关键词: 水下滑翔机, 翼身融合, 神经网络, 双层约束多目标优化, 相关性分析

Abstract: To enhance the performance and optimization efficiency of Blended-Wing-Body Underwater Gliders (BWBUGs), this study proposes a dual-layer multi-objective optimization approach for shape optimization of BWBUGs. The proposed approach involves the development of a fully parameterized model, a performance analysis platform, a performance database, and high-precision neural network mapping models. These mapping models are used to establish correlations between design parameters and performance metrics. Furthermore, a dual-layer constrained multi-objective optimization algorithm is constructed to optimize the BWBUG shape, with the lift-to-drag ratio and internal capacity as objectives. The optimization process is conducted using the developed mapping models. The results demonstrate significant performance enhancements, achieving a 31.49% increase in the lift-to-drag ratio and an 11.33% improvement in internal capacity. Additionally, weighted score analyses under varying weight scenarios further validate the effectiveness and robustness of the proposed optimization approach.

Key words: underwater glider, blended-wing-body, neural network, dual-layer constrained multi-objective optimization, correlation analysis