When Physics Meets Machine Learning: A Survey of Physics-Informed Machine Learning

TL;DR

This survey reviews recent developments in physics-informed machine learning, highlighting how integrating physics knowledge with data-driven models enhances data efficiency, model generalizability, and physical plausibility.

Contribution

It systematically summarizes motivations, methods, and physics knowledge integration strategies in PIML, and discusses future challenges and research directions.

Findings

PIML effectively mitigates training data shortages.

It improves model generalizability and physical plausibility.

The survey identifies key challenges and opportunities in PIML.

Abstract

Physics-informed machine learning (PIML), referring to the combination of prior knowledge of physics, which is the high level abstraction of natural phenomenons and human behaviours in the long history, with data-driven machine learning models, has emerged as an effective way to mitigate the shortage of training data, to increase models' generalizability and to ensure the physical plausibility of results. In this paper, we survey an abundant number of recent works in PIML and summarize them from three aspects: (1) motivations of PIML, (2) physics knowledge in PIML, (3) methods of physics knowledge integration in PIML. We also discuss current challenges and corresponding research opportunities in PIML.