Nhảy đến nội dung

Optimization of Load-Carrying Hierarchical Stiffened Shells: Comparative Survey and Applications of Six Hybrid Heuristic Models


Shun-Peng Zhu, Keshtegar Behrooz, Kuo Tian, Nguyen-Thoi Trung

Source title: 
Archives of Computational Methods in Engineering, 28: 4153-4166, 2021 (ISI)
Academic year of acceptance: 

The accurate result of heuristic models combined by social inspired optimization methods is interesting issue for optimizations of hierarchical stiffened shells (HSS). In this paper, six heuristic combined by social-inspired optimization is compared for both ability and accuracy in optimization of load-carrying capacities of HSS. A three level optimization method is employed as (1) explicit dynamic method to provide the train database of optimization model, (2) six heuristic models including response surface method (RSM), multivariate adaptive regression splines (MARS), Kriging, artificial neural network, radial basis function neural network (RBFNN), and support vector regression (SVR) for approximating load-carrying capacity of HSS and (3) an improved partial swarm optimization (IPSO) to search for the optimum results of HSS. In IPSO as optimizer operator, a random adjusting process is presented to update the positions of particles using best particle by a dynamical bandwidth generated by normal standard distribution. Optimization performances for accuracy and ability of six heuristic models coupled by IPSO are compared for optimum model as maximum load-carrying capacity under mass constraint of HSS. The SVR, Kriging and RSM combined by IPSO can be introduced as efficient and accurate modeling-based optimization method to evaluate the optimum design of HSS. The best optimal result is obtained by RBFNN while the worst optimum result is given using MARS among other models.