A Machine Learning Approach to Classify Vortex Wakes of Energy Harvesting Oscillating Foils

TL;DR

This paper develops a machine learning model combining CNN, LSTM, and autoencoder techniques to classify vortex wake patterns behind oscillating energy-harvesting foils, aiding optimal array design.

Contribution

It introduces a novel hybrid ML approach to categorize wake patterns and links them to foil kinematics for improved tidal energy harvesting.

Findings

Identified four distinct wake patterns via autoencoder and clustering.

Classified wake types with high accuracy using CNN-LSTM model.

Established correlations between wake patterns and foil kinematics.

Abstract

A machine learning model is developed to establish wake patterns behind oscillating foils whose kinematics are within the energy harvesting regime. The role of wake structure is particularly important for array deployments of oscillating foils, since the unsteady wake highly influences performance of downstream foils. This work explores 46 oscillating foil kinematics, with the goal of parameterizing the wake based on the input kinematic variables and grouping vortex wakes through image analysis of vorticity fields. A combination of a convolutional neural network (CNN) with long short-term memory (LSTM) units is developed to classify the wakes into three groups. To fully verify the physical wake differences among foil kinematics, a convolutional autoencoder combined with k-means++ clustering is utilized and four different wake patterns are found. With the classification model, these patterns are associated with a range of foil kinematics. Future work can use these correlations to predict the performance of foils placed in the wake and build optimal foil arrangements for tidal energy harvesting.