Review and Comparative Analysis of System Identification Methods for Perturbed Motorized Systems

This paper reviews recent advancements in system identification methods for perturbed motorized systems, focusing on brushed DC motors, brushless DC motors, and permanent magnet synchronous motors. It examines data acquisition setups and evaluates conventional and metaheuristic optimization algorithms, highlighting their advantages, limitations, and applications. The paper explores emerging trends in model structures and parameter optimization techniques that address specific perturbations such as varying loads, noise, and friction. A comparative performance analysis is also included to assess several widely used optimization methods, including least squares (LS), particle swarm optimization (PSO), grey wolf optimizer (GWO), bat algorithm (BA), genetic algorithm (GA) and neural network for system identification of a specific case of a perturbed DC motor in both open-loop (OL) and closed-loop (CL) settings. Results show that GWO achieves the lowest error overall, excelling in OL scenarios, while PSO performs best in CL due to its synergy with feedback control. LS proves efficient in CL settings, whereas GA and BA rely heavily on feedback for improved performance. The paper also outlines potential research directions aimed at advancing motor modeling techniques, including integration of advanced machine learning methods, hybrid learning-based methods, and adaptive modeling techniques. These insights offer a foundation for advancing motor modeling techniques in real-world applications.