Identifying disturbance sources in shear flows using the degenerate unmixing estimation technique

Disturbance source identification is important in laminar to turbulent flow transition and in active flow control applications. A novel method is presented for blind disturbance source separation (BSS) from a limited number of measured mixtures of sources, in which the mixing processes are unknown beforehand. The method is based on an adaptation of the Degenerate Unmixing Estimation Technique (DUET) for use in shear flow applications. DUET can discover any number of sources by using only two sensors, which renders it very effective relative to other BSS techniques. A theoretical framework is developed for separating N_s disturbance sources generated in a boundary layer using two sensors by using DUET. Linear stability theory for Blasius boundary is employed to model the downstream propagation of small disturbances. Numerical and experimental proofs of concept are provided, that demonstrate the power of the new method. Employing two sensors to separate up to three disturbances generated from N_s sources, the numerical simulations include 2D Tollman-Schlichting waves and 2D wave-packet disturbances. Wind tunnel experiments involving boundary layer flow over a flat plate were carried out, in which two hot-wire anemometers were used to separate disturbances generated by two disturbance generators: a single dielectric barrier discharge (SDBD) plasma actuator and a loudspeaker.