Determination of effective parameters on fragmentation using artificial neural networks

One of the basic stages in open pit mining is blasting operation in which a proper fragmentation has to be reached to optimize overall costs. Therefore, determination of effective parameters on fragmentation seems to be very important. Generally, empirical methods are used to determine these parameters. However, considering complexity of blast design, these methods are not enough efficient, hence new methods have to be applied. In this research work, using artificial neural networks technique a model was developed to determine the most effective parameters on blasting fragmentation in Chadormalu iron mine of Iran. Comparing different types of networks, a four-layer generalized regression neural network (GRNN) with an architecture 7-18-2-1 was found to be optimum. Also sensitivity analysis revealed that blastability index, delay time and blast hole diameter are the most effective parameters and burden to spacing ratio, stemming, water content and powder factor are the least effective parameters on the rock fragmentation in blasting operation.