Experimental study of inverse-grading transport of single coarse particles of different shapes based on the YOLO algorithm

The flow of granular particles is characterized by particle-size sorting called “inverse-grading transport”, and it is important to carry out a series of basic studies on the inverse-grading transport behavior of coarse particles for disaster prevention and mitigation and related theoretical study of particle separation. In order to investigate the influence of shape on the inverse-grading transport characteristics of a single coarse particle, a series of cyclic shear tests were conducted utilizing 3D sand printing technology alongside a self-constructed two-dimensional cyclic shear test device. Using the YOLO target detection algorithm, the inverse-grading transport trajectory, rotation characteristics, and local structure were analyzed. A kinematic equivalent analysis method classified transport behaviors of coarse particles, revealing correlations between single coarse particles of different shapes and macroscopic segregation patterns. The results indicate that: (1) Single coarse particles slowly ascend from the bottom center, with their vertical transport rate increasing until they reach the surface. (2) Particle shape significantly affects the inverse-grading transport of single coarse particles. The closer the coarse particles are to the free surface, the lower is the local volume fraction above them, while the volume fraction below them increases. (3) The inverse-grading transport of coarse particles is significantly correlated with their own rotation and with changes in the local structure of the granular medium around them. Our experiments thus show that the inverse-grading phenomenon of landslide-debris flow is mainly caused by changes in the local structure of the granular medium around the coarse particles.