An active learning kriging model for hybrid reliability analysis with both random and interval variables
- 899 Downloads
Hybrid reliability analysis (HRA) with both random and interval variables is investigated in this paper. Firstly, it is figured out that a surrogate model just rightly predicting the sign of performance function can meet the requirement of HRA in accuracy. According to this idea, a methodology based on active learning Kriging (ALK) model named ALK-HRA is proposed. When constructing the Kriging model, the presented method only finely approximates the performance function in the region of interest: the region where the sign tends to be wrongly predicted. Based on the constructed Kriging model, Monte Carlo Simulation (MCS) is carried out to estimate both the lower and upper bounds of failure probability. ALK-HRA is accurate enough with calling the performance function as few times as possible. Four numerical examples and one engineering application are investigated to demonstrate the performance of the proposed method.
KeywordsHybrid reliability Random variable Interval variable Active learning Kriging model
This work is supported by the Aerospace Support Fund (NBXW0001) and Foundation Research Funds of Northwestern Polytechnical University (JCY20130123).
- Du X (2007) Interval reliability analysis. In: Proceedings of the ASME 2007 Design Engineering Technical Conference and Computers and Information in Engineering Conference, Las Vegas, Nevada, USAGoogle Scholar
- Du X (2008) Unified uncertainty analysis by the first order reliability method. J Mech Design (ASME) 130:1401–1410Google Scholar
- Elishakoff I (1999) Are probabilistic and anti-optimization approaches compatible? whys and hows in uncertainty modelling: probability, fuzziness and antioptimization. Springer, New YorkGoogle Scholar
- Gablonsky J (1998) Implementation of the DIRECT Algorithm. Center for Research in Scientific Computation, Technical Rept. CRSC-TR98-29, North Carolina State Univ, Raleigh, NC, AugGoogle Scholar
- Guo S, Lu Z (2002) Hybrid probabilistic and non-probabilistic model of structural reliability. Chinese J Mech Strength 24:524–526Google Scholar
- Lophaven S, Nielsen H, Sondergaard J (2002) DACE, a matlab Kriging toolbox, version 2.0. Tech. Rep. IMM-TR-2002-12; Technical University of Denmark.Google Scholar