Flow mechanism of a vibrating prism using the combined K-nearest neighbor and dynamic mode decomposition method

Abstract

This study proposes a novel hybrid method, K-nearest neighbor and dynamic mode decomposition (KNN-DMD), for capturing modes of vibrating prisms in the flow field. To validate the effectiveness of the proposed KNN-DMD, the velocity field under structural vibration amplitudes with 0%, 10% and 18% is obtained using large eddy simulation. KNN-DMD is utilized to identify dominant features of the flow field around the prism, such as low-frequency mode, the primary vortex shedding mode, second harmonic mode, and forced vibration mode. The low-frequency mode is associated with fluctuations in the incoming turbulent flow. The primary vortex shedding manifests as the periodic shedding of vortices in the wake region, revealing the dynamics of the shear layer and the process of shedding in the Kármán vortex street. The second harmonic mode, which is a higher-order form of the Kármán vortex street, supplements smaller-scale vortex structures and enriches the vortex characteristics. The forced vibration mode, reflecting the range and intensity of the vibration effect, forms shear layers that propagate to both sides and exhibit an alternate shedding phenomenon. The KNN-DMD reveals the structural modal forms in fluid dynamics more comprehensively and facilitates future research on the fluid-solid interaction and nonlinear stochastic systems.