Real-time event-based particle image velocimetry for active flow control

This work investigates event-based vision (EBV) as a tool for real-time flow diagnostics in configurations analogous to two-dimensional, two-component particle image velocimetry. Owing to its reduced data stream compared to conventional frame-based imaging, EBV enables kilohertz-rate pseudo-framing and efficient processing on standard computing hardware. A pseudo-frame-based implementation called real-time event-based imaging velocimetry is presented, capable of delivering velocity fields at several hundred hertz with O(10^6) vectors per second. The concept is experimentally demonstrated on a small-scale jet in water, where event rates above 100×10^6 events/s and online processing at 250−700 Hz are achieved depending on seeding and interrogation settings. Beyond these validation cases, two active flow control applications on a jet in air are illustrated: open-loop optimization of jet mixing using Bayesian optimization, and closed-loop control of a water jet using reinforcement learning. These results highlight EBV as a cost-effective and scalable sensing technology with strong potential for real-time feedback in flow control.