Neural Network Applied to Blasting Vibration Control Near Communities in a Large-Scale Iron Ore Mine

Abstract

One of the major issues in large open pit mines near communities is related to the human discomfort and possible structural damages caused by blasting vibrations. The usual approach is to determine the attenuation law based on the scaled distance and the explosive charge per delay. Recent research work conducted by different authors focus on the characterization of the local vibration attenuation law based on blasting energy, charge per delay and distance between the blasting and the monitoring points, which have allowed the determination of the maximum charge per delay for a controlled blasting. This paper provides the results of artificial neural network application to predict the blast-induced ground vibration and to allow the assessment of the environmental impact or possible structural damage, using as inputs the blastability index, the distances between the community and the blasting area, the explosive charge per delay and the blasting pattern parameters. The case study is from a large-scale open pit iron mine located at Minas Gerais state in Brazil. The methodology involves blast-induced vibration monitoring, the design, training, validation and testing of a neural network, and the control of the ground vibrations by controlling the maximum charge per delay. The results obtained in the network development phases have provided satisfying correlations between the predicted and measured values, with the values of R2 ranging from 0.8512 to 0.9639 in the different network development phases.