Torque Estimation with Artificial Intelligence Methods in a Brushed Geared Dc Motor

Abstract

Today, with the advances in electricity-electronics, the usage areas of DC motors have increased considerably. DC motors have high starting torques and speed can be adjusted over a wide range. In the present experimental study, different weights connected to the motor shaft were rotated at different speeds, at variable distances, in the angle range of 0º-345º degrees. Thus, different torque values produced by the DC motor were observed. In some cases, the amount of torque produced at low rotational speeds may have non-linear values. This allows the use of artificial intelligence methods for accurate torque estimation. In the present study, different uses of Elman Backpropagation Neural Network (EBNN) and General Regression Neural Network (GRNN) are given for the estimation of the best torque values. Performance comparisons were made according to mean square error (MSE), regression coefficient (R2), root square error (RSE), and mean absolute error (MAE) values.