A Robust Estimation of Blasting-Induced Flyrock Using Machine Learning Decision Tree Algorithms: Random Forest, Gradient Boosting Machine, and XGBoost

In surface mining, blasting operations may have harmful effects on the nearby environment. One of the several undesirable scenarios that should be avoided is flyrock, which is generated during the rock fragmentation process. This paper presents three machine learning decision tree algorithms that we...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Mining, metallurgy & exploration Jg. 42; H. 3; S. 1609 - 1624
Hauptverfasser: Mohammed, Bouhannache, Hamza, Cheniti
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Cham Springer International Publishing 01.06.2025
Schlagworte:
Engineering
Materials Engineering
Metallic Materials
Mineral Resources
Random forest
Gradient boosting machine
Decision trees
Blasting
Flyrock
Extreme gradient boosting machines
ISSN:2524-3462, 2524-3470
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In surface mining, blasting operations may have harmful effects on the nearby environment. One of the several undesirable scenarios that should be avoided is flyrock, which is generated during the rock fragmentation process. This paper presents three machine learning decision tree algorithms that were implemented to predict blasting-induced flyrock. The distinctively developed models are called random forest (RF), gradient boosting machines (GBM), and extreme gradient boosting machines (XGBoost). Accordingly, 61 blasting events were selected from three limestone quarry predestined for cement production in Algeria. Ten controllable parameters are used as the input variables, and the measured flyrock distance is the output variable. The accuracy criteria, including determination coefficient ( R 2 ), root-mean-square error (RMSE), and the variance accounted for (VAF) mean absolute error (MAE), were used for the evaluation of the performance of models. Furthermore, the datasets were also analyzed using the Shapley Additive Explanations (SHAP) method to find out the importance and contribution of parameters to flyrock distance prediction. As a result, the XGBoost and GBM models have demonstrated a comparable high degree of accuracy, with R 2 (0.9935, 1), RMSE (2.04, 0.041), VAF (99.35, 100), and MAE (0.43, 0.016) for the XGBoost model. Followed by nearly the same results from the GBM model with R 2 (0.9935, 1), RMSE (2.05, 0.49), VAF (99.35, 100), and MAE (0.05, 0.049) for both training and test data, while the RF model revealed inferior performance with R 2 (0.86, 0.91), RMSE (9.59, 5.35), VAF (86, 91), and MAE (7.92, 4.42). In addition, the sensitivity analysis indicates that the stemming (ST) parameter had the most significant impact on blasting-induced flyrock prediction.
ISSN:2524-3462
2524-3470
DOI:10.1007/s42461-025-01267-4