An efficient Bayesian inversion method for seepage parameters using a data-driven error model and an ensemble of surrogates considering the interactions between prediction performance indicators
- Resource Type
- ACADEMIC JOURNAL
- Authors
- Yu, Hongling; Wang, Xiaoling ⁎; Ren, Bingyu; Zeng, Tuocheng; Lv, Mingming; Wang, Cheng
- Source
- In Journal of Hydrology January 2022 604
- Subject
- Language
- English
- ISSN
- 0022-1694
- E-ISSN
- DOI
- 10.1016/j.jhydrol.2021.127235
Highlights •A data-driven error model based on Gaussian process regression was integrated into Bayesian inference to address the model structure error.•A novel ensemble of surrogates was proposed to improve the computational efficiency of Bayesian inversion of seepage parameters.•The determination of the weights of the surrogate models has considered the interactions between prediction performance indicators.•The method was validated through a real engineering and showed a significant improvement.