To achieve efficient active flow control using plasma actuator, flow state classification was performed using a subspace-based approach. Proper Orthogonal Decomposition (POD) modes were extracted from time-series data of surface pressure fluctuations on an airfoil, and classification was performed using Grassmann Discriminant Analysis (GDA) to subspaces spanned by POD modes. The surface pressure fluctuations were measured experimentally using a 28ch microphone array mounted on the airfoil. The POD modes exhibited different spatial structures between the separated and attached flows, and modes similar to those of the attached flow were observed under flow control with the plasma actuator. GDA successfully captured the differences among the subspaces, clearly classified the flow states in the training data, and was able to correctly classify data not included in the training set.
Sasaki et al. (Wed,) studied this question.