Modelling of Direct Current (DC) Motor for Performance Improvement using Model Parameter Estimation

Abstract

A key concern with the optimal use of direct current (DC) motor is model updates. In this article, DC motor models from probability density function (PDF) data have been obtained using the system identification method in MATLAB. At first, noise-free input and output data were used to obtain the parameters of the AutoRegressive with eXogenous input (ARX) model structure. Thereafter, the DC motor subjected to random noise had its input and output data used to obtain the ARX model parameters. However, the resulting model from the noisy data differed from the noise-free data model. To minimize this difference, 200 input/output data set for the noisy DC motor were generated and used to obtain the model parameter for each data set. The mean of the 200 data set was obtained, and the resulting model approached the noise-free model as much as possible. The root-mean-squared error of prediction stands at 3.9426E-17, 3.35E-2, and 3.15E-2 for noise-free, noisy, and mean of noisy DC motor model, respectively, showing accuracy of result for noise-free system. The article provides a way by which the model of the DC motor can be updated when it is not convenient or possible to measure the DC motor parameters when used for motion actuation. The parameters of the selected model structure are therefore estimated from the DC motor input/output data instead of the DC motor parameters itself.