Comparison between PSO, NE, QR, SVD methods for least squares DC motor identification

This paper explores the application of the Particle Swarm Optimization (PSO) algorithm for parameter estimation of a Nonlinear Auto-Regressive with Exogeneous Model (NARX) of a Direct Current (DC) motor. The two-step identification step consists of structure selection and parameter estimation. The structure selection process was based on methods from our previous works, while the parameters were estimated using PSO. The propose algorithm was compared with several popular Linear Least Squares (LLS) estimation methods (Normal Equation (NE), QR Factorization (QR) and Singular Value Decomposition (SVD)) found to be comparable with them.