Title: New optimal fast terminal sliding mode control combined with neural networks for modelling and controlling a drone quadrotor
Authors: Najlae Jennan; El Mehdi Mellouli
Addresses: Laboratory of Engineering, Systems and Applications, Sidi Mohammed Ben Abdellah University, Fez, Morocco ' Laboratory of Engineering, Systems and Applications, Sidi Mohammed Ben Abdellah University, Fez, Morocco
Abstract: This study concerns the modelling and control of a drone quadrotor which is a multi-input multi-output nonlinear system. The fast terminal sliding mode controller combined with particle swarm optimisation and neural network is presented to control the system, in order to solve high nonlinearity and cross-coupling problems. Generalising the control laws requires the complete state and all the system dynamics. However, not all states are accessible to the control laws and some dynamics cannot be modelled. Therefore, a triangular observer is proposed to estimate the system hidden states based on neural network to approximate the unmodelled dynamic part. We join a supplementary term in the control laws to reduce modelling errors and disturbances. Then, we use particle swarm optimisation to optimise important coefficients to achieve better results. The control laws come from the stability study in sense of Lyapunov. The results simulations confirm the effectiveness of the proposed control strategy.
Keywords: MiMo nonlinear system; drone quadrotor; fast terminal sliding mode control; triangular observer; neural network; particle swarm optimisation.
DOI: 10.1504/IJAAC.2023.134555
International Journal of Automation and Control, 2023 Vol.17 No.6, pp.595 - 612
Received: 23 Sep 2022
Accepted: 20 Jan 2023
Published online: 27 Oct 2023 *