Efficient storage and interpolation of acoustic transfer functions

The acoustic transfer function is a well-known concept for efficient computations of acoustic analysis with various boundary conditions and sound sources. It is first proposed to accelerate the boundary element analysis by interpolating instead of directly assembling influence matrices. To further improve its computational efficiency, a new frequency-interpolation algorithm is proposed by incorporating the proper orthogonal decomposition and a discrete matrix interpolation method. The former is imposed on those pre-computed transfer functions to construct a reduced-order subspace where the later interpolation is implemented, resulting in efficiency gain in both storage and interpolation. In addition to the computational performance, the error bound is also investigated to verify the accuracy via numerical experiments. In this work, the acoustic transfer functions are mainly employed to compute sound scattering, being different from the common applications of computing sound pressure induced by particle velocity at the boundary. Finally, the capability of the proposed method is demonstrated in a drone noise scattering problem.