Adaptive wind direction strategy for neuro-network-based active disturbance rejection control of tandem twin-rotor aerial-aquatic vehicle

Abstract

The novel slender aerial-aquatic vehicle (AAV) with slender fuselages enhances underwater maneuverability compared to traditional AAVs, which prioritize aerodynamic efficiency. However, these novel vehicles face challenges in positioning accuracy and energy consumption due to rotor control coupling and the significant differences in frontal and lateral surface areas, especially under crosswind and wave interference in open sea areas. This paper proposes an adaptive wind direction (AWD) strategy, combined with the active disturbance rejection control (ADRC) based on a radial basis function (RBF) neural network to enhance positioning accuracy and reduce energy consumption. Firstly, the mathematical models of the slender AAV, varying wind fields and waves are established. Subsequently, an ADRC law is designed for the attitude and position control of the AAV, where an RBF-based Extended State Observer (ESO) is used for disturbance observation instead of the traditional ESO. Then, the adaptive wind direction strategy is employed. This strategy utilizes wind disturbance characteristics, disturbance observations and body surface area parameters to calculate the wind field angle. Based on this angle, the attitude is adjusted to minimize interference. Finally, simulations validated the effectiveness and robustness of the designed ADRC control law based on the RBF ESO. After applying the adaptive wind disturbance strategy, positioning accuracy improved by 2 to 10 times, and energy consumption decreased by 20 % to 80 %, compared to the neural network-based active disturbance rejection control that does not utilize adaptive wind direction strategy.