Efficient uncertainty quantification for dynamic subsurface flow with surrogate by Theory-guided Neural Network

Wang, Nanzhe; Chang, Hong; Zhang, Dongxiao

doi:10.1016/j.cma.2020.113492

Cited by 55 publications

(23 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the TgNN surrogate-based methods, the TgNN surrogate is trained by matching the available simulation data and honoring physical/engineering principles at selected collocation points. The TgNN surrogate can be constructed with limited labeled data, or even in a label-free manner, which is an advantage of the TgNN surrogate as studied in Wang et al (2021). Although the training process of the TgNN surrogate may require some computational cost, the trained surrogate accelerates the inversion procedure significantly and can be used for solving new cases, such as varying prior statistics, and different spatial and temporal observations.…”

Section: Discussionmentioning

confidence: 99%

“…Furthermore, the effect of the number of labeled data and collocation points has been studied in Wang et al. (2021).…”

Section: Deep‐learning Surrogate‐based Inversion Methodsmentioning

confidence: 99%

“…In Wang et al (2020), the TgNN was developed and used to deal with prediction problems in dynamic subsurface flow, which has better accuracy, generalizability, and robustness compared with regular deep neural network. Moreover, the TgNN is further developed as a surrogate with stochastic input parameters in Wang et al (2021), and is used for uncertainty quantification of subsurface flow responses. In this work, however, the proposed TgNN surrogate is used to solve inverse problems.…”

Section: Additional Remarksmentioning

confidence: 99%

“…Moreover, the TgNN is further developed as a surrogate with stochastic input parameters in Wang et al. (2021), and is used for uncertainty quantification of subsurface flow responses. In this work, however, the proposed TgNN surrogate is used to solve inverse problems.…”

Section: Additional Remarksmentioning

confidence: 99%

“…The TgNN, developed by Wang et al. (2020), can incorporate not only physical principles, but also practical engineering theories, and can be further developed as a surrogate for problems with uncertain model parameters (Wang et al., 2021). Since the TgNN surrogate can be constructed with limited simulation data, and even in a label‐free manner, fewer forward simulations are required.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Deep‐Learning‐Based Inverse Modeling Approaches: A Subsurface Flow Example

2021

Self Cite

View full text Add to dashboard Cite

Inverse modeling aims to infer uncertain parameters of a system with noisy observations of the system response, which has been widely utilized in various scientific and engineering practices, such as seismic inversion (Bunks et al., 1995), petroleum reservoir history matching (Oliver et al., 2008), aquifer parameter estimation (Carrera & Neuman, 1986a, 1986b, 1986c), and medical imaging (Arridge, 1999). Under certain conditions, inverse problems can be viewed as optimization problems, in which the model parameters are modified, such that the predictions from forward models can match the measurements. In addition, prior knowledge can be used to regularize the objective function and to obtain the maximum a posteriori estimate of the model parameters. The gradient-based method is a straightforward and common way to perform inverse modeling tasks. Anterion et al. (1989) presented a rigorous analytical method to calculate the gradients of observations with respect to reservoir characteristics, which can then be used to assist the process of parameter adjustment for history matching. For calculating the required gradients in inverse modeling, the adjoint method is a frequently utilized technique (

show abstract

Section: Discussionmentioning

confidence: 99%

“…Furthermore, the effect of the number of labeled data and collocation points has been studied in Wang et al. (2021).…”

Section: Deep‐learning Surrogate‐based Inversion Methodsmentioning

confidence: 99%