Speaker: Şeyma Satıcı

Abstract: This seminar presents the development of an intelligent plasma-activated water (PAW) production system that combines a dielectric barrier discharge (DBD) plasma reactor, real-time sensing, experimental data, and reinforcement learning (RL). PAW production is a nonlinear and time-dependent process in which operating parameters such as plasma power and frequency strongly influence water chemistry. Recent experiments achieved nitrate concentrations above 1,000 ppm, while also revealing trade-offs such as increased acidity at high nitrate levels. The project therefore explores how RL can learn dynamic control policies to optimize PAW production under multiple objectives, providing a practical case study of applying data-driven control to a complex real-world physical system.