The efficiency as computational efforts and accuracy as optimum design condition are two major changes in hybrid intelligent optimization methods for hierarchical stiffened shells (HSS). In this current work, a novel hybrid optimization coupled by adaptive modeling framework is proposed to improve the accuracy of the predicted optimum results of load-carrying capacity for HSS by combining active Kriging (AK) and grey wolf optimizer (AK-GWO) for optimization. In the active learning process, two data sets are introduced to train Kriging model, where active points given from initial data and adaptive points simulated based on optimal point using radial samples. The ability for accuracy of AK-GWO optimization method is compared with several soft computing models including Kriging, response surface method and support vector regression combined by GWO. The accurate results are extracted for simulating the load-carrying strength of HSS using the proposed AK-GWO method. The AK-GWO method is enhanced about 10% the accuracy of optimum load-carrying capacity with superior optimum design condition compared to other models, while the load-carrying using AK-GWO is increased about 2% compared the Kriging model.

中文翻译：

---

AK-GWO：一种用于精确优化分层加筋壳的新型混合优化方法

计算工作的效率和最佳设计条件的准确性是分层加筋壳（HSS）混合智能优化方法的两个主要变化。在目前的工作中，通过结合主动克里金法 (AK) 和灰狼优化器 (AK-GWO) 进行优化，提出了一种新的自适应建模框架耦合的混合优化，以提高高速钢承载能力的预测优化结果的准确性. 在主动学习过程中，引入了两个数据集来训练克里金模型，其中主动点由初始数据给出，自适应点基于最优点使用径向样本进行模拟。AK-GWO优化方法与GWO结合的Kriging、响应面法和支持向量回归等几种软计算模型的精度能力进行了比较。提取准确结果用于使用所提出的 AK-GWO 方法模拟高速钢的承载强度。与其他模型相比，AK-GWO 方法在优化设计条件优越的情况下，优化承载能力的精度提高了约 10%，而使用 AK-GWO 的承载能力比 Kriging 模型提高了约 2%。