Explainable deep reinforcement learning reveals energy-efficient control strategies for turbulent drag reduction
Researchers have developed a novel method combining Multi-Agent Deep Reinforcement Learning (MARL) and eXplainable Deep Learning (XDL) to significantly reduce drag in turbulent flows. This approach utilizes SHAP (SHapley Additive exPlanations) to guide the learning process, focusing on predictions of skin-friction coefficient and wall pressure fluctuations. The resulting strategy achieves a substantial 34.44% drag reduction with minimal energy expenditure, outperforming existing methods by a significant margin. AI
IMPACT This research demonstrates how explainable AI can optimize complex physical systems, potentially leading to more efficient energy use in transportation and industrial applications.