Predictive Modeling of Wavefront Error Using Dynamic Mode Decomposition

In an airborne directed energy system, air density variations around an aircraft distort a beam's wavefront, resulting in degraded performance after propagation. Adaptive Optics (AO) can be used to correct for these aero-optical aberrations in the wavefront to give increased performance. However, in some conditions state-of-the-art AO systems have insufficient spatial and temporal bandwidth to accurately maintain high system performance. Predictive AO control is a promising method to mitigate these problems and therefore improve beam quality. This manuscript builds on previous research with novel Digital Holographic (DH) Wavefront Sensor (WFS) data measuring aero-optical disturbances from the wind tunnel at the Air Force Research Laboratory (AFRL) Aero-Effects Laboratory (AEL) to present a Dynamic Mode Decomposition (DMD) based predictive modeling tool. DMD is a light-weight, equationfree, dimensionality reduction algorithm that can be used to isolate spatio-temporal patterns in a data set into physically meaningful modes. These can be used to predict future states from a given wavefront. This manuscript demonstrates application of this method to data acquired through AFRL as well as discussing the underlying mechanisms.