Research On The Application And Structural Optimization Design Method Based On DOE And Surrogate Model

The structural optimization design is to obtain the best design in accordance with intended target under the given constraints,its application fields are expanded from the original aerospace to the shipping,machinery,automobile and construction,etc.,solving the problems such as reducing the structural weight,reducing the stress level and promoting the service life of structure and reliability.This paper has studied based on the structure optimizing design method and application of DOE and agent model method,and carried out optimization analysis for the flow field of building groups in complex areas like complex super high-rise buildings and chamfering buildings by the DOE and the agent model method;and also had MDO analysis for super high-rise buildings under unidirectional flow soil-fluid coupling effect by adopting the DOE,the agent model method and Monte Carlo Simulation(MCS)method.A method for calculating the fluid solid coupling optimization of buildings is proposed.The main work and conclusions of this paper are as follows:(1)Conducting analysis study on 8 kinds of optimization algorithms in DOE and 4 kinds of surrogate model methods and also carry out optimization design of single subject and single objective of the strength of rectangular,diamond and oval pin-fin models by Latin super cube method in DOE.Through optimization design,the maximum stress of rectangular pin-fin was reduced by 13.497%,diamond pin-fin by 18.091%,and the oval pin-fin by 16.741%.The paper carries out multidisciplinary optimization analysis on rectangular and diamond pin-fin models by adopting the response surface surrogate model of the surrogate model type(The adopted optimization methods include MIGA,NCGA and AMGA).Thus we obtain the iMPact degree of each variable on the model quality and stress.The final optimization results are as below: When using MIGA algorithm,the quality of rectangular pin-fin model reduced by 15.423%,and the quality of diamond pin-fin model reduced by 22.59%;When adopting AMGA algorithm,the quality of rectangular pin-fin model reduced by 19.417%,and the quality of diamond pin-fin model reduced by 24.36%;When applying NCGA algorithm,the quality of rectangular pin-fin model reduced by 16.027%,and the quality of diamond pin-fin model reduced by 21.91%.(2)The DOE method is introduced into the optimization of buildings.Taking planar dimensions of building models as the constraint condition,and the minimum negative pressure in buildings as the optimization goal,using 8 optimization algorithms in DOE(including parameter test,full factorial design,fractional factorial design,orthogonal array,central group and design,Box-Behnken,Latin hypercube design and optimal Latin hypercube design)for optimization analysis of single objective and subject on complex building model.Meanwhile,by coMParison of optimization results,we found that Latin hypercube method was used to optimize the building flow field.(3)The surrogate models method was introduced into the optimization of buildings.Establishing 4 kinds of surrogate models of optimization analysis of chamfer building(Kriging model,neural network model,response surface model and orthogonal polynomial model).And by error coMParison,we selected Kriging surrogate model with rather small error for optimization analysis on chamfer building and finally enhanced the minimum negative pressure for chamfering building by 50.69%.After constructing Kriging surrogate model of chamfer building,we conducted optimization analysis based on Six Sigma design method(Design For Six Sigma,DFSS)and KM model on the two building models.Finally,we greatly enhanced the reliability of optimization objectives and validated the feasibility of DFSS in building optimization.(4)The paper proposed a new research method for unidirectional fluid solid coupling of buildings.The method used ANSYS CFX as fluid calculation module and ABAQUS as the solid computing module.By adopting inverse distance weighted average method(DistanceWeighted Average,DWA),it realized data transmission on coupling surface whose process is realized by Fortran program.Meanwhile,co MParing the result of ANSYS Workbench computing validated the feasibility of the method.Then adopted kriging surrogate model technique and Monte Carlo simulation(Monte Carlo Simulation,abbreviated as MCS)method for MDO analysis on the buildings.Changes of final optimization objectives are as following: While adopting DOE+MIGA combinatorial optimization strategy,the stress reduced by 25.411%;While adopting DOE+AMGA combinatorial optimization strategy,the stress reduced by 24.374%;When adopting MIGA+NLPQL combinatorial optimization strategy,the stress reduced by 23.826%;When adopting MCS method alone,the stress reduced by 25.51%.(5)Emphasizing on the application of surrogate model in flow field optimization of complex buildings,the paper put forward two optimization strategies.The first optimization strategy is to respectively construct KM model for a single building in the building group to conduct optimization analysis based on surrogate model so as to define building form;Then use Latin super cube method of DOE for optimization study on the distance between buildings;The second optimization strategy is to first use Latin super cube method of DOE for optimization analysis on building groups;Then choosing the main factors affecting optimization goals as the input variables to construct KM model.After analyzing,it is found that the two optimization strategies got better results,and the accuracy of agent model was improved greatly.