Prediction System of Blasting Rocks Fragmentaion Based on Machine Learning
The development of mining resources cannot be separated from the key link of engineering blasting.It is crucial to predict the block size generated by blasting in order to ensure job safety.To accurately predict the fragmentation of blasting rocks,this paper proposes a machine learning based fragmentation prediction model,which utilizes the Django framework to implement a corresponding visual interface system,facilitating model training through simple interaction among workers.Based on the actual blasting block size library,Spearman feature correlation analysis is introduced to preprocess the correlation of input parameters.By using three optimization algorithms,namely DE,GWO,and PSO,to predict the rock size in mines,the DE-XGBoost model with the highest prediction accuracy was compared.The root mean square error of the model was 0.0284,the average relative error was 8.401%,and the coefficient of determination was 0.9698.This indicates that the model has improved the accuracy of blasting block size prediction to a certain extent and has good prospects in practical engineering applications.
万方数据
维普
