Search Documents

By A. Jahedsaravani, M. Massinaei, M. Zarie

Deep learning is a subset of machine learning that uses artificial neural networks for extracting high-level features from image data. In the present study, a soft sensor is proposed for the prediction of the flotation performance through froth features generated by the use of pre-trained convolutional neural networks. Several state-of-the-art convolutional neural networks (AlexNet, GoogLeNet, VGGNet, ResNet, and SqueezeNet) pre-trained on the ImageNet database are used to predict the metallurgical performance of two flotation systems. The first case study is a batch copper flotation system video-captured over a wide range of process conditions. The second case study is an industrial coal flotation column equipped with a continuous video recording system. The pre-trained networks are used to extract features from the froth images, and these features are subsequently used to predict the flotation conditions and performance. The prediction results by the pre-trained algorithms were compared with the traditional image processing algorithms. This demonstrates the ability of the pre-trained structures to generalize to images outside the ImageNet database. GoogLeNet outperforms other network architectures and provides more accurate predictions of the flotation process behavior and performance.

May 5, 2023

By R. K. Sinha, B. S. Choudhary, Desh Deepak, A. K. Mishra, Hemant Agrawal, Shankar Kumar

There are several methods (empirical, statistical, and machine learning tools) to calculate blast-induced ground vibration. Still, due to complications in the blasting procedure, under varying blasting loads and strata conditions, it is not easy to predict and simulate blast vibration precisely. Therefore, a study was conducted to simulate the field blast data in Ansys software using explicit dynamics. For this study, monitoring of ground vibration for 20 single hole blasts was carried out. The density of emulsion explosives was in the range of 0.9 to 1.15 g/cc. In the laboratory, rock properties were determined from the rock sample collected from the mine site before blasting. The in-hole velocity of detonation (VoD) and the density of the emulsion explosive were also determined. These parameters were used for numerical analysis to know the ground vibrations at specified points. The results of ground vibration predicted by the model are validated using field data.

By R. Nyirenda, B. Besa, N. Dzimunya

The accurate estimation of peak particle velocity (PPV) is crucial during the design of bench blasting operations in open pit mines, since the vibrations caused by blasting can significantly affect the integrity of nearby buildings and other structures. Conventional models used to predict blast-induced vibrations are not capable of capturing nonlinear relationships between the different blasting-related parameters. Soft computing techniques, i.e., techniques that are founded on the principles of artificial intelligence, effectively model these complexities. In this paper, we use the random forest (RF) algorithm to develop a model to predict blast-induced ground vibrations from bench blasting using 48 data records. The model was trained and tested using WEKA datamining software. To build this model, a feature selection process using several combinations of Attribute Evaluators and Search Methods under the WEKA Select Attributes tab was performed. The correlation coefficient of the actual data and RF model-predicted data was 0.95, and the weighted average of the relative absolute error (RAE) was 10.9%. The RF model performance was also compared to the equivalent-path-based (EPB) equation on the testing data-set, and it was seen that the RF model can effectively be used to predict PPV. The study also demonstrates that the EPB equation is a suitable empirical method for predicting PPV.

Mar 30, 2023

By Pavel A. Torres, Darwin Torres, Sandro Huaman

Given the limited efficiency of blast-induced ground vibrations prediction empirical models, due to the complex geological system from a Peruvian mine. An artificial neural network was built in order to get better results. More accurate and precise predictions allowed the generation of blasting domains for future designs and decision making.

Feb 1, 2020

By S. A. Suhler

The detection of hazardous geological anomalies through remote sensing techniques is the subject of current research efforts sponsored by the U.S. Bureau of Mines. In this review, hazardous geological anomalies are discussed with respect to the "view" presented along with potential operational concepts for each technique. The design and development of a borehole radar is described with emphasis on the potential range and resolution capability of the system under realistic underground conditions. Results of field tests at two coal mines and at a hard rock future subway location are presented.

