Optimization Design Of Structural Reliability Based On Fuzzy Intelligent Multiple Response Surface Method

In order to improve the computational accuracy and computational efficiency of the overall reliability analysis and optimization design with multicomponent and multi-failure mode structure,considering the fuzziness and randomness of the failure criterion and boundary load condition.By combination of the fuzzy reliability theory,support vector machine theory and response surface theory,a reliability analysis method of Fuzzy Intelligent Multi-Response Surface Method(FIMRSM)and Fuzzy Intelligent Multi-Extremum Response Surface Method(FIMERSM)with multi-failure modes was proposed,respectively.On the basis of the fuzzy reliability analysis,the multi-objective intelligent optimization algorithm and decomposition coordination theory are combined to propose the method of reliability optimization with multi-failure modes,which is Multi-object Genetic Algorithm-Fuzzy Intelligent MultiResponse Surface Method(MOGA-FIMRSM)and Fuzzy Intelligent Distributed Collaborative Extremum Response Surface Method(FIDCERSM).The effectiveness of the method which is proposed in this paper is verified by the rotor system components of the aeroengine.(1)To the problem of reliability analysis with multi-physics field structures,based on the method of Multi-Response Surface Method(MRSM)and Fuzzy VLinear Programming Support Vector Regression(FV-LPSVR),a method of Fuzzy Intelligent Multi-Response Surface Method(FIMRSM)was proposed and the fuzzy reliability analysis of aeroengine blisk was completed.Firstly,the distribution of the maximum deformation,maximum stress and maximum strain of the blisk were obtained by the numerical simulation of flow-heat-solid coupling.Secondly,the inlet velocity,rotor speed,temperature and density as input variables.The output response of blisk deformation,stress and strain failure mode was calculated by using Latin hypercube sampling technique.The Fuzzy Intelligent Multi-Response Surface Function(FIMRSF)was constructed by using the model of FV-LPSVR.Finally,the Monte Carlo method(MCM)is used to large amount linkage sampling of the model of FIMRSF and solving the model of fuzzy reliability probability.The comprehensive reliability of blisk is 99.29%.(2)To the problem of dynamic reliability analysis of complex mechanical structures,a method of Fuzzy Intelligent Multi-Extremum Response Surface Method(FIMERSM)was proposed by combination of the Fuzzy Support Vector Regression(FSVR)and Multi-Extremum Response Surface Method(MERSM).The fuzzy dynamic reliability analysis of aeroengine blade was carried out.Firstly,the maximum stress point,the maximum strain point and the minimum life point of the blade are founded by the thermo-structural coupling deterministic analysis.Then,the extremum output response of the blade stress,strain and low cycle fatigue life within in the analysis time domain was obtained by solving the basic equations of the finite element for each set of the input random variables.Finally,the Monte Carlo method(MCM)is used to large amount linkage sampling of the model of FIMERSF and solving the model of generalized fuzzy stochastic reliability probability.The dynamic reliability of blade is 99.46%.(3)To the problem of the reliability optimization design with single component,a method of Multi-Object Genetic Algorithm-Fuzzy Intelligent Multi-Response Surface Method(MOGA-FIMRSM)was proposed by integrating the multi-response surface method and the multi-objective genetic algorithm.The fuzzy reliability-based design optimization(FRBDO)of the aeroengine blade was carried out.Firstly,the Fuzzy Intelligent Multiple Response Surface Function(FIMRSF)was constructed by using the Fuzzy Least Squares Support Vector Regression(FLSSVR).The reliability sensitivity analysis of blade was completed.Secondly,the model of FRBDO is constructed by using the high sensitivity index as the design variables.The model of FRBDO is solved by the NAGA-? multi-objective genetic algorithm,and the reliability optimization design of blade was completed.Finally,through the comparison method of MCM shows that MOGA-FIMRSM improves the index of the blade reliability by 3.45%,and the calculation accuracy is consistent with that of MCM.(4)To the problem of the reliability-based design optimization with overall multi-component,a method of Fuzzy Intelligent Distributed Collaborative Extremum Response Surface Method(FIDCERSM)was proposed by combination of the Fuzzy V-Support Vector Regression(FV-SVR)and the Distributed Collaborative Extremum Response Surface Method(DCERSM).The fuzzy decomposition coordination reliability-based design optimization(FDCRBDO)of aeroengine blade-disk was carried out.Firstly,the model of FVSVR is used to construct the Fuzzy Intelligent Decomposition Collaborative Extremum Response Surface Function(FIDCERSF).Then,the sensitivity index of the input random variables was obtained by using the MCM dynamic reliability analysis of the FIDCERSF.Finally,the model of FDCRBDO is established based on the high sensitivity degree as the design variables,the reliability and fatigue-creep damage of GH4133 B superalloy as the constraint function,and the FIDCERSF is used to instead of the limit state function of the blade-disk.The fatigue-creep interaction reliability-based design optimization of blade-disk was completed by the decoupling coordination between the various levels of planning.