A comparative study on the application of various artificial neural networks to simultaneous prediction of rock fragmentation and backbreak

Sayadi, Ahmad Reza; Monjezi, Masoud; Talebi, N.; Khandelwal, Manoj

doi:10.1016/j.jrmge.2013.05.007

Cited by 113 publications

(31 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Levenberg-Marquardt (LM) algorithm is used for training the network because it has good generalization ability and has the capability of providing good predictions. Generally, Log-Sigmoid ((Logsig), Hyperbolic tangent Sigmoid (Tansig), Positive Linear (Poslin) and Linear (Purelin) transfer functions are used in back propagation neural network (BPNN) [10]. The network was optimised for the number of hidden layers and type of the transfer function.…”

Section: Ann Modelmentioning

confidence: 99%

Minimisation of Surface Mining Costs using Artificial Neural Network

2017

IJRCMCE

View full text Add to dashboard Cite

Section: Ann Modelmentioning

confidence: 99%

Minimisation of Surface Mining Costs using Artificial Neural Network

2017

IJRCMCE

View full text Add to dashboard Cite

“…Previously, researchers considered burden, spacing, hole depth, and stemming as input variables for evaluating backbreak (Sayadi et al, 2013;Mohammadnejad et al, 2013;Monjezi et al, 2013;Monjezi & Dehghani, 2008;Khandelwal & Monjezi, 2012;Sari et al, 2013). Certain additional parameters such as drillhole height, specific charge (Sayadi et al, 2013), specific drilling (Sayadi et al, 2013;Mohammadnejad et al, 2013;Monjezi et al, 2013), number of rows (Faramarzi et al, 2012;Monjezi et al, 2013), powder factor (Faramarzi et al, 2012;Monjezi et al, 2013;Sari et al, 2013), delay per metre (Monjezi et al, 2013), charge per delay , rock density (Monjezi et al, 2013;, geometric stiffness (Sari et al, 2013), and rock factor (Monjezi et al, 2013) have also been included as inputs for developing backbreak models. In most cases, the backbreak phenomenon is nonlinear; therefore, artificial intelligence has been preferred to multivariate regression in the development of backbreak prediction and control models.…”

Section: Modelling the Backbreak Phenomenonmentioning

confidence: 99%

“…Others identified different combinations of sensitive parameters from the model for different situations. Some researchers have reported that increased backbreak is due to increases in stemming, burden (Sayadi et al, 2013), stiffness ratio, and improper delay timing (Gates et al, 2005). Monjezi & Dehghani (2008) observed that the most important parameters concerning the backbreak phenomenon are the ratios of stemming to burden and last-row charge to total charge, powder factor, total charge per delay, and the number of rows in a blasting round.…”

Section: Modelling the Backbreak Phenomenonmentioning

confidence: 99%

“…Artificial intelligence has been diversely applied in earth sciences recently, using fuzzy logic (Demicco & Klir, 2004;Muhammad & Glass, 2011), neural networks (Bonaventura et al, 2017;Chatterjee et al, 2010;Izadi et al, 2017;Rogiers et al, 2012;Roslin & Esterle, 2016;Muhammad et al, 2014), and neuro-fuzzy modelling techniques (Cherkassky et al, 2006;Kar et al, 2014;Valdés & Bonham-Carter, 2006;Yegireddi & Uday Bhaskar, 2009;Yurdakul et al, 2014;Zoveidavianpoor et al, 2013). Recently, several researchers have solved backbreak problems through applying neural networks (Jang & Topal, 2013;Monjezi & Dehghani, 2008;Monjezi et al, 2013;Saadat et al, 2014;Sayadi et al, 2013;Ebrahimi et al, 2016), neuro-fuzzy techniques (Ghasemi et al, 2016), stochastic optimisation (Sari et al, 2013), and machine learning techniques (Khandelwal & Monjezi, 2012;Mohammadnejad et al, 2013). The findings differed according to rock type, as the characteristics of explosives and geometry of blast design are different in different conditions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Minimising Backbreak at the Dewan Cement Limestone Quarry Using an Artificial Neural Network