Jan 1, 1978

By M. Hepokoski, M. Klein, L. Yan

"NIOSH research has shown that heat/humidity buildup is a major concern within coal mine refuge alternatives (RAs). These high temperature and humidity levels inside an RA may expose occupants to heat stress. Due to the safety risks associated with testing using human subjects, NIOSH partnered with ThermoAnalytics to create detailed thermal simulation models of RAs with human occupants. The objective of this effort was to predict a miner’s core temperature response and moisture loss in environments that may be encountered in a coal mine RA. These parameters were studied across a range of temperatures and relative humidity values to determine if the current 95°F apparent temperature limit for RAs is reasonable. The results indicated that the 95°F apparent temperature limit is protective, provided that miners are supplied with sufficient water. The results also indicated that the body core temperature did not reach dangerous levels even at an apparent temperature of 130°F. However, the results show that moisture loss increases with apparent temperature. Therefore, if the apparent temperature limit were increased, the water provided in an RA would have to be increased to offset moisture loss. INTRODUCTION The purpose of this effort was to predict a miner’s core temperature and moisture loss (sweating and respiration) in environments that are likely to be encountered in a coal mine refuge alternative (RA). A range of temperatures and relative humidity were studied to determine whether the current 95°F Steadman apparent temperature (AT) limit and 2.25 quart/day water requirements for RAs as mandated by the Mine Safety and Health Administration (MSHA) are reasonable [1, 2]. The effects of metabolic heat rate (activity level), miner size (height and weight), and pose (sitting or lying) on core temperature and moisture loss were also studied. MODEL DESCRIPTION TAITherm thermal modeling software (ThermoAnalytics, Inc.) was used to simulate humans in RA-like environments within a range of temperature and relative humidity combinations. The TAITherm Human Thermal Model (HTM) predicts the dynamic thermal response of the simulated miners while accounting for clothing, human tissue thermal properties, and active thermoregulation. The following sections describe the HTM in more detail, the thermal model setup for this analysis, and core temperature and moisture loss metrics found in published literature. Human Thermal Model Overview The TAITherm human thermal model was used to simulate the heat generated by the shelter occupants, and to predict the miners’ skin and core temperatures along with their sweat evaporation rates. The thermoregulation model simulates the process by which the body attempts to maintain a constant core temperature by considering the effects of shivering, sweating, and changes in skin blood flow. This model calculates surface and deep tissue temperatures within the human body as described by a surface mesh."

Jan 1, 2017

By M. Hepokoski, M. Klein, D. S. Yantek, L. Yan

Research by the U.S. National Institute for Occupational Safety and Health (NIOSH) has shown that heat/humidity buildup is a major concern within coal mine refuge alternatives. High temperature and humidity levels inside a refuge alternative may expose occupants to heat stress. Due to the safety risks associated with testing using human subjects, NIOSH partnered with ThermoAnalytics Inc. to create detailed thermal simulation models of refuge alternatives with human occupants. The objective of this effort was to predict a miner’s core temperature response and moisture loss in environments that may be encountered in a coal mine refuge alternative. These parameters were studied across a range of temperatures and relative humidity values to determine if the current 35 °C (95 °F) apparent temperature limit for refuge alternatives is reasonable. The results indicate that the apparent temperature limit is protective, provided that miners are supplied with sufficient water. The results also indicate that the body core temperature does not reach dangerous levels even at an apparent temperature of 54 °C (130 °F). However, the results show that moisture loss increases with apparent temperature. Therefore, if the apparent temperature limit were raised, the water provided in a refuge alternative would have to be increased to offset moisture loss.

Jan 1, 2017

By M. Hepokoski, M. Klein, D. S. Yantek, L. Yan

Research by the U.S. National Institute for Occupational Safety and Health (NIOSH) has shown that heat/humidity buildup is a major concern within coal mine refuge alternatives. High temperature and humidity levels inside a refuge alternative may expose occupants to heat stress. Due to the safety risks associated with testing using human subjects, NIOSH partnered with ThermoAnalytics Inc. to create detailed thermal simulation models of refuge alternatives with human occupants. The objective of this effort was to predict a miner’s core temperature response and moisture loss in environments that may be encountered in a coal mine refuge alternative. These parameters were studied across a range of temperatures and relative humidity values to determine if the current 35 °C (95 °F) apparent temperature limit for refuge alternatives is reasonable. The results indicate that the apparent temperature limit is protective, provided that miners are supplied with sufficient water. The results also indicate that the body core temperature does not reach dangerous levels even at an apparent temperature of 54 °C (130 °F). However, the results show that moisture loss increases with apparent temperature. Therefore, if the apparent temperature limit were raised, the water provided in a refuge alternative would have to be increased to offset moisture loss.

