PINNs algorithm and its application in geotechnical engineering

被引：0
|
作者：
Lan P. [1 ]
Li H.-C. [1 ]
Ye X.-Y. [1 ]
Zhang S. [1 ]
Sheng D.-C. [1 ,2 ]
机构：
[1] School of Civil Engineering, Changsha Central South University, Changsha
[2] School of Civil and Environmental Engineering, Sydney University of Technology, Sydney, 2007, NSW
来源：
Yantu Gongcheng Xuebao/Chinese Journal of Geotechnical Engineering | 2021年 / 43卷 / 03期
关键词：
Automatic differentiation; Continuous drainage boundary condition; Mesh-free algorithm; Parameter inversion; Physical information neural networks;
D O I：
10.11779/CJGE202103023
中图分类号：
学科分类号：
摘要：
The physical information neural networks (PINNs) algorithm, a new mesh-free algorithm, uses the automatic differential method to embed the partial differential equation directly into the neural networks so as to realize the intelligent solution of the partial differential equation, which has the advantages of fast convergence speed and high computational accuracy. The PINNs algorithm has a promising application in geotechnical engineering because it can solve the complex partial differential equations (PDEs) and inverse the unknown parameters of the PDEs. In order to verify the feasibility of the PINNs algorithm in geotechnical engineering, the one-dimensional consolidation process with the continuous drainage boundary condition is taken as an example to illustrate the procedures of the PINNs algorithm in terms of both the forward and inverse problems. The results show that the PINNs solution is highly consistent with the analytical one, indicating that the PINNs algorithm can provide an alternative approach for solving the related problems in geotechnical engineering. © 2021, Editorial Office of Chinese Journal of Geotechnical Engineering. All right reserved.
引用
收藏
页码：586 / 592
页数：6
相关论文
共 16 条
  • [1] RAISSI M, PERDIKARIS P, KARNIADAKIS G E, Et al., Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, Journal of Computational Physics, 378, pp. 686-707, (2019)
  • [2] LU L, MENG X, MAO Z, Et al., DeepXDE: A deep learning library for solving differential equations, (2019)
  • [3] TARTAKOVSKY A M, MARRERO C O, PERDIKARIS P, Et al., Physics-informed deep neural networks for learning parameters and constitutive relationships in subsurface flow problems, Water Resources Research, 56, 5, (2020)
  • [4] MAO Z, JAGTAP A D, KARNIADAKIS G E., Physics-informed neural networks for high-speed flows, Computer Methods in Applied Mechanics and Engineering, 360, (2020)
  • [5] RAISSI M, KARNIADAKIS G E., Hidden physics models: machine learning of nonlinear partial, differential equations, Journal of Computational Physics, 357, pp. 125-141, (2018)
  • [6] FANG Z, ZHAN J., Deep physical informed neural networks for metamaterial design, IEEE Access, 8, pp. 24506-24513, (2019)
  • [7] PANG G, LU L, KARNIADAKIS G E, Et al., FPINNs: Fractional physics-informed neural networks, SIAM Journal on Scientific Computing, 41, 4, pp. A2603-A2626, (2019)
  • [8] ABADI M, BARHAM P, CHEN J, Et al., TensorFlow: a system for large-scale machine learning, 12th USENIX Symposium on Operating Systems Design and Implementation, pp. 265-283, (2016)
  • [9] ADAM P, GROSS S, CHINTALA S, Et al., Automatic differentiation in PyTorch, 31st Conference on Neural Information Processing Systems(NIPS 2017), (2017)
  • [10] ROBBINS H, MONRO S., A stochastic approximation method, Annals of Mathematical Stats, 22, 3, pp. 400-407, (1951)
12