Reinforcement learning-based plasma flow control of asymmetric vortices over a slender body at high angles of attack

Slender-body aircraft operating at high angles of attack often experience nonlinear, asymmetric multi-vortex flow structures that generate random, unsteady lateral forces, undermining stability, and maneuverability. Dielectric barrier discharge plasma actuators can eliminate these lateral forces. However, conventional open-loop plasma control method cannot adapt to dynamic flow fields in real time, limiting the overall effectiveness of active flow control. This study introduces a plasma control framework grounded in physical principles and develops plasma actuator design methods to regulate vortex interactions, stabilize flow dynamics, and optimize control efficiency. An intelligent closed-loop flow control strategy based on proximal policy optimization, a deep reinforcement learning algorithm, is utilized to enable real-time plasma parameter adjustments for suppressing lateral force at high angle of attack. The spatiotemporal interaction of plasma-induced and asymmetric vortices was investigated through synchronized pressure measurements and particle image velocimetry. The proximal policy optimization-based parameter optimization model was trained online in an educational open-return wind tunnel and subsequently deployed in a low-speed closed-return wind tunnel. Based on vortex stability analysis and comprehensive results, the closed-loop control algorithm significantly mitigates lateral forces, achieving a 68.5% reduction compared to steady plasma actuation, while improving energy efficiency by 70% over conventional methods.