期刊
KNOWLEDGE-BASED SYSTEMS
卷 179, 期 -, 页码 77-91出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.knosys.2019.05.005
关键词
Fruit fly optimization; Evolutionary algorithm; Cooperative learning search; High-dimensional and multimodal optimization; Aerodynamic design optimization
资金
- National Key Basic Research Program of China [2015CB755800]
This paper presents a novel improved fruit fly optimization algorithm (IFOA) to effectively solve the aerodynamic design optimization problems, when used in conjunction with computational fluid dynamics (CFD). In IFOA, a new inertia weight function is employed to balance the global exploration and local fine-tuning by adjusting the search scope dynamically. To make this algorithm more universal and effective for continuous optimization problems, especially for the high-dimensional and multimodal complex optimization problems, a new clustering evolution strategy that incorporates the cooperative learning search and crossover operation introduced from real coded GA is developed to enhance the exploration ability and convergence rate of the original FOA. The numerical experiments and comparisons on several benchmark functions are provided, which indicates that the proposed IFOA significantly outperforms the original FOA as well as other competition evolutionary algorithms from literature in terms of comprehensive performance. The IFOA was finally integrated with CFD solver to perform three practical aerodynamic shape design optimization where the objective and constraint functions are highly nonlinear. The inverse design case has been successfully solved to illustrate the algorithm's convergence capability and computational efficiency. The drag minimization problem with constraints on the lift coefficient and geometry is carefully conducted on a transonic airfoil and an isolated wing, and a remarkable drag reduction of 22.79% and 17.41%, respectively, is achieved, confirming the potential of IFOA for aerodynamic design optimization. (C) 2019 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据