Simulating pressure coefficients on a circular cylinder at Re=106 by cognitive classifiers

A system formed by multiple fuzzy ARTMAP neural algorithms, operating in parallel, was applied to capture the dynamics of surface pressures from a circular cylinder in cross flow and to forecast the time histories. Pressure data measured at all locations around the section at earlier instants of time were needed to yield accurate temporal predictions of the lower order statistics of pressure, lift and drag. However, using data from only a few adjacent pressure taps at earlier times was not sufficient since pressure in incompressible turbulent flows depends on velocity fluctuations over the entire flow domain.