罗文俊, 王德禹. 基于兴趣子域动态代理模型的船舶结构可靠性优化[J]. 中国舰船研究, 2021, 16(4): 96–107. doi: 10.19693/j.issn.1673-3185.02043
引用本文: 罗文俊, 王德禹. 基于兴趣子域动态代理模型的船舶结构可靠性优化[J]. 中国舰船研究, 2021, 16(4): 96–107. doi: 10.19693/j.issn.1673-3185.02043
LUO W J, WANG D Y. Reliability-based optimization of ship structure based on interest subdomain dynamic surrogate model[J]. Chinese Journal of Ship Research, 2021, 16(4): 96–107. doi: 10.19693/j.issn.1673-3185.02043
Citation: LUO W J, WANG D Y. Reliability-based optimization of ship structure based on interest subdomain dynamic surrogate model[J]. Chinese Journal of Ship Research, 2021, 16(4): 96–107. doi: 10.19693/j.issn.1673-3185.02043

基于兴趣子域动态代理模型的船舶结构可靠性优化

Reliability-based optimization of ship structure based on interest subdomain dynamic surrogate model

  • 摘要:
      目的  针对常规船舶结构可靠性优化设计中因高度非线性导致难以同时保证代理模型拟合精度和优化效率的问题,提出基于兴趣子域动态代理模型的可靠性优化方法。
      方法  运用序贯优化与可靠性评估法(SORA)的兴趣子域概念,确定兴趣子域的范围,基于信息熵函数H制定自适应空间减缩规则,以缩减设计空间,进而构造基于兴趣子域的自适应空间减缩序贯抽样策略,从而采用尽可能少的样本点构造对兴趣子域进行局部高度拟合的Kriging动态代理模型, 并将代理模型和多岛遗传算法(MIGA)嵌入SORA中进行可靠性优化。提出概率约束可行性检查方法,以减少不必要的可靠性评估过程。同时,通过数学算例对该可靠性优化方法进行验证。
      结果  结果显示,最优解与理论解的相对误差为0.066 8%,与文献中的最优方法相比减少了40.6%的功能函数调用次数,证明了本方法的高精度和高效性。
      结论  将所提方法应用于舱段结构可靠性优化设计中,舱段总质量与文献结果相比减少了0.511%,且有限元计算调用次数减少了94次,证明了所提方法的高效性和适用性。

     

    Abstract:
      Objectives  Aiming at the difficulty of ensuring the fitting accuracy and optimization efficiency of surrogate models due to the high nonlinearity in ship structure reliability-based optimization design, a reliability-based ship structure optimization method based on the interest subdomain dynamic surrogate model is proposed.
      Methods  This method puts forward the concept of the interest subdomain based on the sequential optimization and reliability assessment (SORA) method, determines the range of interest subdomains and formulates adaptive spatial reduction rules based on information entropy function H, then proposes an adaptive spatial reduction sequential sampling strategy based on interest subdomains, thereby constructing a dynamic Kriging surrogate model that highly fits the subdomain of interest locally with as few sample points as possible, and embedding the surrogate model and multi-island genetic algorithm (MIGA) in the SORA method to undertake reliability-based optimization.This study proposes a probability constraint feasibility checking method to reduce unnecessary reliability assessment processes. A mathematical example is given to verify the reliability-based optimization method.
      Results  The relative error between the optimal solution and theoretical solution is 0.066 8%, and the number of function calls is 40.6% less than those of the optimal method in the references, which proves the accuracy and efficiency of this method.
      Conclusions  When the proposed method is applied to the reliability optimization design of a cabin structure, the total cabin mass is reduced by 0.511% compared with the references, and 94 fewer finite element calculations are required, proving the efficiency and applicability of this method.

     

/

返回文章
返回