Zhiyuan Wang

PhD Thesis

AI-Enhanced Flow Sensing for Control of Unsteady Flows.

Supervisors

Andrea Ianiro (UC3M) & Stefano Discetti (UC3M)

Abstract

Efficient flow sensing and control are crucial for enhancing aerodynamic performance and facilitating the development of high-efficiency, sustainable transportation systems. However, the time-varying and strongly nonlinear characteristics of complex unsteady flows present substantial challenges to achieving these goals. Recent advances in artificial intelligence and neuromorphic computing offer new pathways to address these limitations.
The research will focus on the following key aspects. First, for typical unsteady flows such as dynamic stall and gust encounters, efficient flow dimensionality reduction and feature extraction will be achieved through manifold learning and multimodal learning techniques, aiming to uncover the intrinsic physics and low-dimensional representations of unsteady aerodynamics. Besides, under control actions, investigations of actuation manifolds and multi-actuator coordination control strategies will be conducted. Finally, within an event-driven asynchronous framework, new paradigms for event-based sensors and spiking neural network architectures will be explored, paving the way for practically deployable online flow sensing and control systems.