Abstract:Efficient real-time solvers for forward and inverse problems are essential in engineering and science applications. Machine learning surrogate models have emerged as promising alternatives to traditional methods, offering substantially reduced computational time. Nevertheless, these models typically demand extensive training datasets to achieve robust generalization across diverse scenarios. While physics-based approaches can partially mitigate this data dependency and ensure physics-interpretable solutions, addressing scarce data regimes remains a challenge. Both purely data-driven and physics-based machine learning approaches demonstrate severe overfitting issues when trained with insufficient data. We propose a novel Tikhonov autoencoder model-constrained framework, called TAE, capable of learning both forward and inverse surrogate models using a single arbitrary observation sample. We develop comprehensive theoretical foundations including forward and inverse inference error bounds for the proposed approach for linear cases. For comparative analysis, we derive equivalent formulations for pure data-driven and model-constrained approach counterparts. At the heart of our approach is a data randomization strategy, which functions as a generative mechanism for exploring the training data space, enabling effective training of both forward and inverse surrogate models from a single observation, while regularizing the learning process. We validate our approach through extensive numerical experiments on two challenging inverse problems: 2D heat conductivity inversion and initial condition reconstruction for time-dependent 2D Navier-Stokes equations. Results demonstrate that TAE achieves accuracy comparable to traditional Tikhonov solvers and numerical forward solvers for both inverse and forward problems, respectively, while delivering orders of magnitude computational speedups.

Solving Inverse Heat Transfer Problems Without Surrogate Models: A Fast, Data-Sparse, Physics Informed Neural Network Approach

Physics-Informed Deep Neural Network for Inverse Heat Transfer Problems in Materials

Physics-informed neural networks for solving nonlinear diffusivity and Biot's equations

PHYSICS-INFORMED NEURAL NETWORK FOR INVERSE HEAT CONDUCTION PROBLEM

Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

Solving inverse-PDE problems with physics-aware neural networks

A Physics-Informed Deep Learning Method for Solving Direct and Inverse Heat Conduction Problems of Materials

Physics-informed neural networks for studying heat transfer in porous media

Physics Constrained Learning for Data-driven Inverse Modeling from Sparse Observations

Data-driven and physics-informed deep learning operators for solution of heat conduction equation with parametric heat source

Solving Inverse Problems in Steady State Navier-Stokes Equations using Deep Neural Networks

An Analysis of Physics-Informed Neural Networks

Physics-Informed Neural Networks for Heat Transfer Problems

TAEN: A Model-Constrained Tikhonov Autoencoder Network for Forward and Inverse Problems

Temperature Field Reconstruction of 3d Components Through a Neural Network-Based Method

Learning a Neural Solver for Parametric PDE to Enhance Physics-Informed Methods

Physics-constrained convolutional neural networks for inverse problems in spatiotemporal partial differential equations

Physics-informed and graph neural networks for enhanced inverse analysis

The Unreasonable Effectiveness of Solving Inverse Problems with Neural Networks

Data-Guided Physics-Informed Neural Networks for Solving Inverse Problems in Partial Differential Equations

Physics-informed PointNet: A deep learning solver for steady-state incompressible flows and thermal fields on multiple sets of irregular geometries