摘要
采用Abaqus软件对典型涡轮叶片燕尾榫结构周围的最小接触压力和应力进行有限元分析,并基于Isight软件进行仿真流程自动化和优化设计研究,以期在满足设计要求的前提下,找到一种高效的优化设计方案。首先利用Isight提供的多种算法进行优化设计,并对比不同算法的优化结果和优化运行时间,发现序列二次规划法(NLPQLP)和下山单纯型法(Downhill Simplex)虽然优化时间短,但是优化结果陷入局部最优,多岛遗传算法(MIGA)和非支配排序遗传算法(NSGA-II)虽然优化结果理想,但是优化效率较低。然后,提出了两种改进方案。第一种,首先基于Isight提供的工具Latin Hypercube创建代理模型,再利用代理模型进一步优化。第二种,采用组合优化策略的方式,先采用MIGA或NSGA-II进行少量的全局优化,再以优化结果作为下一步优化的初始方案,使用NLPQLP或Downhill Simplex进行局部优化。最终对多种方案的优化结果和优化效率进行对比和评价。
In this paper,Abaqus software is used to analyze the minimum contact pressure and stress around the dovetail of typical turbine blades,and the simulation process automation and optimization design are studied based on the multidisciplinary optimization design platform Isight,in order to find an efficient optimization design scheme on the premise of meeting the design requirements.At the beginning of the design,a variety of algorithms provided by Isight were used for optimization design,and the optimization results and time of different algorithms were compared.It was found that although the optimization time of Nonlinear Sequential Quadratic Programming(NLPQLP)and Downhill Simplex was short,the optimization results fell into local optimization,and the optimization results of Multi-Island Genetic Algorithm(MIGA)and the Non-dominated Sorting Genetic Algorithm(NSGAII)were ideal,but the optimization efficiency is low.On this basis,two improvement schemes are proposed.First,based on the tools provided by Isight,the Optimal Latin Hypercube is used to collect design samples to create an agent model,and then the agent model is used for further optimization.In the second way,a combination optimization strategy is adopted.First,MIGA or NSGA-II is used for a small amount of global optimization,and then the optimization results are used as the initial scheme for the next optimization.NLPQLP and Downhill Simplex are used for local optimization.Finally,the optimization results and efficiency of various schemes are compared and evaluated.
作者
禹燕飞
张伟
Yan-fei Yu;Wei Zhang(Dassault Systèmes)
出处
《风机技术》
2024年第1期48-54,共7页
Chinese Journal of Turbomachinery
