Artificial intelligence control applied to drag reduction of the fluidic pinball
Résumé
The aim of our work is to advance a self-learning, model-free control method to tame complex nonlinear flows-building on the pioneering work of Dracopoulous [1]. The cornerstone is the formulation of the control problem as a function optimization problem. The control law is derived by solving a nonsmooth optimization problem thanks to an artificial intelligence technique, genetic programming (GP). Metaparameters optimization of the algorithm and complexity penalization have been our main contribution and have been tested on a cluster of three equidistant cylinders immersed in a incoming flow, the fluidic pinball. The means of control is the independent rotation of the cylinders. GP derived a control law associated to each cylinder in order to minimize the net drag power and managed to outperform past open-loop studies with a 46.0 % net drag power reduction by combining two strategies from literature. This success of MIMO control including sensor history is promising for exploring even more complex dynamics.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...