Physics-informed neural networks (PINNs) for dynamic pile-soil interaction problems

Key Points

• Improved modeling of dynamic pile-soil interactions enhances predictive accuracy with physics-informed neural networks.
• This study shows that using PINNs can significantly reduce computational time while maintaining accuracy through a machine learning approach.
• Dynamic simulations were conducted to assess the pile-soil interaction, validating the efficacy of PINNs in computational modeling.
• These findings highlight the potential of using machine learning methods for complex geotechnical engineering problems.