Acoustic localization of incipient cavitation in marine propeller using greedy-type compressive sensing

Abstract

A precise evaluation of propeller cavitation inception is imperative in prediction of ship's underwater radiated noise performance. In addition to a simple indication on whether the cavitation has occurred or not, it would be much more valuable if the location of cavitation occurrence can be identified for a further confirmation. Hence, this work concentrates on establishing an efficient acoustic localization for incipient cavitation using the compressive sensing (CS) theory. Rather than the convex minimization approach, the greedy-type orthogonal matching pursuit (OMP) algorithm is chosen for solving the CS localization problem due to a low computational complexity making it suitable for online applications. It is also proposed to employ the singular value decomposition (SVD) - based rank estimation for a provision of user input parameter that is essential for initializing the greedy algorithm. The suggested localization procedure is validated through the propeller cavitation data obtained in a water tunnel facility.