APPLICATION OF PHYSICS AI IN RAILWAY VEHICLE DESIGN – A REVIEW

Abstract

Modern railway vehicle design must satisfy numerous requirements related to safety, reliability, weight reduction, energy efficiency, and performance. Numerical methods such as FEM and multibody system models provide reliable results, but their application to complex problems and numerous operating conditions can require significant computational time and resources. Physics AI combines artificial intelligence with physical knowledge and offers a promising way to overcome these limitations. This paper reviews its applications in railway engineering, including collisions and braking, vehicle dynamics and control, structural analysis and fatigue, wheel–rail and vehicle–track interaction, aerodynamics, pantograph–catenary systems, and maglev systems. The reviewed studies show that incorporating physical knowledge into AI models can improve accuracy, robustness, physical consistency, and generalization while reducing data requirements and computational cost. These results highlight the significant potential of Physics AI for future railway system design, analysis, monitoring, and control.