voltage source. The results disclose that the SP PM BLDC performance is enhanced and more robust during load disturbance. ANSYS and MATLAB environments are used for obtaining finite element analysis and dynamic analysis of single-phase PM brushless DC external rotor motors, respectively. The merits of the proposed approach are validated through implementing a low-scale experimental setup.