Optimizing Sensor Placement in Turbulent Flows: A Correlation-Aware Attribution Framework

Optimal sensor placement (OSP) is critical for efficient, accurate monitoring, control, and inference in complex real-world systems such as turbulence. We propose a feature attribution framework for OSP that considers the correlated nature of turbulent flow. Feature attribution quantifies the contribution of input variables to a model's output, making it useful for OSP, but it often struggles in the presence of highly correlated data. To address this, the Correlation-Aware Attribution Framework (CAAF) introduces a clustering step before attribution to reduce redundancy and improve generalizability. We first illustrate the core principles of the proposed framework through a series of validation cases, then demonstrate its effectiveness in airfoil lift prediction and quantity estimation in turbulent channel flows, where traditional OSP methods struggle due to nonlinear dynamics, chaotic behavior, or multi-scale interactions. The results show that the CAAF outperforms alternative approaches in sensor placement and enables the effective application of feature attribution for identifying OSP in real-world environments.