TKLE-BPINN: A Bayesian physics-informed inversion framework for high-dimensional parameter identification in geotechnical subsurface systems

Key Points

• Improved parameter identification is achieved using a physics-informed inversion framework.
• The framework utilizes Bayesian principles to enhance accuracy in geotechnical modeling.
• Analysis evaluated high-dimensional systems, indicating significant advancements over traditional methods.
• Implications support better predictions in subsurface behavior, though further testing in diverse conditions is needed.