BUILDING SOFTWARE FOR ANALYZING MUCK PILES AFTER BLASTING IN LABORATORY CONDITIONS WITH INTEGRATED ARTIFICIAL INTELLIGENCE

Abstract

In the mining and tunneling industry, the mean particle size (D_tb) of a muck pile after blasting is a key parameter for evaluating blasting efficiency. Numerous solutions exist for determining mean particle size by photographing muck piles after blasting using commercial software such as Split-Desktop and WipFrag, but these incur annual costs of thousands of US dollars. To address this limitation, the study develops software that analyzes images of muck piles after blasting, providing Swebrec, Gate-Gaudin-Schumann, and Rosin-Rammler particle size distribution (PSD) functions and their corresponding mean particle sizes. The software utilizes open-source resources and integrates the SAM2 artificial intelligence model to facilitate editing of each rock fragment detected in the image. The software’s mean particle size results are compared with those from traditional sieving analysis, with differences ranging from 5% to 7% for image segmentation data. These results demonstrate the software’s feasibility as a low-cost, reliable alternative to the sieving process.