Search

Four techniques, namely Voronoi, orthogonal, randomized fracture and an innovative geometric-statistical algorithm are modeled and the results are compared with the size distribution from image analysis of the field data from rock fragmentation by blasting. The results show that the proposed algorithm has higher ability in accurate prediction of real size distribution compared to image analysis. The coefficient of correlation and mean square error of prediction obtained through the algorithm were 0.9991 and 1.726, respectively. Since all these algorithms use statistical logics, 25 outputs of the models were recorded and compared. These results show that the proposed model has the greatest variance and estimates the mean size of distribution data more accurately. Also the uniformity index of proposed algorithm shows the average variance while it keeps the non-uniformity as comes from the real data. Other models show more uniform predictions. It could be concluded that the proposed model has the ability to be used in 3D modeling of rock fragmentation by blasting. [ABSTRACT FROM AUTHOR]

Blasting operation is considered as one of the cheapest methods to break the rock into small pieces in surface and underground mines. Ground vibration is a side effect of blasting and can result in damage to, or failure of, nearby structures. Therefore, it is imperative to predict ground vibration in the blasting sites. The primary objective of this paper is to propose a new model to predict ground vibration based on itemset mining (IM) and neural networks (NN), called IM–NN. It is worth mentioning that no research has tested the efficiency of IM–NN to predict ground vibration yet. IM–NN is composed of three steps; firstly, frequent and confident patterns (itemsets) were extracted by using IM. Secondly, for each test instance, the most appropriate instances were selected based on the extracted patterns. Thirdly, NN was only trained by the selected instances. To achieve the objective of this research, a dataset including 92 instances was collected from blasting events of two surface mines in Iran, Kerman province. To demonstrate the acceptability of IM–NN, the classical NN as well as several empirical equations were also developed in this study. The results indicated that IM–NN with the correlation squared (R2) of 0.944 has better performance than NN with R2 of 0.898 and may be a promising alternative to the NN for predicting ground vibration. Thus, the use of IM was a good idea to optimize and improve the NN performance. [ABSTRACT FROM AUTHOR]

Drilling and blasting is a widely-used method for rock fragmentation in open-pit mines, tunneling and civil projects. Flyrock, as one of the most dangerous effects induced by blasting, can cause substantial damage to structures and injury to human. Therefore, the ability to make proper predictions of flyrock distance is important to reduce and minimize the environmental side effects caused by blasting operation. The main goal of the present research is to develop a precise equation for predicting flyrock through particle swarm optimization (PSO) approach. For comparison purpose, multiple linear regression (MLR) was also used. In this regard, a database including several controllable blasting parameters was collected from 76 blasting events in three quarry sites, Malaysia. In modeling procedures, five effective parameters on the flyrock including burden, spacing, stemming, powder factor and rock density were used as input parameters, while flyrock was considered as output parameter. In order to check the performance of the developed models, several statistical functions, i.e., root-mean-square error, Nash and Sutcliffe and coefficient of multiple determination ( R ), were computed. The results revealed that the proposed PSO equation is more reliable than MLR in predicting the flyrock. Based on sensitivity analysis results, it was also found that the RD was the most effective parameter on the flyrock in the studied cases. [ABSTRACT FROM AUTHOR]

This study constructs and verifies a new statistical meta based-model to predict tunnel-boring machine (TBM) performance, namely, polynomial chaos expansion (PCE). To test the validity of the proposed PCE, two well-known mathematical models, namely, response surface method (RSM) and multivariate adaptive regression spline (MARS) were developed. According to the results, it can be found that the PCE model, with a coefficient of determination (R2) of 0.843, was superior in comparison with the RSM and MARS models as well as those formerly presented in the literature for the same database and rock conditions. Abbreviations: ANFIS: Adaptive Neuro-Fuzzy Inference System; ANN: Artificial Neural Networks; AR: Advance Rate; BI: Rock Brittleness; BTS: Brazilian Tensile Strength; CP: Cutterhead Power; CT: Cutterhead Torque; d: Modified Agreement Index; DNN: Deep Neural Networks; DPW: Distance between Planes of Weakness; ICA: Imperialist Competitive Algorithm; MAE: Mean Absolute Error; MARS: Multivariate Adaptive Regression Spline; NSE: Modified Nash and Sutcliffe Efficiency; NTNU: Norwegian Institute of Technology; PCE: Polynomial Chaos Expansion; PR: Penetration Rate; PSI: Point Strength Index; PSO: Particle Swarm Optimisation; R2: Coefficient of Determination; RF: Random Forests; RMR: Rock Mass Rating; RMSE: Root Mean Square Error; RQD: Rock Quality Designation; RSM: Response Surface Method; RSR: Rock Structure Rating; SE: Specific Energy; SVR: Support Vector Regression; TBM: Tunnel-Boring Machine; TF: Thrust Force; UCS: Uniaxial Compressive Strength; WZ: Weathering Zone; α: Planes Of weakness. [ABSTRACT FROM AUTHOR]

• Indirect measurement of driven piles capacity is presented in this study. • Developing an FA-ANFIS model for prediction of driven piles capacity. • Results signify effectiveness of FA-ANFIS to predict driven piles capacity. Parametric studies of driven piles response during dynamic and static load experiments, especially in the field, are generally too costly to measure. This study constructs and verifies a computational intelligence technique for predicting the bearing capacity of driven piles. The proposed technique is a hybridization of adaptive neuro-fuzzy inference system (ANFIS) and firefly algorithm (FA). It is worth mentioning that this is the first work that explores the ability of FA-ANFIS in the field of piles capacity prediction. In modeling, different FA-ANFIS and ANFIS models were constructed based on different combination of input parameters, and finally, the performances of all models were compared using statistical functions. The results demonstrated that the proposed FA-ANFIS model with coefficient of determination (R2) of 0.997 was more effective than the ANFIS in predicting the pile capacity, and thereby the effectiveness of FA to optimize ANFIS was proved. [ABSTRACT FROM AUTHOR]

Air overpressure is one of the most undesirable destructive effects induced by blasting operation. Hence, a precise prediction of AOp has vital importance to minimize or reduce the environmental effects. This paper presents the development of two artificial intelligence techniques, namely artificial neural network (ANN) and ANN based on genetic algorithm (GA) for prediction of AOp. For this purpose, a database was compiled from 97 blasting events in a granite quarry in Penang, Malaysia. The values of maximum charge per delay and the distance from the blast-face were set as model inputs to predict AOp. To verify the quality and reliability of the ANN and GA-ANN models, several statistical functions, i.e., root means square error (RMSE), coefficient of determination (R2) and variance account for (VAF) were calculated. Based on the obtained results, the GA-ANN model is found to be better than ANN model in estimating AOp induced by blasting. Considering only testing datasets, values of 0.965, 0.857, 0.77 and 0.82 for R2, 96.380, 84.257, 70.07 and 78.06 for VAF, and 0.049, 0.117, 8.62 and 6.54 for RMSE were obtained for GA-ANN, ANN, USBM and MLR models, respectively, which prove superiority of the GA-ANN in AOp prediction. It can be concluded that GA-ANN model can perform better compared to other implemented models in predicting AOp. [ABSTRACT FROM AUTHOR]