Analysis of motion performance and motion mode recognition in filament propulsion

Abstract

Abstract In this study, the full free motion of the filament in vortex structures is numerically simulated, and the influence of model parameters and horizontal position on the propulsion efficiency of the filament is investigated. With the “data + physics” approach with the immersed boundary method and Random Forest (RF) method, an RF model is initially employed to screen 189 sets of numerical experiments using a minimal amount of flow field physical data to select experimental groups capable of generating vortex bands. The selected experimental groups are then used to complete numerical experiments, followed by the re-introduction of a minimal amount of data and the use of the RF model to identify the motion modes of the filament. The experimental results indicate that the stretching coefficient of the filament has a negative effect on the thrust coefficient when the filament is in the vortex. The thrust of the filament decreases with increasing horizontal distance. The RF model demonstrates excellent identification capabilities, correctly classifying the vortex street structure with only three sets of data from three test points, achieving a high accuracy rate of 95% in identifying the motion modes of the filament.