NEW APPROACHES TO EVALUATION OF REFLECTED FRAGMENTATIONTHE EXPLOSION OF THE MINING MASS AT THE QUARRY'S STEP

The article reviews the results of research on the assessment of the distribution of the granulometric composition of the destroyed rock mass after a mass explosion at the quarry ledge of LLC "Uman Granite Quarry" using the U-CARFnet and U-Net models by segmenting images of rock sample fragments based on the visual attention mechanism, the residual research module and the focus loss function. A comparative evaluation of the proposed model with traditional image segmentation methods was carried out. To improve the segmentation of rock sample fragment images, the maximum interclass variance method of the genetic algorithm based on the improved Otsu algorithm was used. The experimental results showed high accuracy and efficiency of the U-CARFnet model, which reached 97.11% with an average error of 5.50%. compared to the traditional image segmentation method. The proposed approach ensures the effectiveness of statistical analysis of images of the reflected rock mass. The quality of the blasted rock on the quarry ledge was determined on an average piece using the U-CARFnet and WipFrag methods, which were processed using the WipFrag©Win Version 2.6 computer program. The results of processing the grain size distribution data are presented in the form of a distribution curve, a histogram, and data on the grain size distribution by class (in %).