Prediction of local scour around complex piers using GEP and M5-Tree

Local scour around piers is one of the main causes of bridge failures. In this study, three robust techniques, artificial neural networks (ANNs), M5-Tree, and Gene Expression Programming (GEP), were employed for prediction of scour depth around complex piers. The clear water condition was chosen for all experimental tests. The results indicated that pier diameter (b _c) and foundation level (Y) are the main parameters for local scour. Furthermore, the minimum scour depth occurs in range of Y/b _c = 1.1~1.3. In next step, to evaluate the mentioned techniques, a wide range of dataset was collected from the present study and literature. The radial base function (RBF) with R ² = 0.945 and RMSE = 0.031 provides better prediction in comparison with conventional equations, M5-Tree (R ² = 0.883, RMSE = 0.292) and the GEP techniques (R ² = 0.811 and RMSE = 0.263). The equations developed by M5-Tree and GEP are more useful for practical purposes and can be easily employed to predict the depth of scour at complex piers.     

Effect of the form of data on the quality of mine tremors hazard forecasting using neural networks

During hard coal mining operations conducted under conditions of rockburst hazard, one of the most important preventive measures can be the prediction of occurrence time and location of the strong seismic mine tremors of energy E _s ⩾ 10⁴ J. This is a very difficult task and the way it is being currently performed appears to be unsatisfactory. Therefore, attempts have been made to use neural networks, specifically trained for this application. The paper presents an approach for determining an influence of the type and shape of the input data on the efficiency of such a prediction. The considerations are based on a selected example of the seismic activity recorded during longwall mining operations conducted in one of the Polish mines.     

Artificial neural networks approach for zeta potential of Montmorillonite in the presence of different cations

In this study, the zeta potential of montmorillonite in the presence of different chemical solutions was modeled by means of artificial neural networks (ANNs). Zeta potential of the montmorillonite was measured in the presence of salt cations, Na⁺, Li⁺ and Ca²⁺ and metals Zn²⁺, Pb²⁺, Cu²⁺, and Al³⁺ at different pH values, and observed values pointed to a different behavior for this mineral in the presence of salt and heavy metal cations. Artificial neural networks were successfully developed for the prediction of the zeta potential of montmorillonite in the presence of salt and heavy metal cations at different pH values and ionic strengths. Resulting zeta potential of montmorillonite shows different behavior in the presence of salt and heavy metal cations, and two ANN models were developed in order to be compared with experimental results. The ANNs results were found to be close to experimentally measured zeta potential values. The performance indices such as coefficient of determination, root mean square error, mean absolute error, and variance account for were used to control the performance of the prediction capacity of the models developed in this study. These indices obtained make it clear that the predictive models constructed are quite powerful. The constructed ANN models exhibited a high performance according to the performance indices. This performance has also shown that the ANNs seem to be a useful tool to minimize the uncertainties encountered during the soil engineering projects. For this reason, the use of ANNs may provide new approaches and methodologies.     

Application of neural network and ANFIS model for earthquake occurrence in Iran

This study examined the spatial-temporal variations in seismicity parameters for the September 10^th, 2008 Qeshm earthquake in south Iran. To this aim, artificial neural networks and Adaptive Neural Fuzzy Inference System (ANFIS) were applied. The supervised Radial Basis Function (RBF) network and ANFIS model were implemented because they have shown the efficiency in classification and prediction problems. The eight seismicity parameters were calculated to analyze spatial and temporal seismicity pattern. The data preprocessing that included normalization and Principal Component Analysis (PCA) techniques was led before the data was fed into the RBF network and ANFIS model. Although the accuracy of RBF network and ANFIS model could be evaluated rather similar, the RBF exhibited a higher performance than the ANFIS for prediction of the epicenter area and time of occurrence of the 2008 Qeshm main shock. A proper training on the basis of RBF network and ANFIS model might adopt the physical understanding between seismic data and generate more effective results than conventional prediction approaches. The results of the present study indicated that the RBF neural networks and the ANFIS models could be suitable tools for accurate prediction of epicenteral area as well as time of occurrence of forthcoming strong earthquakes in active seismogenic areas.     

