Please wait a minute...
工程设计学报
Chinese Journal of Engineering Design  2009, Vol. 16 Issue (4): 266-270    DOI:
    
Research on improved EGO algorithm based on Kriging surrogate model
 WANG  Hong-Tao, ZHU  Xiao-Cheng, DU  Chao-Hui
School of Mechanical and Power Engineering, Shanghai Jiaotong University, Shanghai 200240, China
Download: HTML     PDF(426KB)
Export: BibTeX | EndNote (RIS)      

Abstract  Surrogate model is a key technology for complex engineering optimization design. EGO algorithm, a Bayesian analysis optimization algorithm based on Kriging surrogate model, makes use of the EI function to select the next sampling point to ensure the global convergence. An improved EGO algorithm was proposed due to the shortcomings of original algorithm. Four representative numerical and one engineering examples were selected to test the improved EGO algorithm's performance. Finally, comparison of the improved EGO and other algorithms was conducted from the aspect of convergence efficiency and accuracy. The results show that the improved EGO algorithm is more efficient than the original one when the optimization process reaches the same accuracy and superior to response surface optimization algorithm in accuracy and converging efficiency. The method has strong adaptability and good engineering practical value．

Key words： improved EGO algorithm      global optimization      Kriging model      experimental design method     
Published: 28 August 2009
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
WANG Hong-Tao
ZHU Xiao-Cheng
DU Chao-Hui
Cite this article:

WANG Hong-Tao, ZHU Xiao-Cheng, DU Chao-Hui. Research on improved EGO algorithm based on Kriging surrogate model. Chinese Journal of Engineering Design, 2009, 16(4): 266-270.

URL:

https://www.zjujournals.com/gcsjxb/     OR     https://www.zjujournals.com/gcsjxb/Y2009/V16/I4/266


基于Kriging代理模型的改进EGO算法研究

代理模型是复杂工程优化设计问题的关键技术之一．基于Kriging代理模型的EGO算法作为一种贝叶斯全局优化算法引入了EI函数来确定校正点,保证了算法的全局收敛性．首先针对原始EGO算法的不足之处,提出改进EGO算法．然后采用改进EGO算法对4个经典函数和1个工程算例进行测试,最后从算法的收敛速度和精度两方面将不同的算法进行比较．结果表明改进后的EGO算法达到原始EGO算法精度时所需迭代步数更少,与基于响应面的优化算法相比在收敛速度和精度方面更具有优势．说明该方法适应性强,具有很高的工程实用价值．

关键词： 改进EGO算法,  全局优化,  Kriging模型,  试验设计方法 
[1] YANG Yang, SHU Le-shi. Design optimization of micro-aerial vehicle fuselage structure based on sequential hierarchical Kriging model[J]. Chinese Journal of Engineering Design, 2018, 25(4): 434-440.
[2] ZHU Wei, LAI Xiong-Ming. Reliability sensitivity analysis for multi-parameters influential mechanism[J]. Chinese Journal of Engineering Design, 2011, 18(2): 124-129.
[3] MA Xiao-Li, MA Lü-Zhong, ZHOU Zhao-Zhong. Study and application of orthogonal-genetic algorithm test method[J]. Chinese Journal of Engineering Design, 2005, 12(6): 334-337.