This project utilizes a Physics-Informed Neural Network (PiNN) to analyze slope stability in multi-layered soil systems. By integrating physical principles with neural network capabilities, it effectively predicts slope deformation and failure across different soil layers.
The PiNN models a 2D slope problem with multi-layered soil using physics equations (linear elasticity and gravity). It aims to predict slope stability similarly to traditional methods like the Finite Element Method (FEM), while effectively capturing the interactions between different soil layers under load.
- Predicts slope stability in multi-layered soil systems using PiNN.
- Compares PiNN results with FEM.
- Shows displacement and deformation.
- Handles gravity loading.
- Exports displacement data for plotting.
The PiNN is trained using:
- Physics equations: 2D linear elastic
- Material parameters:
- Young's Modulus soil layer 1 (E): 50000 kN/m² (for example) to compare with FEM from Plaxis2d
- Young's Modulus soil layer 2 (E): 20000 kN/m²
- Poisson's Ratio (ν): 0.3
- Unit Weight (γ): 18 kN/m³ (for example) to compare with FEM from Plaxis2d
- Boundary Conditions: Fixed displacements on the bottom, left, and right sides.
Install the necessary tools using:
pip install torch matplotlib numpy
pip install pandas
pip install pytorch