Application of an Expert System to Predict Maximum Explosive Charge Used Per Delay in Surface Mining

The present paper mainly deals with the prediction of maximum explosive charge used per delay (Q _MAX) using an artificial neural network (ANN) incorporating peak particle velocity (PPV) and distance between blast face to monitoring point (D). One hundred and fifty blast vibration data sets were monitored at different vulnerable and strategic locations in and around major coal producing opencast coal mines in India. One hundred and twenty-four blast vibrations records were used for the training of the ANN model vis-à-vis to determine site constants of various conventional vibration predictors. The other 26 new randomly selected data sets were used to test, evaluate and compare the ANN prediction results with widely used conventional predictors. Results were compared based on coefficient of correlation (R), mean absolute error and mean squared between measured and predicted values of Q _MAX. It was found that coefficient of correlation between measured and predicted Q _MAX by ANN was 0.985, whereas it ranged from 0.316 to 0.762 by different conventional predictor equations. Mean absolute error and mean squared error was also very small by ANN, whereas it was very high for different conventional predictor equations.     

Intelligent methods for predicting nuclear magnetic resonance of porosity and permeability by conventional well-logs: a case study of Saharan field

In the well-log data processing, the principal advantage of the nuclear magnetic resonance (NMR) method is the measurement of fluid volume and pore size distribution without resorting to parameters such as rock resistivity. Preliminary processing of the well-log data allowed first to have the petrophysical parameters and then to evaluate the performances of the transverse relaxation time T ₂ NMR. Petrophysical parameters such as the porosity of the formation as well as the effective permeability can be estimated without having recourse the fluid type. The well-log data of five wells were completed during the construction of intelligent models in the Saharan oil field Oued Mya Basin in order to assess the reliability of the developed models. Data processing of NMR combined with conventional well data was performed by artificial intelligence. First, the support vector regression method was applied to a sandy clay reservoir with a model based on the prediction of porosity and permeability. NMR parameters estimated using intelligent systems, i.e., fuzzy logic (FL) model, back propagation neural network (BP-NN), and support vector machine, with conventional well-log data are combined with those of NMR, resulting in a good estimation of porosity and permeability. The results obtained during the processing are then compared to the FL and NN regression models performed by the regression method during the validation stage. They show that the correlation coefficients R ² estimated vary between 0.959 and 0.964, corresponding to the root mean square error values of 0.20 and 0.15.     

Digital soil mapping in a Himalayan watershed using remote sensing and terrain parameters employing artificial neural network model

Digital soil mapping relies on field observations, laboratory measurements and remote sensing data, integrated with quantitative methods to map spatial patterns of soil properties. The study was undertaken in a hilly watershed in the Indian Himalayan region of Mandi district, Himachal Pradesh for mapping soil nutrients by employing artificial neural network (ANN), a potent data mining technique. Soil samples collected from the surface layer (0–15 cm) of 75 locations in the watershed, through grid sampling approach during the fallow period of November 2015, were preprocessed and analysed for various soil nutrients like soil organic carbon (SOC), nitrogen (N) and phosphorus (P). Spectral indices like Colouration Index, Brightness Index, Hue Index and Redness Index derived from Landsat 8 satellite data and terrain parameters such as Terrain Wetness Index, Stream Power Index and slope using CartoDEM (30 m) were used. Spectral and terrain indices sensitive to different nutrients were identified using correlation analysis and thereafter used for predictive modelling of nutrients using ANN technique by employing feed-forward neural network with backpropagation network architecture and Levenberg–Marquardt training algorithm. The prediction of SOC was obtained with an R² of 0.83 and mean squared error (MSE) of 0.05, whereas for available nitrogen, it was achieved with an R² value of 0.62 and MSE of 0.0006. The prediction accuracy for phosphorus was low, since the phosphorus content in the area was far below the normal P values of typical Indian soils and thus the R² value observed was only 0.511. The attempts to develop prediction models for available potassium (K) and clay (%) failed to give satisfactory results. The developed models were validated using independent data sets and used for mapping the spatial distribution of SOC and N in the watershed.     

