Optimization strategy for bio-inspired lateral line sensor arrays in autonomous underwater helicopter

Conventional optical and acoustic sensing methods exhibit significant limitations in extreme environments, hindering the autonomous sensing capabilities of Autonomous Underwater Helicopters (AUH) during complex operations. While prior research has concentrated on sensor array optimization for bio-inspired underwater vehicles, such as airfoil-based robotic fish, the optimization of sensor layouts for AUH lateral line systems remains underexplored. This study introduces an optimized sensor placement strategy for AUHs by integrating CFD with advanced feature selection techniques to enhance sensing performance in complex flow fields. The importance of features was evaluated using Maximal Relevance Minimal Redundancy (mRMR), the Pearson Correlation Coefficient (PCC), and the Feature Distance-Based Method (FBDM), facilitating the determination of optimal sensor candidate weights. These features were applied to classification algorithms, including Random Forest (RF), Support Vector Machine (SVM), and k-Nearest Neighbors (KNN), to assess the impact of sensor quantity and layout on classification accuracy. The results demonstrate that the RF algorithm, in conjunction with FBDM, achieved the optimal sensor configuration while minimizing the number of sensors required, enhancing classification accuracy. These findings provide a theoretical foundation for optimizing AUH lateral line systems and establish a framework for improving autonomous sensing capabilities in extreme underwater environments, presenting strong potential for real-world applications.