Learning Wake Regimes from Snapshot Data

Abstract

Fluid wakes are often categorized by visual inspection according to the number and grouping of vortices shed per cycle (e.g., 2S, 2P, P+S). While such categorizations have proven useful for describing and comparing wakes, the criterion excludes features that are essential to a wake's evolution (i.e., the relative positions and strengths of the shed vortices). For example, not all 2P wakes exhibit the same dynamics; thus, the evolution of wake patterns among 2P wakes can be markedly distinct. Here, we explore the notion of labeling wakes according their dynamics based on empirical snapshot data. Snapshots of the velocity field are reduced to a representative feature vector, which is then processed using machine learning techniques tailored to the task of wake regime learning. The wake regime learning framework is evaluated on an idealized 2P wake model, which can be configured to simulate different (known) dynamical regimes. A simple version of the wake regime learning framework successfully discriminates between dynamically distinct 2P wakes. The results presented here suggest that the wake regime learning perspective may facilitate the development of a new dynamics-based wake labeling convention in the future.

Quantitative analysis of the kinematics and induced aerodynamic loading of individual vortices in vortex-dominated flows: a computation and data-driven approach

“…This identification of distinct regimes has similarities to previous studies on the wake of oscillating cylinders [44,45,46,4,47] as well as in biomimetic propulsion [48,49,50]. However, the visual identification employed in these past studies is impractical for more complex flows; a fact that has motivated data-driven approaches to this problem [51,52,53]. However, these prior data-driven efforts have focused on idealized wakes, using point vortices for example, and have mostly assumed a-priori knowledge of the possible wake patterns in order to classify observed wakes into these known categories.…”

Detecting exotic wakes with hydrodynamic sensors

“…The dynamical differences between each of these 2P wakes suggests that there may be much more information to glean from a wake signature simply by considering the wake dynamics; associating a wake signature with a particular wake regime can allow inferences about the wake generating system. Although future work is still needed to establish connections between specific wake dynamics and various swimming characteristics [36], the study here will demonstrate that different wake regimes impart distinct hydrodynamic signatures that can enable wake detection and classification from sensor measurements. For example, a carangiform swimmer (2S wake generator) can be distinguished from an anguilliform swimmer (2P wake generator) from hydrodynamic wake signatures alone.…”

Classifying Exotic Wakes with a Flow Speed Sensor

“…In the event that the wake type is unknown during the library construction stage, then an unsupervised wake regime learning strategy can be devised to group and label wakes according to similarities in their dynamics. 8 Once the library is constructed, we can proceed to classify unknown wakes by comparing against entries in the library. To do so, the measured time-series signal is converted into a static feature vector V test , then compared against entries in the library to determine the "closest match" and the most likely classification of the unknown wake type (see Fig.…”

“…If the wake types were unknown at this point, unsupervised wake regime learning techniques could be used to assign these labels automatically. 8 The short-time Fourier algorithm described in Section II can also be used to extract feature vectors V i , now from datastreams in online applications. Here, we used an overlap percentage of 50%.…”