Structural Failure Mode Identification And Reliability Analysis Based On Multiple Response Surfaces Method

For complex engineering structures or systems with uncertainties,the identification of dominant failure modes can provide valuable information on realization of safe design and accurate reliability estimation.The traditional failure mode identification and the reliability analysis method of the structural system mainly exist the two shortcomings of the dominate failure mode identification efficiency and the combined failure probability calculation of multi-failure modes.This paper has carried on the corresponding research to this question,and we propose an improved method for searching dominant failure mode.This method combines the multiple response surfaces(MRS)method with the iterative algorithm to obtain the representative sample points effectively.Then,based on the representative sample points,the deterministic mechanical analysis is carried out,and the dominate failure modes corresponding to the sample points are obtained efficiently.In order to verify the efficiency of this method,a comparative analysis was made with the widely used method of the ?-bound method.The main contents are as follows:(1)Based on multiple response surfaces method,the use of Matlab platform and the Python programming language,adopting modular,systematic design thought to write the system reliability calculation program,and for the whole program frame and functionality has carried explanation is introduced.The example analysis is given to verify the validity and the efficiency of the method and the program based on the subsets simulation method,line sampling method and monte carlo method.The results show that the method and the program can reduce the workload and analysis difficulty of the workers to a large extent,thus improving the computational efficiency.(2)By using the iterative technique of multiple response surface method,the convergence is achieved steadily,so the basic sample point set containing the design points are obtained efficiently.Identify the important domain that contribute the most to the whole structure,and the important domain can be divided into multiple sub-domains(SD).The new sample points are added by iteration strategies to improve the coverage density of sample points in important domain so as to obtain the dominant failure modes corresponding to the representative sample points.By using the ?-unzipping method and the proposed method,respectively,simple frame structure and six bar truss structure are used to identify the dominant failure modes,and to get their system failure probability.Compared with the existing literature,the results show that the method has higher identification efficiency.(3)Combined with the mentioned method and the techniques proposed in chapter 3 and chapter 4,the dominant failure modes.are identified with higher dimensional random variables of large span steel arch structures and transmission tower structures,and their system failure probability is obtained.