A hybrid metaheuristic approach using random forest and particle swarm optimization to study and evaluate backbreak in open-pit blasting
- Dai, Yong, Khandelwal, Manoj, Qiu, Yingui, Zhou, Jian, Monjezi, Monjezi, Yang, Peixi
- Authors: Dai, Yong , Khandelwal, Manoj , Qiu, Yingui , Zhou, Jian , Monjezi, Monjezi , Yang, Peixi
- Date: 2022
- Type: Text , Journal article
- Relation: Neural Computing and Applications Vol. 34, no. 8 (2022), p. 6273-6288
- Full Text:
- Reviewed:
- Description: Backbreak is a rock fracture problem that exceeds the limits of the last row of holes in an explosion operation. Excessive backbreak increases operational costs and also poses a threat to mine safety. In this regard, a new hybrid intelligence approach based on random forest (RF) and particle swarm optimization (PSO) is proposed for predicting backbreak with high accuracy to reduce the unsolicited phenomenon induced by backbreak in open-pit blasting. A data set of 234 samples with six input parameters including special drilling (SD), spacing (S), burden (B), hole length (L), stemming (T) and powder factor (PF) and one output parameter backbreak (BB) is set up in this study. Seven input combinations (one with six parameters, six with five parameters) are built to generate the optimal prediction model. The PSO algorithm is integrated with the RF algorithm to find the optimal hyper-parameters of each model and the fitness function, which is the mean absolute error (MAE) of ten cross-validations. The performance capacities of the optimal models are assessed using MAE, root-mean-square error (RMSE), Pearson correlation coefficient (R2) and mean absolute percentage error (MAPE). Findings demonstrated that the PSO–RF model combining L–S–B–T–PF with MAE of 0.0132 and 0.0568, RMSE of 0.0811 and 0.1686, R2 of 0.9990 and 0.9961 and MAPE of 0.0027 and 0.0116 in training and testing phases, respectively, has optimal prediction performance. The optimal PSO–RF models were compared with the classical artificial neural network, RF, genetic programming, support vector machine and convolutional neural network models and show that the PSO–RF model has superiority in predicting backbreak. The Gini index of each input variable has also been calculated in the RF model, which was 31.2 (L), 23.1 (S), 27.4 (B), 36.6 (T), 23.4 (PF) and 16.9 (SD), respectively. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
- Authors: Dai, Yong , Khandelwal, Manoj , Qiu, Yingui , Zhou, Jian , Monjezi, Monjezi , Yang, Peixi
- Date: 2022
- Type: Text , Journal article
- Relation: Neural Computing and Applications Vol. 34, no. 8 (2022), p. 6273-6288
- Full Text:
- Reviewed:
- Description: Backbreak is a rock fracture problem that exceeds the limits of the last row of holes in an explosion operation. Excessive backbreak increases operational costs and also poses a threat to mine safety. In this regard, a new hybrid intelligence approach based on random forest (RF) and particle swarm optimization (PSO) is proposed for predicting backbreak with high accuracy to reduce the unsolicited phenomenon induced by backbreak in open-pit blasting. A data set of 234 samples with six input parameters including special drilling (SD), spacing (S), burden (B), hole length (L), stemming (T) and powder factor (PF) and one output parameter backbreak (BB) is set up in this study. Seven input combinations (one with six parameters, six with five parameters) are built to generate the optimal prediction model. The PSO algorithm is integrated with the RF algorithm to find the optimal hyper-parameters of each model and the fitness function, which is the mean absolute error (MAE) of ten cross-validations. The performance capacities of the optimal models are assessed using MAE, root-mean-square error (RMSE), Pearson correlation coefficient (R2) and mean absolute percentage error (MAPE). Findings demonstrated that the PSO–RF model combining L–S–B–T–PF with MAE of 0.0132 and 0.0568, RMSE of 0.0811 and 0.1686, R2 of 0.9990 and 0.9961 and MAPE of 0.0027 and 0.0116 in training and testing phases, respectively, has optimal prediction performance. The optimal PSO–RF models were compared with the classical artificial neural network, RF, genetic programming, support vector machine and convolutional neural network models and show that the PSO–RF model has superiority in predicting backbreak. The Gini index of each input variable has also been calculated in the RF model, which was 31.2 (L), 23.1 (S), 27.4 (B), 36.6 (T), 23.4 (PF) and 16.9 (SD), respectively. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
- Qiu, Yingui, Zhou, Jian, Khandelwal, Manoj, Yang, Haitao, Yang, Peixi, Li, Chuanqi
- Authors: Qiu, Yingui , Zhou, Jian , Khandelwal, Manoj , Yang, Haitao , Yang, Peixi , Li, Chuanqi
- Date: 2022
- Type: Text , Journal article
- Relation: Engineering with Computers Vol. 38, no. (2022), p. 4145-4162
- Full Text: false
- Reviewed:
