Twelve POD modes account for ~80% of the aberration of energy.
Using the predictive method allows us to make use of previous eigenvalues of POD modes to predict future eigenvalves at some latency time.
To demonstrate this process, we used a neural network as a prediction method on the AAOL wavefront data at a viewing angle of 120 degrees, at a WFS frame rate of 25 kHz.
The neural network approach was able to predict rapid changes in the eigenvalues with a 200 microsecond latency using only four previous eigenvalue windows.
We have conducted a number of high-fidelity simulation activities at a variety of elevation angles and flow fields.