Multi-fidelity shape optimization design method for autonomous underwater vehicle

• Main Authors: YAN WEISHENG, SONG BAOWEI, WANG PENG, JIANG JUN, SHI YAO, DU XIAOXU
• Format: Patent
• Language: Chinese; English
• Published: 24.07.2018
• Subjects: CALCULATING; CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TOTRANSPORTATION; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; PHYSICS; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS; TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINSTCLIMATE CHANGE

The invention provides a multi-fidelity shape optimization design method for an autonomous underwater vehicle. The multi-fidelity shape optimization design method is characterized by including the steps: performing parametric three-dimensional modeling on the autonomous underwater vehicle, initializing parameter space, determining a parameter range and selecting parameters by taking minimum zero-lift drag of the autonomous underwater vehicle as an optimization objective; performing partial research by a minimum prediction method and limiting optimizing step length by a trust region method. Therefore, design space can be efficiently searched and optimized without gradient information at a sample point. A multi-fidelity model is managed by an improved trust region model management method, and a global search strategy is increased. 本发明提供了种自主水下航行器的外形多保真度优化设计方法，其特征在于，包括以下步骤：对自主水下航行器进行参数化三维建模，初始化参数空间，确定参数范围，其中参数的选择以自主水下航行器零升阻力最小为优化目标；利用最小预测方法进行局部探索，同时用信赖域方法来限制寻优的步长，这样可以在没有样本点处梯度信息的情况下对设计空间进行高效地探索和寻优。利用改进的信赖域模型管理方法对