Reliability Analysis And Optimization Design Based On Support Vector Machine And Grasshopper Parameterized Platform

The structural reliability is an important measurement of the quality and safety of the engineering structure.Therefore,reliability theories,methods and applications are important topics in the scientific research and engineering.Structural reliability analysis and reliabilitybased optimization design are two key parts that have a decisive influence on the safety and service life of the structure.At present,the reliability analysis method for simple explicit performance function is relatively mature.However,the performance function of the actual engineering structure is usually implicit and non-linear.For this case,non-convergence may occur in the calculation using the traditional reliability analysis method.To solve this problem,surrogate model can be introduced to replace the actual performance function for reliability analysis.At the same time,with the rapid development of computer technology,a series of parametric modeling software such as Grasshopper have been widely used in the field of structural design.The combination of parametric modeling technology and reliability method can significantly improve the efficiency of the reliability analysis of actual engineering structures.Therefore,the main contents in this thesis are as follows:(1)The basic theory and calculation method of structural reliability are introduced,including the concepts of structural reliability,limit state function,failure probability and reliability index;traditional reliability calculation methods such as first-order reliability method,Monte Carlo simulation method and reliability analysis based on surrogate model method.The support vector machine method is a machine learning method based on mathematical statistics theory,which has good learning and prediction ability for small sample size problems.So this thesis uses support vector machine as the basis for the study.(2)A surrogate-model reliability analysis method is proposed.Based on the existing adaptive support vector machine method,the searching process is improved by using a new adding-point criterion.This modified adaptive support vector machine is combined with the Monte Carlo simulation method to get a new reliability analysis method.Some numerical examples verify the accuracy and efficiency of the proposed method.(3)The proposed reliability analysis method and the commercial finite element software SAP2000 are integrated in the parametric modeling software Grasshopper to perform reliability analysis of the engineering structure.Firstly,based on the customized re-development function of Grasshopper,the SAP2000 OAPI tool is used to connect Grasshopper with SAP2000 to realize the interaction between the structural parameterized model and the finite element software.Secondly,integrate the proposed reliability analysis method in the form of a customized plug-in.The reliability analysis of several actual structures vetify the correctness and effectiveness of the developed reliability analysis platform.(4)The sequential approximate programming strategy is integrated in Grasshopper to perform the reliability-based design optimization of engineering structures.Based on the reliability analysis and reliability-based design optimization platform developed in Grasshopper,the parametric modeling technology can be used to construct the structural geometric model quickly and easily.And the customized plug-in can be used to input structural stochastic parameter information and add reliability analysis methods.Several structural examples show that the platform has good engineering application prospects.