Muhammad

Shah²

2017

Archives of Mining Sciences

View full text Add to dashboard Cite

Backbreak, defined as excessive breakage behind the last row of blastholes in blasting operations at a quarry, causes destabilisation of rock slopes, improper fragmentation, minimises drilling efficiency. In this paper an artificial neural network (ANN) is applied to predict backbreak, using 12 input parameters representing various controllable factors, such as the characteristics of explosives and geometrical blast design, at the Dewan Cement limestone quarry in Hattar, Pakistan. This ANN was trained with several model architectures. The 12-2-1 ANN model was selected as the simplest model yielding the best result, with a reported correlation coefficient of 0.98 and 0.97 in the training and validation phases, respectively. Sensitivity analysis of the model suggested that backbreak can be reduced most effectively by reducing powder factor, blasthole inclination, and burden. Field tests were subsequently carried out in which these sensitive parameters were varied accordingly; as a result, backbreak was controlled and reduced from 8 m to less than a metre. The resulting reduction in powder factor (kg of explosives used per m 3 of blasted material) also reduced blasting costs.Keywords: Neural Network, backbreak, sensitivity analysis, modeling, blast design, quarry Kruszenie części złoża poza obszarem prowadzonych prac strzałowych oznacza nadmierne pęka-nie skał poza ostatnim rzędem otworów strzałowych w trakcie prac w kamieniołomach i prowadzi do destabilizacji górotworu poprzez zmianę nachylenia warstw skalnych, powoduje niepotrzebną fragmentację skał i obniża efektywność prac wiertniczych. W pracy tej wykorzystano sztuczną sieć neuronową (ANN) do przewidywania zasięgu kruszenia dalszej części złoża przy wykorzystaniu 12 parametrów wejściowych. Parametry te opisują różne zmienne czynniki, np. charakterystyka materiału wybuchowego czy przyjęty plan prac strzałowych w kamieniołomie Deewan w regionie Hattar w Pakistanie. Prowadzono proces uczenia sieci dla różnej architektury modelu, wybrano model 12-2-1 ANN, jako model najprostszy, zapewniający najlepszy wynik a współczynniki korelacji uzyskane dla fazy uczenia i walidacji wyniosły odpowiednio 0.98 i 0.97. Przeprowadzona analiza wrażliwości modelu wykazała że zasięg kruszenia dalszych części złoża obniżyć można poprzez zmianę parametrów ładunku strzelniczego, zmianę nachylenia otworów strzałowych oraz zmianę przybitki. Badania terenowe w czasie których ulegały zmianie wartości wyżej wymienionych wrażliwych parametrów wykazały, że zasięg kruszenia złoża poza obszarem prac strzałowych ograniczono z uprzednich 8 m do wielkości poniżej jednego metra. Obniżenie współczynnika charakteryzującego ładunek (kg zastosowanego materiału wybuchowego przypadający na 1 m 3 rozkruszonego materiału skalnego) pozwoliło także na obniżenie kosztów prac strzałowych.Słowa kluczowe: sieci neuronowe, zasięg kruszenia, analiza wrażliwości, modelowanie, projektowanie robót strzałowych, kamieniołom

show abstract

“…A large number of tools have been developed in AI to solve the most difficult problems and ANN is one of them. Literature review indicates that a lot of work has been done on the application of ANN to predict various aspects of blast-induced ground vibrations [21][22][23][24], air overpressure [25], fly rock [26][27][28], back break [29][30][31][32], powder factor [33][34][35], estimation of blast geometry [36][37][38], estimation of fragmentation [32,[39][40][41][42][43][44][45] etc.…”

Section: Introductionmentioning

confidence: 99%

Assessment of the Effect of Blast Hole Diameter on the Number of Oversize Boulders Using ANN Model

Dhekne

Pradhan

Jade

2015

J. Inst. Eng. India Ser. D

View full text Add to dashboard Cite

Now-a-days, blasts are planned using large diameter blast holes. The loading density (kg/m) and subsequently the energy available for the breakage of the rockmass increase with the diameter. The in-hole velocity of detonation (VoD) of non-ideal explosive also boosts up with the increase in diameter till the optimum diameter is reached. The increase in the energy content and in-hole VoD cause a sizable effect on the rock fragmentation. The effect can be assessed by counting the number of oversize boulders. This paper explains as to how the technique of artificial neural network modeling was used to predict the number of oversize boulders resulting from ANFO and SME blasts with blast holes of different diameters. The results from ANFO blasts indicated that there was no significant variation in the number of oversize boulders with the diameter whereas a perceptible variation was noticed in case of SME blasts with the change in the diameter. The change in the number of oversize boulders in ANFO blasts was negligible because mean energy factor remained almost same even when the diameter of the blast holes was altered. The decrease in the number of oversize boulders in SME blasts was on account of increase in mean energy factor when the blast hole diameter was increased. The increase in the in-hole VoD due to increase in the diameter of the hole was not found to have an effect on the generation of oversize boulders as this increase was not substantial both in SME and ANFO blasts.

show abstract

A comparative study on the application of various artificial neural networks to simultaneous prediction of rock fragmentation and backbreak

Cited by 113 publications

References 21 publications

Minimisation of Surface Mining Costs using Artificial Neural Network

Minimisation of Surface Mining Costs using Artificial Neural Network

Minimising Backbreak at the Dewan Cement Limestone Quarry Using an Artificial Neural Network

Assessment of the Effect of Blast Hole Diameter on the Number of Oversize Boulders Using ANN Model

Contact Info

Product

Resources

About