Physics-constrained symbolic regression for discovering closed-form equations of multimodal water retention curves from experimental data

Researchers developed a physics-constrained machine learning framework that uses genetic programming to automatically discover closed-form mathematical equations for modeling water retention in porous materials with complex pore structures. The approach represents mathematical expressions as binary trees and incorporates physical constraints to ensure scientifically valid solutions.

• →A new AI framework combines genetic programming with physics constraints to discover mathematical equations from experimental data.
• →The method addresses limitations of traditional hydraulic models that struggle with multimodal pore size distributions.
• →Mathematical expressions are evolved as binary trees using genetic algorithms guided by physical constraints.
• →The framework can automatically generate interpretable closed-form equations without requiring separate parameter identification for each pore mode.
• →Full implementation is made available as open-source software for third-party validation and extension.