Research On Particle Size Analysis Of Open Pit Mine Blasting Based On Improved U-Net

In the mining process,the particle size of blasting ore directly affects the production cost of the mine.In order to improve the blasting effect of the ore,it is necessary to statistically analyze the particle size distribution of the blasting pile.The particle size analysis of blasting heap ore has always been a research hotspot in the field of mining.The traditional particle size analysis methods of blasting reactors have limitations and shortcomings such as low efficiency and high labor cost in practical application.In the current period of rapid development of computer technology,image recognition technology is also progressing accordingly.In order to improve the efficiency and accuracy of explosion image segmentation,and to promote the precise and intelligent development of the mining industry,this paper adopts the method of deep learning to realize U-Net Algorithmic blast heap ore particle size distribution statistics.(1)A dataset of ore segmentation from blast piles in open pit mines.First,the largearea explosion image is cut into equal parts,and Labelme software is used for image annotation to obtain the original data set.Then,the dataset is augmented with data augmentation methods of rotation flip,zoom translation,and color dithering.Finally,an ideal self-made open-pit mine blasting heap ore segmentation dataset is obtained,which provides a solid foundation for the subsequent model training.(2)Explosive heap ore segmentation model based on improved U-Net algorithm.Aiming at the problems of insufficient and inaccurate segmentation of the ore in the segmented image,this paper improves the network structure on the basis of the U-Net model and adds depthwise separable convolution to the model to reduce the model size and computational cost.Batch regularization is employed in the training phase to speed up the convergence of the model.The experimental results show that the improved model can basically complete the blasting ore segmentation task,and the accuracy of the entire model for ore segmentation is better than the U-Net model before the improvement,and the average accuracy on the self-made open pit blasting ore segmentation data set An increase of 1.53%.(3)Statistics and analysis of particle size distribution of blasting heap.In order to test the effectiveness of the model in counting the particle size of detonation,five groups of detonation images are set for the experiment,and the traditional image segmentation method is used to select the K-means algorithm with the best segmentation effect and the method in this paper.The particle size distribution of blasted ore.The experimental results show that the statistical accuracy of the method in this paper is higher than that of the Kmeans algorithm.Finally,the characteristic distributions of five groups of detonation reactors were compared and analyzed in terms of particle size distribution,perimeter distribution,area distribution and volume distribution.Experiments show that compared with other methods,the particle size analysis method of blasting mine in this paper has the advantages of less limiting factors,high efficiency and higher accuracy.Length distribution,ore projected area distribution and volume distribution,and ore particle size distribution are more suitable for analyzing the particle size distribution of detonation in actual work.