SYSTEM IDENTIFICATION OF UAV DYNAMICS USING ECHO STATE NETWORKS AND CLASSICAL METHODS
Y. Ghanbary, A. M. Khoshnood
Aerospace Engineering Department, K. N. Toosi University of Technology, Tehran, Iran
Abstract: System identification plays a crucial role in determining the behavior and dynamics of flying objects and in modeling their dynamic systems. The availability of a suitable model of UAV can reduce some of the challenges of a vehicle control system designing. The selection of the appropriate method for identifying a system depends on the system structure and its complexity. Therefore, according to these conditions, in this paper, classical methods of system identification and Echo State Networks (ESNs) are employed to identify the dynamics of a small fixed-wing UAV. To evaluate the accuracy of the identification methods used, system output comparisons and transfer function coefficient analyses are conducted using appropriate metrics. The results indicate that the Least Squares method provides the best unbiased estimator under noise-free conditions, while the ESN and some of the classical methods examined demonstrate robust performance in noisy environments.
Keywords: System identification, UAV, least square, Echo state networks, Root mean square.
VOLUME 9 ISSUE 11 2025: 163 – 177