- Description: Accurate prediction of ground vibration caused by blasting has always been a significant issue in the mining industry. Ground vibration caused by blasting is a harmful phenomenon to nearby buildings and should be prevented. In this regard, a new intelligent method for predicting peak particle velocity (PPV) induced by blasting had been developed. Accordingly, 150 sets of data composed of thirteen uncontrollable and controllable indicators are selected as input dependent variables, and the measured PPV is used as the output target for characterizing blast-induced ground vibration. Also, in order to enhance its predictive accuracy, the gray wolf optimization (GWO), whale optimization algorithm (WOA) and Bayesian optimization algorithm (BO) are applied to fine-tune the hyper-parameters of the extreme gradient boosting (XGBoost) model. According to the root mean squared error (RMSE), determination coefficient (R2), the variance accounted for (VAF), and mean absolute error (MAE), the hybrid models GWO-XGBoost, WOA-XGBoost, and BO-XGBoost were verified. Additionally, XGBoost, CatBoost (CatB), Random Forest, and gradient boosting regression (GBR) were also considered and used to compare the multiple hybrid-XGBoost models that have been developed. The values of RMSE, R2, VAF, and MAE obtained from WOA-XGBoost, GWO-XGBoost, and BO-XGBoost models were equal to (3.0538, 0.9757, 97.68, 2.5032), (3.0954, 0.9751, 97.62, 2.5189), and (3.2409, 0.9727, 97.65, 2.5867), respectively. Findings reveal that compared with other machine learning models, the proposed WOA-XGBoost became the most reliable model. These three optimized hybrid models are superior to the GBR model, CatB model, Random Forest model, and the XGBoost model, confirming the ability of the meta-heuristic algorithm to enhance the performance of the PPV model, which can be helpful for mine planners and engineers using advanced supervised machine learning with metaheuristic algorithms for predicting ground vibration caused by explosions. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
Performance evaluation of rockburst prediction based on PSO-SVM, HHO-SVM, and MFO-SVM hybrid models
- Zhou, Jian, Yang, Peixi, Peng, Pingan, Khandelwal, Manoj, Qiu, Yingui
- Authors: Zhou, Jian , Yang, Peixi , Peng, Pingan , Khandelwal, Manoj , Qiu, Yingui
- Date: 2023
- Type: Text , Journal article
- Relation: Mining, Metallurgy and Exploration Vol. 40, no. 2 (2023), p. 617-635
- Full Text:
- Reviewed:
- Description: Rockburst prediction is crucial in deep hard rock mines and tunnels to make safer working conditions. Due to the complex interaction of many factors involved in rockburst prediction, such as multi-variable and multi-interference factors, three hybrid support vector machine (SVM) models optimized by particle swarm optimization (PSO), Harris hawk optimization (HHO), and moth flame optimization (MFO) are proposed to predict rockburst hazard level (RHL). The RHL is determined according to four kinds of microseismic characteristic parameters including angular frequency ratio, total energy, apparent stress, and convexity radius. Then, six types of microseismic characteristic parameters are taken as input variables in 343 sets of data, including angular frequency ratio and total energy, etc. And the RHL is taken as the output target of rockburst prediction. The classification performance of PSO-SVM, HHO-SVM, and MFO-SVM hybrid models is evaluated by accuracy (ACC), precision (PRE), and kappa coefficient. Findings reveal that the MFO-SVM model performs best in terms of accuracy, with ACC, PRE, and kappa coefficients reaching 0.9559, 0.9063, and 0.9094 respectively, while PSO-SVM and HHO-SVM have similar performances. However, the PSO-SVM, HHO-SVM, and MFO-SVM all perform better than the unoptimized SVM model. This confirms that the three optimization algorithms significantly enhance the rockburst prediction capacity of the SVM model to help mine practitioners apply machine learning methods to rockburst prediction problems appropriately. © 2023, Society for Mining, Metallurgy & Exploration Inc.
- Authors: Zhou, Jian , Yang, Peixi , Peng, Pingan , Khandelwal, Manoj , Qiu, Yingui
- Date: 2023
- Type: Text , Journal article
- Relation: Mining, Metallurgy and Exploration Vol. 40, no. 2 (2023), p. 617-635
- Full Text:
- Reviewed:
- Description: Rockburst prediction is crucial in deep hard rock mines and tunnels to make safer working conditions. Due to the complex interaction of many factors involved in rockburst prediction, such as multi-variable and multi-interference factors, three hybrid support vector machine (SVM) models optimized by particle swarm optimization (PSO), Harris hawk optimization (HHO), and moth flame optimization (MFO) are proposed to predict rockburst hazard level (RHL). The RHL is determined according to four kinds of microseismic characteristic parameters including angular frequency ratio, total energy, apparent stress, and convexity radius. Then, six types of microseismic characteristic parameters are taken as input variables in 343 sets of data, including angular frequency ratio and total energy, etc. And the RHL is taken as the output target of rockburst prediction. The classification performance of PSO-SVM, HHO-SVM, and MFO-SVM hybrid models is evaluated by accuracy (ACC), precision (PRE), and kappa coefficient. Findings reveal that the MFO-SVM model performs best in terms of accuracy, with ACC, PRE, and kappa coefficients reaching 0.9559, 0.9063, and 0.9094 respectively, while PSO-SVM and HHO-SVM have similar performances. However, the PSO-SVM, HHO-SVM, and MFO-SVM all perform better than the unoptimized SVM model. This confirms that the three optimization algorithms significantly enhance the rockburst prediction capacity of the SVM model to help mine practitioners apply machine learning methods to rockburst prediction problems appropriately. © 2023, Society for Mining, Metallurgy & Exploration Inc.
- «
- ‹
- 1
- ›
- »