Jan 1, 2018

By W. E. Wallace, J. A. Warren, D. Josell, D. Wheeler

"We present the results of experiments and modeling of flip-chip geometry solder joint shapes under shear loading. Modeling, using Surface Evolver, included development of techniques that use an applied vector force (normal and shear loading) as input to determine a vector displacement of the pads connected by the solder joint (standoff height and misalignment). Previous solutions solved the converse problem: fixed displacements used to determine required applied force. Such solutions were inconvenient for applications, where the applied force (chip weight) is known. Also, for geometric and materials studies of solder joint shapes involving multiple parameters, determining the equilibrium displacement from applied force by bracketing solutions could become computationally expensive. Measurements of solder joint standoff height and misalignment as functions of the applied force (normal and shear), solder volume and pad diameter are presented. Experiments were carried out for solder ball diameters from 0.38 mm (0.029 mm3 volume) to 0.15 mm (0.0019 mm3 volume) on pads of diameter 0.64 mm and 0.35 mm. Fitting of simulation to experimental results gave optimized values for the contact angle and surface tension of the solder which were consistent with measured and literature values. IntroductionIn packaging technology flip-chip has become an important method for attaching semiconductor devices to substrates. Generally, the chip and substrate are bumped with solder and then the chip is flipped onto the substrate and the solder reflowed to form conductive interconnects. An epoxy underfill is normally flowed between the substrate and chip to reduce the thermal fatigue on the solder joints. A flip-chip assembly is illustrated in Figure 1.In flip-chip applications, the use of an integrated underfill/solder bump system [1, 2) removes an extra process step, namely infiltration of the underfill. However, the presence of an underfill will retard the self-alignment of the flip-chip. A first step in understanding and predicting this retardation is to develop a model to describe normal and lateral force-displacements curves when underfill material is not present. At this stage it is unclear whether static surface tension theory is sufficient to describe a complete model for flip-chip realignment even without underfill."

Jan 1, 2001

By J. Kwon, D. Lee, H. Cho

"The use of the combined grinding and liberation model to predict liberation characteristics of iron ore comminuted by a ball mill was studied. Comminution characteristics were obtained using the one-size fraction method. The iron ore samples were crushed and separated into a number of narrow size fractions by screening. The size fractions were further divided by magnetic separation into three classes: concentrate, middling, and tailings. Subsequently, various size-grade classes were subjected to ball milling, and the results were analyzed in terms of the kinetic grinding model. The mineralogical textures of the grinding products were analyzed using mineral liberation analysis. The distribution of iron content within the size fractions was characterized by the beta distribution combined with the Andrew–Mica diagram. It was found that the grinding characteristics of the iron ore samples varied slightly with the iron content. However, the liberation characteristics were well described by the same set of liberation model parameters, regardless of composition. The variation in grinding kinetics with composition and the liberation model parameters was used in the combined comminution–liberation model to predict the evolution of the size-composition matrix as grinding proceeded. The predicted results agreed well with the experimental results. The developed model can be used to construct the yield–recovery graph and to calculate the liberation efficiency to determine the size reduction threshold for efficient separation. INTRODUCTION In mineral processing, comminution is usually the first step that ores undergo to liberate valuable minerals from gangue minerals. Generally, as the particle size decreases, higher liberation can be achieved. However, excessively small particles are unfavorable to subsequent processes, such as separation. Moreover, comminution is a highly energy-consuming process itself that accounts for approximately 60% of all energy used for processing. Therefore, there is a definite need to increase energy efficiency by applying adequate stress so that a high degree of liberation can be acquired at the coarsest size without overgrinding. To this end, it is essential to develop a model to predict the liberation of minerals as a function of size reduction."

