Evaluation Of Artificial Neural Networks As A Reliable Tool In Blast Design

This paper is an evaluation of Artificial Neural Networks (ANN) as a tool in the design of the geometry of surface blast patterns. The built model uses eight different parameters, which affect the design of the pattern. Those parameters are rock type, stratification, blasthole diameter, bench height, type of explosive, priming position, powder factor and fragmentation size required. The network was trained to predict burden and spacing of the blast pattern. The model was built and trained (back-propagation technique) using 43 case histories collected from the literature. The model has then been validated using 16 cases from operational quarries.