Surrogate-Based Optimization for Autonomous Underwater Vehicle´s Shell Design

According to the AUVs´ mission requirement of large operational depth, this paper provides a kind of ring-stiffened multiple intersecting spherical shell (RSMISS) for AUVs. In order to get the performance of the new shell, a surrogate-based optimization (SBO) method is proposed. The SBO method is employed to explore the kriging-based model. Simultaneously, a multi-start optimization approach is repeat-ed to do a global search. Furthermore, an enhanced "Minimizing the Predictor" (EMP) method is presented as the local infill criteria which can dynamically adjust the search direction and the length of sampling radius. Ultimately, the proposed SBO method is applied to look for the agreeable solution of the RSMISS. The results show that EMP can find the satisfactory solution more quickly with smaller computational cost.