Jan 1, 2018

By Claudio L. Schneider, Peter R. King, Reiner Neumann

"Predicting the liberation spectra that are produced from breakage of two-phase ores under the random fracture assumption can be accomplished by measuring the conditional, on length, linear grade distribution directly from images obtained from a representative sample of the unbroken texture. The technique tremendously simplifies the prediction problem by avoiding the multiple convolutions that arise from the traditional method based on a description of texture from a measurement of the chord length distributions associated with each phase in the unbroken texture. For a 3-phase system, the phase transition probabilities associated with the textural information make the traditional calculation an almost insurmountable mathematical problem. The direct measurement procedure on the other hand can be expanded to three phases comparatively easily.A coal sample is used here to demonstrate the procedure. It is concluded that the conditional, on length, linear grade distribution can be predicted in the same fashion for multi-phase systems. Accurate stereological conversion is the practical hurdle to be overcome for multi-phase systems, but it is already a reality for 3- phase systems."

Jan 1, 2003

By Hang Goo Kim

A computer simulation has been carried out to predict the distribution behavior of minor elements such as Pb. Zn, Bi, Sb and As in recently proposed copper smelting and converting processes with submerged oxygen injection in which tonnage oxygen is used with a shielding gas. In this study the technique of stepwise equilibrium modeling was employed and some of the thermodynamic data were updated or reassessed. The volatilization and slagging of undesirable elements were investigated in copper smelting (steady-state) system, copper matte converting (batch) system, and the combined (smelting + converting) system at extremly high oxygen-enrichment levels. The predicted minor-element behavior under high oxygen- enrichment was compared with that in the Noranda process (normally 42% oxygen-enrichment in smelting and 24% in converting) and the reverberatory process using air. The effects of various operating parameters such as temperature, the final matte grade in smelting, the initial matte grade in converting and the O/Fe ratio in slag on the behavior of minor elements are discussed.

Jan 1, 1991

By H Mansouri, M A. Ebrahimi Farsangi, F Faramarzi

A blast fragmentation model is developed based on the gamma function to describe the run-of-mine fragmentation distribution. The model presented is aimed to benefit from simplicity in application and also to be efficient in description of the fine to coarse part of the blast rock fragments. In fact, the gamma based model (KC-KUM), gives a unit distribution curve, which is made up of two separate equations, covering different fragment sizes in the muck pile. The generated values of the fines to central part of the equation and the central to coarse part of the equation are connected at the 40 per cent grafting point. For validation, eight production blasts, which were specially designed for this purpose, were carried out at an iron ore mine (Jalal-Abad iron ore mine/Iran). Results achieved by KC-KUM and two of the most well-known models. Kuz-Ram and Swebrec, are compared with measured data. Moreover, to assess the models’ efficiency in presenting the best distribution curve, a Logarithmic error method is chosen. Based on the achievements. KC-KUM has the least deviations by the least absolute log error of 0.52, whereas the values obtained for the Swebrec and Kuz-Ram models were 0.79 and 1.34 respectively. Furthermore, the models were applied to predict X80 for 12 production blasts carried out at Gol-e-Gohar iron ore mine/Iran. These predictions were compared with the results from these trials. The coefficient of determination (R2) and root mean square error (RMSE) are calculated. Consequently, the highest R2 was obtained for KC-KUM at 0.837 with the least RMSE of 11.31. On the other hand, R2 and RMSE are equal to 0.832 and 20.19 for Swebrec; and 0.813 and 19.75 for Kuz-Ram respectively. The results obtained suggest the superiority of the KC-KUM model over the other two models in assessment and prediction of blast fragmentation in a reliable way.CITATION:Faramarzi, F, Ebrahimi Farsangi, M A and Mansouri, H, 2015. Prediction of rock fragmentation using a gamma-based blast fragmentation distribution model, in Proceedings 11th International Symposium on Rock Fragmentation by Blasting, pp 685–692 (The Australasian Institute of Mining and Metallurgy: Melbourne).

