S. M. Mahdi Mirabedi - School of Mining, College of Engineering University of Tehran, Tehran, Iran.
Mehdi Rahmanpour - School of Mining, College of Engineering University of Tehran, Tehran, Iran.
Yousef Azimi - Research Centre for Environment and Sustainable Development, RCESD, Department of Environment, Tehran, Iran.
Hassan Bakhshandeh Amnieh - School of Mining, College of Engineering University of Tehran, Tehran, Iran.

The shape of a blasted rock mass, or simply muckpile, affects the efficiency of loading machines. Muckpile is defined with two main parameters known as throw and drop, while several blasting parameters will influence the muckpile shape. This paper studies the prediction of muckpile shape in open-pit mines by applying an artificial neural network designed by a genetic algorithm. In that regard, a genetic algorithm has been used in preparing the neural network architecture and parameters. Moreover, input variables have been reduced using the principal component analysis. Finally, the best models for predicting throw and drop are determined. Analyzing the performance of the proposed models indicates their superiority in predicting muckpile shape. As a result, the Mean Squared Error of throw is ۰.۵۳ for train data and ۱.۲۴ for test data. While for the drop, the errors are ۰.۴۵ and ۰.۵۸ for the training and testing data. Furthermore, sensitivity analysis shows that specific-charge effects drop and throw more.

Hybrid genetic algorithm neural network, blasting, muckpile, Principal component analysis