Regular Article - Flowing Matter
Machine learning for optimal flow control in an axial compressor
Univ. Lille, CNRS, ONERA, Arts et Métiers Institute of Technology, Centrale Lille, UMR 9014, LMFL - Laboratoire de Mécanique des Fluides de Lille - Kampé de Fériet, 59000, Lille, France
2 Arts et Métiers Institute of Technology, CNAM, DynFluid, HESAM Université, 75013, Paris, France
Accepted: 19 March 2023
Published online: 12 April 2023
Air jets for active flow control have proved effective in postponing the onset of stall phenomenon in axial compressors. In this paper, we use a combination of machine learning and genetic algorithm to explore the optimal parameters of air jets to control rotating stall in the axial compressor CME2. Three control parameters are investigated: the absolute injection angle, the number of injector pairs and the injection velocity. Given an experimental dataset, the influence of the air jet parameters on the surge margin improvement and power balance is modeled using two shallow neural networks. Parameters of the air jets are then optimized using a genetic algorithm for three rotational velocities, i.e., . First, surge margin improvement and power balance are being maximized independently. Then, a bi-objective optimization problem is posed to explore the trade-off between the two competing objectives. Based on the Pareto front, results suggest that a globally optimal set of parameters is obtained for a velocity ratio (defined as the ratio of the injection velocity to the rotor tip speed) ranging from 1.1 to 1.6 and an injection angle attack varying from . These outcomes point out a potential generalization of the control strategy applicable to other compressors.
© The Author(s), under exclusive licence to EDP Sciences, SIF and Springer-Verlag GmbH Germany, part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.