Solution of advection-diffusion-reaction inverse problems with Physics-Informed Neural Networks


  • Roberto Mamud
  • Carlos T. P. Zanini
  • Helio S. Migon
  • Antônio J. Silva Neto



Inverse Source Problem, Parameter Estimation, Physics-Informed Neural Network


In this work, two inverse problems related to pollutant dispersion in a river considering the advection-dispersion-reaction equation are studied along with a Neural Network approach. The first inverse problem concerns the estimation of the reaction parameter in an homogeneous equation, and the second one concerns the estimation of source pollution location in the non-homogeneous case. Both inverse problems are solved by two multiplayer perceptron networks: the usual Artificial Neural Network (ANN) and the Physics-Informed Neural Network (PINN), which is a special type of neural network that includes the physical laws that describes the phenomena in its formulation . Numerical experiments related to both inverse problems with ANN and with PINN are presented, demonstrating the feasibility of the proposed approach.


Não há dados estatísticos.

Biografia do Autor

Roberto Mamud


Carlos T. P. Zanini


Helio S. Migon


Antônio J. Silva Neto



M Abadi et al. TensorFlow: Large-scale machine learning on heterogeneous systems. Software available from 2015.

M Andrle and A El Badia. “On an inverse source problem for the heat equation. Application to a pollution detection problem, II”. In: Inverse Problems in Science and Engineering 23.3 (2015), pp. 389–412.

A G Baydin et al. “Automatic Differentiation in Machine Learning: a Survey”. In: Journal of Machine Learning Research 18 (2018), pp. 1–43.

A El Badia, T Ha-Duong, and A Hamdi. “Identification of a point source in a linear advection–dispersion–reaction equation: application to a pollution source problem”. In: Inverse Problems 21 (2005), pp. 1121–1136.

L C Evans. Partial Differential Equations. 1st Edition. Americal Mathematical Society, 1998.

D P Kingma and J L Ba. “ADAM: A METHOD FOR STOCHASTIC OPTIMIZATION”. In: Proceedings of the International Conference on Learning Representations - ICLR 2015. San Diego, California, 2015.

F. D. Moura Neto and A. J. Silva Neto. An Introduction to Inverse Problems with Applications. 1st Edition. New York: Springer, 2013.

M Raissi, P Perdikaris, and G E Karniadakis. “Physics-informed neural networks: A deep learning framework for solving forward and inverse probems involving nonlinear partial differential equations”. In: Journal of Computational Physics 378 (2019), pp. 686–707.






Trabalhos Completos