6DOF Aircraft Landing Gear System with Magnetorheological Damper in Various Taxing and Touchdown Scenarios.

This manuscript presents a new approach to describe aircraft landing gear systems equipped with magnetorheological (MR) dampers, integrating a reinforcement learning-based neural network control strategy. The main target of the proposed system is to improve the shock absorber efficiency in the touchdown phase, in addition to reducing the vibration due to rough ground in the taxing phase. The dynamic models of the aircraft landing system in the taxing phase with standard landing ground roughness, one-point touchdown, two-point touchdown, and third-point touchdown are built as the first step. After that, Q-learning-based reinforcement learning is developed. In order to verify the effectiveness of the controller, the co-simulations based on RECURDYN V8R4-MATLAB R2019b of the proposed system and the classical skyhook controller are executed. Based on the simulation results, the proposed controller provides better performance compared to the skyhook controller. The proposed controller provided a maximum improvement of 16% in the touchdown phase and 10% in the taxing phase compared to the skyhook controller.