Research on the Mechanical Properties of Xiyu conglomerate Based on Deep Learning Digital Image Analysis

Abstract Xiyu conglomerate is a special geological body that lies between rock and soil. Its mechanical properties are determined by the microscopic characteristics of the framework particles. How to identify its microscopic structure and evaluate its impact on mechanical properties has attracted much attention in the geotechnical field. In this paper, based on the typical Xiyu conglomerate of the Momoke Key Water Control Project, conducted indoor experiments. It introduced the deep learning digital image recognition method to identify its fabric characteristics, and used particle flow numerical simulation to analyze the Mesoscopic Mechanical Properties. It also explored the effects of cement content, cement composition, and stone content on the strength characteristics of the Xiyu conglomerate. The results show that the higher the content of calcareous cement, the more closely the particles are interlocked, and the greater their density will be. When the stone content is fixed at 50%, as the proportion of calcareous cementation increases, the internal friction angle of the Western Regions conglomerate keeps increasing. The cohesion gradually increases and then decreases. When the content of calcareous cementation increases from 25% to 75%, the internal friction angle rises from 16.3°to 32.1°. When the proportion of cementing material is fixed at 50%, as the content of stone increases, the squeezing and biting effects between the boulders become significant, and the internal friction angle and cohesion also increase significantly. The research results can provide a reference for engineering construction in the Xiyu conglomerate area.