A novel structural reliability method based on active Kriging and weighted sampling
Autor: | Qisong Qi, Wenzhao Li, Guangli Zhao, Qing Dong, Ruigang Yang |
---|---|
Rok vydání: | 2021 |
Předmět: |
0209 industrial biotechnology
Mathematical optimization Active learning (machine learning) Computer science Mechanical Engineering Design of experiments Sampling (statistics) Probability density function Sample (statistics) 02 engineering and technology 020303 mechanical engineering & transports 020901 industrial engineering & automation 0203 mechanical engineering Mechanics of Materials Kriging Limit state design Reliability (statistics) |
Zdroj: | Journal of Mechanical Science and Technology. 35:2459-2469 |
ISSN: | 1976-3824 1738-494X |
Popis: | With complex limit state functions and small failure probability, the analysis of engineering structure reliability is a challenging problem. Most of the traditional methods require a lot of calculation time, which results in delaying the progress of solving engineering. To overcome this issue, a new structural reliability analysis method aiming to select a set of better initial design of experiment (DoE) is proposed in this study. The proposed method combines a weighted sampling based on sample probability and a novel selection strategy to select DoE and conduct active learning. Weighted sampling based on sample probability density makes the DoE uniformly distributed in the sampling space. The novel selection strategy is proposed to make selected DoE near limit state surface (LSS) and has better predictive ability. Numerical examples and engineering examples show that this method can perform energy efficiency analysis in important areas. The results show that the method is accurate and efficient in solving low failure probability and nonlinearity problems. |
Databáze: | OpenAIRE |
Externí odkaz: |