Complex variable neural network controller for port-Hamiltonian systems with non-holonomic constraints
Authors
Fernando Serrano, Saim Ahmed, Ahmad Taher Azar, Ahmed Redha Mahlous
Abstract
This study presents a complex variable neural network (CVNN) controller designed for unmanned aerial vehicles (UAVs) using the port-Hamiltonian formulation with non-holonomic constraints. The approach begins by deriving the dynamic model of a quadrotor UAV in port-Hamiltonian form, where the Hamiltonian is constructed from the system’s total kinetic and potential energies. Dirac structures are implemented to preserve the system’s structural properties while incorporating non-holonomic constraints specifically in the attitude loop. The proposed control architecture decouples the attitude and position control loops, with each utilizing a CVNN controller designed through Lyapunov stability analysis. Theoretical proofs establish the stability of both control loops, while numerical simulations demonstrate the effectiveness of the proposed approach for trajectory tracking tasks. Comparative analysis against existing controllers shows superior performance in reducing root mean square errors while maintaining reasonable control effort. The results confirm that the CVNN controller effectively manages the UAV’s dynamic behavior even with non-holonomic constraints limiting the vehicle’s orientation range.
Citation
- Journal: An International Journal of Optimization and Control: Theories & Applications (IJOCTA)
- Year: 2026
- Volume: 0
- Issue: 0
- Pages: 025450195
- Publisher: AccScience Publishing
- DOI: 10.36922/ijocta025450195
BibTeX
@article{Serrano_2026,
title={{Complex variable neural network controller for port-Hamiltonian systems with non-holonomic constraints}},
volume={0},
ISSN={2146-5703},
DOI={10.36922/ijocta025450195},
number={0},
journal={An International Journal of Optimization and Control: Theories \& Applications (IJOCTA)},
publisher={AccScience Publishing},
author={Serrano, Fernando and Ahmed, Saim and Azar, Ahmad Taher and Mahlous, Ahmed Redha},
year={2026},
pages={025450195}
}