Data-Driven and Machine Learning-Based Optimization Design for AUV Shape

This study presents a novel AUV shape optimization method integrating data-driven approaches and machine learning, focusing on head, tail, fin, and rudder parameters. A parameterized hydrodynamic analysis workflow was developed to automatically generate required parameters. A data-driven framework for multi-objective optimization was constructed to minimize drag. Surrogate models were built using Multilayer Perceptron neural networks. The optimization problem was solved using the NSGA-II algorithm. Results showed significant improvements in hydrodynamic performance of optimized AUV shapes, confirming the method's feasibility and effectiveness.