Intelligent construction method for rock slope fracture network model based on discrete element and neural network.

Area of Science:

• Geotechnical Engineering
• Artificial Intelligence in Geoscience

Background:

• Characterizing complex fracture networks in steep rock slopes is a significant engineering challenge.
• Existing methods often struggle with the intricate nature of these geological structures.

Purpose of the Study:

• To develop an intelligent Discrete Fracture Network (DFN) construction method integrating field adit data and AI algorithms.
• To accurately model complex internal fracture networks in high-steep rock slopes.

Main Methods:

• Utilized engineering geological surveys to gather spatial fracture data from adit sites.
• Established a 3D DFN generation model framework using Monte Carlo principles and 3DEC software simulations.
• Employed Back Propagation (BP) and Cascade Correlation (CC) neural networks to map adit fracture parameters to DFN model parameters.

Main Results:

• Optimized neural network models achieved high-precision inversion of fracture networks from limited adit data.
• Demonstrated prediction errors as low as 8.0% in cataclastic rock regions using a three-layer BP network.
• Validated the applicability of intelligent algorithms for fracture network modeling in diverse lithological units.

Conclusions:

• The proposed intelligent DFN construction method offers a novel approach for modeling complex rock mass structures.
• This technique enhances the accuracy of fracture network characterization in engineering applications.
• Findings have direct relevance to large-scale infrastructure projects like the Dadu River Grand Bridge.