Burden prediction in blasting operation using rock geomechanical properties

Burden prediction is a vital task in the production blasting. Both the excessive and insufficient burden can significantly affect the result of blasting operation. The burden which is determined by empirical models is often inaccurate and needs to be adjusted experimentally. In this paper, an attempt was made to develop an artificial neural network (ANN) in order to predict burden in the blasting operation of the Mouteh gold mine, using considering geomechanical properties of rocks as input parameters. As such here, network inputs consist of blastability index (BI), rock quality designation (RQD), unconfined compressive strength (UCS), density, and cohesive strength. To make a database (including 95 datasets), rock samples are used from Iran’s Mouteh goldmine. Trying various types of the networks, a neural network, with architecture 5-15-10-1, was found to be optimum. Superiority of ANN over regression model is proved by calculating. To compare the performance of the ANN modeling with that of multivariable regression analysis (MVRA), mean absolute error (E _a), mean relative error (E _r), and determination coefficient (R ²) between predicted and real values were calculated for both the models. It was observed that the ANN prediction capability is better than that of MVRA. The absolute and relative errors for the ANN model were calculated 0.05 m and 3.85%, respectively, whereas for the regression analysis, these errors were computed 0.11 m and 5.63%, respectively. Moreover, determination coefficient of the ANN model and MVRA were determined 0.987 and 0.924, respectively. Further, a sensitivity analysis shows that while BI and RQD were recognized as the most sensitive and effective parameters, cohesive strength is considered as the least sensitive input parameters on the ANN model output effective on the proposed (burden).     

Earthquake risk assessment in NE India using deep learning and geospatial analysis

Earthquake prediction is currently the most crucial task required for the probability, hazard, risk mapping, and mitigation purposes. Earthquake prediction attracts the researchers' attention from both academia and industries. Traditionally, the risk assessment approaches have used various traditional and machine learning models. However, deep learning techniques have been rarely tested for earthquake probability mapping. Therefore, this study develops a convolutional neural network (CNN) model for earthquake probability assessment in NE India. Then conducts vulnerability using analytical hierarchy process (AHP), Venn's intersection theory for hazard, and integrated model for risk mapping. A prediction of classification task was performed in which the model predicts magnitudes more than 4 Mw that considers nine indicators. Prediction classification results and intensity variation were then used for probability and hazard mapping, respectively. Finally, earthquake risk map was produced by multiplying hazard, vulnerability, and coping capacity. The vulnerability was prepared by using six vulnerable factors, and the coping capacity was estimated by using the number of hospitals and associated variables, including budget available for disaster management. The CNN model for a probability distribution is a robust technique that provides good accuracy. Results show that CNN is superior to the other algorithms, which completed the classification prediction task with an accuracy of 0.94, precision of 0.98, recall of 0.85, and F1 score of 0.91. These indicators were used for probability mapping, and the total area of hazard (21,412.94 km²), vulnerability (480.98 km²), and risk (34,586.10 km²) was estimated.     

广义回归神经网络预测加筋土支挡结构高度

土工合成材料加筋支挡结构（Geosythetics-Reinforced Retaining Wall, 简称GRW）设计方法主要是建立在似粘聚力理论基础之上的半经验设计法。由于土性及加筋机理的复杂性，常常要对它们进行人为假定，导致计算结果差强人意。神经网络方法与传统方法的不同之处在于不需要主观假定，而是模拟人脑思维，通过数据样本的学习来获得预测结果。引入神经网络技术来预测加筋土支挡结构的设计高度是一种新尝试。由于本问题具有样本容量非常有限、影响因素复杂多样的特点。因此，采用适用于稀土样本数据的广义回归网络（General Regression Neural Network)来预测加筋土支挡结构设计高度。基于MATLAB神经网络工具箱及文献[1]的挡墙离心模型试验结果，建立了一个可用于加筋支挡结构设计高度预测的GRNN网络。通过对足尺试验，实际工程及模型试验结果的检验，表明网络的学习是成功的，具有一定指导意义。     

Extreme carbon monoxide pollution of the atmospheric boundary layer in Moscow region in the summer of 2010

Using data of spectroscopic measurements in Moscow and Moscow region and data of ecological monitoring from Ostankino TV Tower, it has been found that, in the period of intense smoke blanketing of the atmosphere in summer 2010 due to large-scale forest-peat fires on the territory of the European part of the Russian Federation, the carbon monoxide content in the boundary layer and in the atmospheric depth of Moscow region reached the extremely high level of 8 g/m², or 17 ×10¹⁸ mol/cm², and the carbon monoxide concentration in the near-ground atmospheric layer increased to 37.5 mg/m³, i.e., an unprecedentedly large value for Moscow and more than a factor of 7 larger than the one-time maximum permissible concentration for carbon monoxide MPC_ot = 5 mg/m³. In the first decade of August, when intense smoke blanketing of the Moscow region was observed, the average carbon monoxide concentrations varied in the range from approximately 3 to 7 m/g³, i.e., an order of magnitude larger than the annual average concentrations calculated according to data of measurements in 2009. The probabilities of exceeding MPC_ot and multiples of MPC_ot are calculated. Analysis of data of thermal sensing showed that an important feature of the atmospheric boundary layer over the city was a high stability of the lower atmospheric layer with the thickness of 150–200 m, and also long-term nighttime and morning inversions of the air temperature in this layer.     

