Enhancing predictions of blast-induced ground vibration in open-pit mines: Comparing swarm-based optimization algorithms to optimize self-organizing neural networks

The objective of this paper is to present a method for predicting blast-induced ground vibration in open-pit mines that is based on the use of self-organizing neural networks (SONIA) and metaheuristic algorithms. In order to improve the accuracy of the SONIA model, several metaheuristic algorithms w...

Full description

Saved in:

Bibliographic Details
Published in:	International journal of coal geology Vol. 275; p. 104294
Main Authors:	Nguyen, Hoang, Bui, Xuan-Nam, Topal, Erkan
Format:	Journal Article
Language:	English
Published:	Elsevier B.V 01.07.2023
Subjects:	Applied soft computing Blast-induced ground vibration Metaheuristic optimization algorithm Open-pit coal mine Peak particle velocity Self-organizing neural network FFA DE HPSOGWO PR BNN dh UCS ELM QRNN RVM MRFO NMRA Peak particle velocity BBO MLP CSmj QC ANFIS DOE ML Hh BGAMs BCF SpaSO AI MAPE VOD HKM PSO AO FCM GEP RF HGS JA SONIA ν AutoencoderNN CART ANN RS ABC B D E BI El H SaDE SRH KNN K SVM BO HHO SC Q BSVR R GBM S XGBoost MFO PPV GA Nh Self-organizing neural network GD Tdl Applied soft computing CA CSO GWO ST f ICA Nblast_group Open-pit coal mine WOA CL Pν SCA Sd AGPSO CRANFIS Blast-induced ground vibration PE MARS PF RMR Metaheuristic optimization algorithm HD IM-ANN GFNN
ISSN:	0166-5162
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	The objective of this paper is to present a method for predicting blast-induced ground vibration in open-pit mines that is based on the use of self-organizing neural networks (SONIA) and metaheuristic algorithms. In order to improve the accuracy of the SONIA model, several metaheuristic algorithms were employed, including the Manta Ray Foraging Optimization (MRFO), Hunger Games Search (HGS), Aquila Optimization (AO), and Naked Mole-Rat Algorithm (NMRA). Additionally, the k-fold cross-validation technique was used to identify the best parameters for the algorithms, which were then used to retrain the models for predicting blast-induced ground vibration. The effectiveness of the proposed method was evaluated using a case study of an open-pit coal mine in Vietnam, which had 288 blasting events. The study found that SONIA was a suitable neural network for predicting blast-induced ground vibration due to its self-organizing structure, even with a small dataset containing complex relationships. However, the SONIA model could be further optimized using the metaheuristic algorithms to improve its accuracy. The study found that the MRFO-SONIA model was the most reliable and accurate, with the lowest error (MAE = 0.379, RMSE = 0.453, MAPE = 0.08) and the highest reliability (R2 = 0.896). Meanwhile, the HGS-SONIA, AO-SONIA, and NMRA-SONIA models provided lower performance, with MAE values of 0.455, 0.500, and 0.492, RMSE values of 0.552, 0.603, and 0.580, MAPE values of 0.100, 0.112, and 0.111, and R2 values of 0.845, 0.815, and 0.829, respectively. The results of this study demonstrated the potential of using metaheuristic-based SONIA models to enhance predictions of blast-induced ground vibration in open-pit mines. This approach could prove useful in other operations in open-pit mines where there is a need to predict vibrations or other disturbances/negative effects resulting from specific mining activities. •Self-organizing neural network is applied to predict PPV at open-pit coal mine.•Advanced metaheuristic algorithms are applied to optimize the predictive model.•Advanced techniques in data mining are applied to evaluate the models.•The MRFO-SONIA model is proposed as the best model for predicting PPV.
ISSN:	0166-5162
DOI:	10.1016/j.coal.2023.104294