Machine learning open-loop control of a mixing layer

Abstract

We develop an open-loop control system using machine learning to destabilize and stabilize the mixing layer. The open-loop control law comprising harmonic functions is explored using the linear genetic programming in a purely data-driven and model-free manner. The best destabilization control law exhibits a square wave with two alternating duty cycles. The forced flow presents a 2.5 times increase in the fluctuation energy undergoing early multiple vortex-pairing. The best stabilization control law tames the mixing layer into pure Kelvin-Helmholtz vortices without following vortex-pairing. The 23% reduction of fluctuation energy is achieved under the dual high-frequency actuations.