The influence of wind speed on surface layer stability and turbulent fluxes over southern Indian peninsula station

Surface to atmosphere exchange has received much attention in numerical weather prediction models. This exchange is defined by turbulent parameters such as frictional velocity, drag coefficient and heat fluxes, which have to be derived experimentally from high-frequency observations. High-frequency measurements of wind speed, air temperature and water vapour mixing ratio (eddy covariance measurements), were made during the Integrated Ground Observation Campaign (IGOC) of Cloud Aerosol Interaction and Precipitation Enhancement Experiment (CAIPEEX) at Mahabubnagar, India (16^°44^′N, 77^°59^′E) in the south-west monsoon season. Using these observations, an attempt was made to investigate the behaviour of the turbulent parameters, mentioned above, with respect to wind speed. We found that the surface layer stability derived from the Monin–Obukhov length scale, is well depicted by the magnitude of wind speed, i.e., the atmospheric boundary layer was under unstable regime for wind speeds >4 m s^?1; under stable regime for wind speeds <2 m s^?1 and under neutral regime for wind speeds in the range of 2–3 m s^?1. All the three stability regimes were mixed for wind speeds 3–4 m s^?1. The drag coefficient shows scatter variation with wind speed in stable as well as unstable conditions.     

Predicting longitudinal dispersion coefficient using ANN with metaheuristic training algorithms

A reliable prediction of dispersion coefficient can provide valuable information for environmental scientists and river engineers as well. The main objective of this study is to apply intelligence techniques for predicting longitudinal dispersion coefficient in rivers. In this regard, artificial neural network (ANN) models were developed. Four different metaheuristic algorithms including genetic algorithm (GA), imperialist competitive algorithm (ICA), bee algorithm (BA) and cuckoo search (CS) algorithm were employed to train the ANN models. The results obtained through the optimization algorithms were compared with the Levenberg–Marquardt (LM) algorithm (conventional algorithm for training ANN). Overall, a relatively high correlation between measured and predicted values of dispersion coefficient was observed when the ANN models trained with the optimization algorithms. This study demonstrates that the metaheuristic algorithms can be successfully applied to make an improvement on the performance of the conventional ANN models. Also, the CS, ICA and BA algorithms remarkably outperform the GA and LM algorithms to train the ANN model. The results show superiority of the performance of the proposed model over the previous equations in terms of DR, R ² and RMSE.     

Assessment of a conceptual hydrological model and artificial neural networks for daily outflows forecasting

Artificial neural networks (ANNs) are used by hydrologists and engineers to forecast flows at the outlet of a watershed. They are employed in particular where hydrological data are limited. Despite these developments, practitioners still prefer conventional hydrological models. This study applied the standard conceptual HEC-HMS’s soil moisture accounting (SMA) algorithm and the multi layer perceptron (MLP) for forecasting daily outflows at the outlet of Khosrow Shirin watershed in Iran. The MLP [optimized with the scaled conjugate gradient] used the logistic and tangent sigmoid activation functions resulting into 12 ANNs. The R ² and RMSE values for the best trained MPLs using the tangent and logistic sigmoid transfer function were 0.87, 1.875 m³ s^?1 and 0.81, 2.297 m³ s^?1, respectively. The results showed that MLPs optimized with the tangent sigmoid predicted peak flows and annual flood volumes more accurately than the HEC-HMS model with the SMA algorithm, with R ² and RMSE values equal to 0.87, 0.84 and 1.875 and 2.1 m³ s^?1, respectively. Also, an MLP is easier to develop due to using a simple trial and error procedure. Practitioners of hydrologic modeling and flood flow forecasting may consider this study as an example of the capability of the ANN for real world flow forecasting.     

Prediction of residual friction angle of clays using artificial neural network

The residual strength of clay is very important to evaluate long term stability of proposed and existing slopes and for remedial measure for failure slopes. Various attempts have been made to correlate the residual friction angle (_r) with index properties of soil. This paper presents a neural network model to predict the residual friction angle based on clay fraction and Atterberg's limits. Different sensitivity analysis was made to find out the important parameters affecting the residual friction angle. Emphasis is placed on the construction of neural interpretation diagram, based on the weights of the developed neural network model, to find out direct or inverse effect of soil properties on the residual shear angle. A prediction model equation is established with the weights of the neural network as the model parameters.     

