A feedback flow control is designed to reduce the aero-optical distortions behind a backward facing step. The delayed-detached eddy simulation, based on the Spalart-Allmaras turbulence model, is developed to simulate the unsteady density variations caused by the shear layer. A database from simulations of both an unforced baseline flow and open-loop forced flows is built using the proper orthogonal decomposition. An Artificial Neural Network and an Auto-Regressive eXogenous model are used to build a reduced order model of the flow field. Closed loop controllers for a suction-blowing actuator were simulated using the reduced order model. Finally, the actuation results from the closed-loop simulation are validated by re-computing the forced flow. The results show that aero-optical distortions can be reduced via the feedback flow controller based on the reduced order model.