Pumping kite flight control based on deep learning
Year: 2026
Type: conference-paper
Abstract: This paper proposes a flight control method for pumped kites based on proximal policy optimization. The coordinate system for pumped kites is established, and the basic kinematic equations are constructed. Subsequently, the fundamental idea of proximal policy optimization is discussed, and a flowchart is drawn. Combining the necessary factors for kite control and the preset trajectory, a control scheme based on deep learning is designed. A simple yet comprehensive kite dynamics model is proposed, and the calculation results are analyzed, along with the characteristics of parameter variations. Finally, through simulation verification, the PPO algorithm can converge to a fixed value and remain relatively stable during the training process, demonstrating good exploration and wind energy collection capabilities.
Authors: Honglin Zhang, Jing Wang, Shurui Shen
Institution: Xi'an Siyuan University
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204968336