Summary
This review examines the integration of physics-based constraints into machine learning models, particularly for civil engineering applications where conventional ML trained on laboratory data often fails in real-world deployment due to data shift. Physics-based ML models combine data, partial differential equations, and mathematical representations of physical laws to improve generalisation and robustness. The authors survey the historical development of this approach and its relevance to fluid dynamics, quantum mechanics, and computational resource optimisation.
Regional applicability
This paper is a methodological review of ML approaches in civil engineering and does not directly address agricultural, soil, or food system applications. The findings on physics-based ML generalisation may have tangential relevance to agricultural engineering or soil modelling, but transferability to United Kingdom farming practice is not evident from the scope.
Key measures
Not applicable — review paper without empirical measurements
Outcomes reported
This is a narrative review of physics-based machine learning approaches and their application to civil engineering problems. The paper examines how integrating physical laws and partial differential equations into ML models addresses data shift and real-world generalisation challenges.
Topic tags
Dig deeper with Pulse AI.
Pulse AI has read the whole catalogue. Ask about this record, its theme, or how the findings apply to UK farming and policy — every answer cites the underlying studies.