Aug 24, 2015

By A M. Suchowerska, J P. Carter, J P. Hambleton

"In order to effectively predict the roof collapse of underground openings using continuum models, it is imperative that a realistic failure criterion is used to represent the rock mass. However, for coal mining operations, there is still uncertainty as to the appropriate failure criterion that should be used for the coal measure strata. This study compares stability numbers and collapse mechanisms for a rectangular underground opening predicted based on three commonly used failure criteria: the standard Mohr-Coulomb, Mohr-Coulomb with a tension cut-off and Hoek-Brown. Three methods of analysis are considered: an analytical upper bound method an upper and lower bound finite element formulation the displacement finite element method.The theoretical results are also compared with field measurements of subsidence above longwall coal mining. For the standard Mohr-Coulomb failure criterion, the friction angle governs the shape of the collapse mechanism, and the so-called critical width (the cavity width above which failures extend to the ground surface) ascertained from field observations in the New South Wales coalfields corresponds to a friction angle of approximately 30°. On the other hand, predictions obtained based on the Hoek-Brown failure criterion consistently overestimate the critical width. The results also show that tensile failure governs the stability of the cavity and shape of the collapse mechanism when a tension cut-off is introduced for the Mohr-Coulomb yield criterion.CITATION:Suchowerska, A M, Hambleton, J P and Carter, J P, 2014. Prediction of roof collapse for rectangular underground openings, in Proceedings AusRock 2014: Third Australasian Ground Control in Mining Conference, pp 367–374 (The Australasian Institute of Mining and Metallurgy: Melbourne)."

Nov 5, 2014

By S. Swanepoel, A. V. Cherkaev, Q. G. Reynolds, M. Erwee

Ferrochrome (FeCr) is a vital ingredient in stainless steel production and is commonly produced by smelting chromite ores in submerged arc furnaces. Silicon (Si) is a componrnt of the FeCr alloy from the smelting process. Being both a contaminant and an indicator of the state of the process, its content needs to be kept within a narrow range. The complex chemistry of the smelting process and interactions between various factors make Si prediction by fundamental models infeasible. A data-driven approach offers an alternative by formulating the model based on historical data. This paper presents a systematic development of a data-driven model for predicting Si content. The model includes dimensionality reduction, regularized linear regression, and a boosting method to reduce the variability of the linear model residuals. It shows a good performance on testing data (R2 = 0.63). The most significant predictors, as determined by linear model analysis and permutation testing, are previous Si content, carbon and titanium in the alloy, calcium oxide in the slag, resistance between electrodes, and electrode slips. Further analysis using thermodynamic data and models, links these predictors to electrode control and slag chemistry. This analysis lays the foundation for implementing Si content control on a ferrochrome smelter.

Mar 14, 2024

By Nemick JA

Recent advances in computer simulation together with field measurements of caving and microseismic activity about longwall panels, has allowed a much better understanding of the caving process and the variability due to geology. The joint research between SCT Operations and CSIRO Division of Exploration and Mining has initiated new methods of computational modelling predicting various caving patterns and strata failure far ahead of the longwall face. This work was validated by field measurements of caving and microseismic activity at the longwall face.The rock fracture distribution and the caving characteristics of a range of strata sections have been simulated by computer methods. Validation studies of the method were addressed together with case studies. The interaction of caving with support convergence and face control is presented. The method allows the simulation of longwall support behaviour under various geological conditions. The system also allows a prediction of the monitoring data, which is best suited to give an early warning of weighting events or signal various key caving characteristics.

Jan 1, 1998

By R. H. Zeng

Many subsidence researchers in the U. S. have developed new empirical function methods to predict subsidence, or attempted to validate some empirical functions developed by foreign researchers for use in the U. S. An attempt is made in this paper to develop a new empirical function to predict a surface subsidence basin due to longwall mining. In the U. S., the emphasis of empirical methods for the subsidence prediction over longwall panels has been in large part on determining the profile functions for the major cross-sections of a subsidence basin. In this approach, the surface monuments are usually laid out along the major cross-sections and thus, the subsidence profiles developed are restricted for subsidence prediction along the major cross-sections. However in practice many monuments or surface structures for which subsidence prediction for damage evaluation is required, are not laid out along the major cross-section of the basin. In such cases, it is difficult to get either the empirical profile function for the cross-sections where these structures are located or to predict their subsidence based on the profile function developed for the major cross-sections. Therefore, there exists a strong need to develop a subsidence basin function with a reasonable degree of accuracy, which can be determined or confirmed by the data of scattered monuments on a surface basin. The same subsidence basin function can also be used for prediction of subsidence for any arbitrary points over a longwall panel. In 1982, He (1) proposed the clastic theory, for which he developed the basic differential equations describing the movement and deformation of overburden due to underground mining, and performed the selection of best solution for the equations. He found that the Weibull dietribution is the optimal solution. The Weibull distribution has three parameters that vary to fit the subsidence profile so well that the profile function derived from it provides a high degree of accuracy in subeidence prediction. However the results

