AK-GWO: a novel hybrid optimization method for accurate optimum hierarchical stiffened shells

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.