Prediction of Flyrock Throw Using Gaussian Process Regression Machine Learning Models

Flyrock is a by-product of blasting that can pose dangers to man, equipment, infrastructure, and neighbouring mining communities. As such, there is the need to minimise the throw of flyrock using predictive tools. Hence, in this research, the GPR is used for flyrock throw prediction. The results from the different GPR models were compared with those of BPNN, namely LM, BR and SCG algorithm. The accuracy of prediction is Matern 5/2 GPR with R2 of 1.00 and RMSE of 0.000386, then SEGPR with R2 of 1.00 and RMSE of 0.000386; and then RQGPR with of R2 of 1.00 and RMSE of 0.000387 with the last EGPR with R2 of 0.99 and RMSE of 0.14. Keywords – Flyrock, BPNN, Gaussian Process Regression, SDG, Mining.