An efficient hierarchical fuzzy simulation method for estimating failure possibility

doi:10.1007/s00366-022-01692-9

2022-08-03

PRE

Xia Jiang

Zhenzhou Lü *

Kaixuan Feng

DOI：10.1007/s00366-022-01692-9

原文链接

原文求助

摘要

En 中文

The failure possibility (FP) can reasonably measure the safety degree of the structure under fuzzy uncertainty, and the estimation of FP can be transformed into searching the point with the maximum joint membership function (MF) of fuzzy input vector in the failure domain (also known as fuzzy design point). In the current fuzzy simulation (FS) method, the fuzzy design point is searched in the maximum value region of the fuzzy input vector corresponding to the lowest membership level which is equal to 0 and the computational efficiency is low. In this paper, an efficient hierarchical fuzzy simulation (HFS) method is proposed for estimating FP. In the proposed method, by the nature that the fuzzy design point is a failure point with the maximum joint MF, the fuzzy design point is first searched in the smaller value region of input vector corresponding to the larger membership level, and the membership level is automatically reduced layer by layer to expand the search region until the failure points appear. Compared with the traditional FS method, the proposed HFS method not only guarantees the search accuracy of the fuzzy design point, but also reduces the total search region; thus the computational efficiency is improved. In addition, an adaptive Kriging model is also embedded in the search process of HFS. Since the adaptively updated Kriging model is used to replace the real performance function for recognizing the state of the simulated sample points during the search process, the strategy of combining the Kriging model with HFS method can further improve the search efficiency of the fuzzy design point. The results of examples show that the proposed HFS method is reasonable and efficient.

Keyword:

Failure possibility

Membership level

Fuzzy design point

Hierarchical fuzzy simulation

Adaptive Kriging model