Deep Learning Based Predictive Analysis of BLDC Motor Control

Abstract

Predictive Analysis using deep learning techniques are emerging in several engineering domains. Artificial Neural Networks consists of network of layers to learn, process and predict from the data. Their implementation in electrical and electronics engineering has consistently helped in producing intelligent industries with effective results. The brushless direct current (BLDC)motor generates magnetic fields by switching DC current to the motor windings using electronic closed-loop controllers. BLDC motors require low maintenance, and exhibit high speed and sufficient torque capacity and hence are used in various applications. This motor has an edge over other motors because of its superior performance and ease with which the power converters can regulate its speed. This article describes a technique for varying the speed of a BLDC motor that involves altering the voltage of bridge converter, which feeds the motor winding. The speed control is carried out with a speed controller (PI-based). The motor is modelled in MATLAB/Simulink, and a PI controller is employed to provide the speed control. Simulated waveforms of EMF signals are achieved along with rotor speed, Hall Effect signals, electromagnetic torque, and PWMsignals. Artificial neural networks (ANN) are used to forecast the corresponding parameters, and they are fed with the gathered data to produce results that are reasonably close to the results from the simulations. Hence, both the simulation-based approach as well as the predictions from the data provided, yield satisfactory outcomes.