Optimization of Rock Blasting Fragmentation in Open Pit Using Predictive Models

Noriega-Vidal E.M.; Cotrina-Teatino M.A.; Anticona-Cueva T.J.; Anticona-Cueva J.Y.; Vera-Encarnación J.K.; Valdiviezo-Velarde A.Y. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2023) · DOI: 10.18687/leird2023.1.1.414

Resumen

The main objective of this research was to optimize the fragmentation of open pit rock blasting through predictive models. To do this, we worked with 47 samples of blasting records from an open pit mine and an artificial neural network (ANN) model was programmed in Python 3.11.4. 80% of the samples were randomly chosen to carry out the ANN training and 20% for the testing. In the architecture of the artificial neural network, three layers were added (input, hidden and output), in which 8 neurons were generated for the input parameters and 13 hidden neurons, focusing on P80, P50 and P20 as the output result. Finally, the Adam training algorithm was used, with a learning rate of 0.01 and a total of 600 cycles, in order to minimize the mean square error. As a result, a connection coefficient R2 of 0.91 was obtained for the P80. For P20 and P50, the R2 was 0.62 and 0.84 respectively. Concluding that, it was possible to predict in an acceptable way the fragmentation of the blast using the ANN, and with this, to optimize the fragmentation of the rock blast in open pit.

Instituciones: Universidad Nacional de Trujillo; Universidad César Vallejo