Jan 1, 1986

By P. Myllykoski, J. Nylander, A. S. Korhonen, J. Larkiola

The coefficient of friction and the deformation resistance have been determined from the measured rolling parameters by applying the Bland-Ford-Ellis rolling force model and the artificial neural network model (ANN). The dependence of friction on the parameters such as reduction, rolling speed, front and back tensions, contact pressure and deformation resistance have been studied by the ANN model. Measured data of over 10 000 coils have been used in the training stage. Various dependencies have been examined and the results show that the friction coefficient does not depend much on the rolling speed. The reduction and deformation resistance seem to have the strongest effect on the friction coefficient. A friction model based on neural computing has been included into the existing rolling force model.

Jan 1, 1994

By A. McLean, W. Q. Chen, W. Yan, M. Barati, Y. D. Yang

In order to maximally remove the harmful impurities phosphorus and reduce the loss of valuable elements from Si-based alloy for production of solar-grade silicon, thermodynamic properties of the Si-based alloy were studied. In this work a new model, the Molecular Interaction Volume Model (MIVM) derived from statistical thermodynamics, was successfully used in the Si-based and Si-Fe-based alloy. The calculation started from binary Si-Fe, Si-P and Fe-P systems, then the thermodynamic parameters of ternary systems could be derived based sub-binary systems. Using the activity coefficients of the elements in the Si-based alloys and the saturated vapor pressure of the pure substance of the components, mass fraction of P during vapor-liquid equilibrium can be calculated. The calculated thermodynamic parameters were compared with experimental data available in different publications, and it is found that the thermodynamic parameters calculated using MIVM model are in good agreement with the measured results. The calculated result shows that the mass fraction of phosphorus during vapor-liquid equilibrium of Si-Fe-P melt increases with the increasing phosphorus content and decreasing temperature. This is an important direction for experimental practice in vacuum or low-pressure EML refining of ferrosilicon for potential production of solar-grade silicon without significant vaporization loss of valuable element.

Jan 1, 2015

By Xiaodong Yu, Jinxia Zhang, Fusheng Niu, Hongmei Zhang

The three-dimensional (3D) fractal dimension is an important parameter to analyze the 3D structure and flocculation effect of the hematite flocs. In this work, the 3D fractal dimension of hematite flocs was predicted by establishing the particle swarm optimization (PSO) algorithm optimized with the back propagation (BP) neural network (PSO-BP). The proposed model considers four factors, namely, the flocculant quantity, flocculation time, stirring speed, and temperature, as the input parameters. We normalize the input data during the preprocessing stage. The BP neural network was optimized by using the PSO algorithm for realizing the 3D fractal dimension. The prediction accuracy and effectiveness of the model were evaluated. The experimental results showed that the predictions performed by the proposed PSO-BP neural network were better than that of BP neural network. In addition, the root mean square error (RMSE), mean squared relative error (MSRE), mean absolute error (MAE), and mean absolute relative error (MARE) of the proposed model are lower, compared with the errors of the BP network. Similarly, the measurement coefficient R2 of the proposed model is higher, compared with the BP network. The maximum absolute error of the model is 0.0772, the maximum relative error is 0.02405, and the regression coefficient r is 0.98592. These results showed that the proposed model has a good performance and high prediction accuracy. When verifying the practicability and the effectiveness of the proposed PSO-BP model, the prediction results of 10 groups of hematite flocculation experimental data under different conditions showed that the prediction value of the proposed PSO-BP model was closer to the ground truth as compared to the BP model. Therefore, the proposed model is suitable for predicting the 3D fractal dimension of hematite flocculation.

Oct 3, 2022