摘要
Due to the good balance between high efficiency and accuracy, meta-model based optimization algorithm is an important global optimization category and has been widely applied. To better solve the highly nonlinear and computation intensive en- gineering optimization problems, an enhanced hybrid and adaptive meta-model based global optimization （E-HAM） is first proposed in this work. Important region update method （IRU） and different sampling size strategies are proposed in the opti- mization method to enhance the performance. By applying self-moving and scaling strategy, the important region will be up- dated adaptively according to the search results to improve the resulting precision and convergence rate. Rough sampling strategy and intensive sampling strategy are applied at different stages of the optimization to improve the search efficiently and avoid results prematurely gathering in a small design space. The effectiveness of the new optimization algorithm is verified by comparing to six optimization methods with different variables bench mark optimization problems. The E-HAM optimization method is then applied to optimize the design parameters of the practical negative Poisson＇s ratio （NPR） crash box in this work. The results indicate that the proposed E-HAM has high accuracy and efficiency in optimizing the computation intensive prob- lems and can be widely used in engineering industry.
基金
supported by the Research Project of State Key Laboratory of Mechanical System and Vibration(Grant Nos.MSV201507&MSV201606)
the National Natural Science Foundation of China(Grant No.51375007)
the Natural Science Foundation of Jiangsu Province(Grant No.SBK2015022352)
the Fundamental Research Funds for the Central Universities(Grant No.NE2016002)
the Open Fund Program of the State Key Laboratory of Vehicle Lightweight Design,P.R.China(Grant No.20130303)
the Visiting Scholar Foundation of the State Key Lab of Mechanical Transmission in Chongqing University(Grant Nos.SKLMT-KFKT-2014010&SKLMT-KFKT-201507)
