军事知识服务平台
多层极限学习机自编码器算法在船型优化中的应用研究
Application of Multi-Layer Extreme Learning Machine Auto-Encoder in Ship Shape Optimization
摘要 由于船型控制参数之间以及优化目标之间存在复杂的耦合关系,船型优化问题的有效降维一直是该领域的难题之一.传统降维方法需要开展大量的数值仿真计算,效率低且降维效果无法保证.论文将多层极限学习机自编码器算法引入到船型优化中的数据降维中,无需进行数值仿真分析即可实现优化变量维度的缩减.将该方法应用到KCS船型总阻力优化中,结果表明该方法在保证优化效果的前提下,显著提升了优化效率.
Abstract:
Owing to the complex coupling relationships among ship hull control parameters and between these parameters and optimization objectives,achieving effective dimensionality reduction for ship shape optimization problems remains a significant challenge in the field.Conventional dimensionality reduction methods necessitate extensive numerical simulation computations,exhibiting low efficiency and unreliable dimensionality reduction outcomes.This study proposes the application of the multi-layer extreme learning Machine Auto-Encoder(M-ELM-AE)algorithm for dimensionality reduction within ship hull optimization,enabling the reduction of optimization variable dimensions without requiring numerical simulation analysis.The proposed method was employed for total resistance optimization of the KCS hull.Results demonstrate that the proposed method significantly enhances optimization efficiency while maintaining the desired optimization performance.
刊 名 中国造船 ISTICEIPKU
年,卷(期): 2026, 67(4)
分类号: U662.2
机标分类号: TP391.41 U661.31+1 U441.5
在线出版日期: 2026年9月23日
基金项目: 52471339:国家自然科学基金U2441289:国家自然科学基金,