NARMAX identification of DC motor model using repulsive particle swarm optimization

This paper explores the usage of repulsive particle swarm optimization (RPSO) to perform non-linear auto-regressive with exogenous input (NARMAX) system identification of direct current (DC) motor. The NARMAX model was constructed using a recurrent artificial neural network (ANN) model by Rahim and Taib and Yassin et al. The comparison result was made between RPSO method and inertia weight-based PSO method by Yassin et al. to train the NARMAX model. The result shows that RPSO yielded comparable performance to the inertia weight-based PSO method in determining NARMAX coefficients in the model.