Estimation of Hydraulic Conductivity and Its Uncertainty from Grain-Size Data Using GLUE and Artificial Neural Networks

Various approaches exist to relate saturated hydraulic conductivity (K _s) to grain-size data. Most methods use a single grain-size parameter and hence omit the information encompassed by the entire grain-size distribution. This study compares two data-driven modelling methods??multiple linear regression and artificial neural networks??that use the entire grain-size distribution data as input for K _s prediction. Besides the predictive capacity of the methods, the uncertainty associated with the model predictions is also evaluated, since such information is important for stochastic groundwater flow and contaminant transport modelling. Artificial neural networks (ANNs) are combined with a generalised likelihood uncertainty estimation (GLUE) approach to predict K _s from grain-size data. The resulting GLUE-ANN hydraulic conductivity predictions and associated uncertainty estimates are compared with those obtained from the multiple linear regression models by a leave-one-out cross-validation. The GLUE-ANN ensemble prediction proved to be slightly better than multiple linear regression. The prediction uncertainty, however, was reduced by half an order of magnitude on average, and decreased at most by an order of magnitude. This demonstrates that the proposed method outperforms classical data-driven modelling techniques. Moreover, a comparison with methods from the literature demonstrates the importance of site-specific calibration. The data set used for this purpose originates mainly from unconsolidated sandy sediments of the Neogene aquifer, northern Belgium. The proposed predictive models are developed for 173 grain-size K _s-pairs. Finally, an application with the optimised models is presented for a borehole lacking K _s data.     

Combining neural network with fuzzy,certainty factor and likelihood ratio concepts for spatial prediction of landslides

The landslide studies can be categorized as pre- and postdisaster studies. The predisaster studies include spatial prediction of potential landslide zones known as landslide susceptibility zonation (LSZ) mapping to identify the areas/locales susceptible to landslide hazard. The LSZ maps provide an assessment of the safety of existing habitations and infrastructural/functional elements and help plan further developmental activities in the hilly regions. Landslides are one of the natural geohazards that affect at least 15% of land area of India. Different types of landslides occur frequently in geodynamical active domains of the Himalayas. In India, various techniques have been developed and adopted for LSZ mapping of different regions. However, the technique for LSZ mapping is not yet standardized. The present research is an attempt in this direction only. In our earlier work (Kanungo et al. 2006), a detailed study on conventional, artificial neural network (ANN)- black box-, fuzzy set-based and combined neural and fuzzy weighting techniques for LSZ mapping in Darjeeling Himalayas has been documented. In this paper, other techniques such as combined neural and certainty factor concept along with combined neural and likelihood ratio techniques have been assessed in comparison with combined neural and fuzzy technique for the preparation of LSZ maps of the same study area in parts of Darjeeling Himalayas. It is observed from the present study that the LSZ map produced using combined neural and fuzzy approach appears to be the most accurate one as in this case only 2.3% of the total area is found to be categorized as very high susceptibility zone and contains 30.1% of the existing landslide area. This approach can serve as one of the key objective approaches for spatial prediction of landslide hazards in hilly terrain.     

An intelligent neural network model for evaluating performance of immobilized cell biofilter treating hydrogen sulphide vapors

Biofiltration has shown to be a promising technique for handling malodours arising from process industries. The present investigation pertains to the removal of hydrogen sulphide in a lab scale biofilter packed with biomedia, encapsulated by sodium alginate and poly vinyl alcohol. The experimental data obtained under both steady state and shock loaded conditions were modelled using the basic principles of artificial neural networks. Artificial neural networks are powerful data driven modelling tools which has the potential to approximate and interpret complex input/output relationships based on the given sets of data matrix. A predictive computerised approach has been proposed to predict the performance parameters namely, removal efficiency and elimination capacity using inlet concentration, loading rate, flow rate and pressure drop as the input parameters to the artificial neural network model. Earlier, experiments from continuous operation in the biofilter showed removal efficiencies from 50 to 100 % at inlet loading rates varying up to 13 g H₂S/m³h. The internal network parameter of the artificial neural network model during simulation was selected using the 2^k factorial design and the best network topology for the model was thus estimated. The results showed that a multilayer network (4-4-2) with a back propagation algorithm was able to predict biofilter performance effectively with R² values of 0.9157 and 0.9965 for removal efficiency and elimination capacity in the test data. The proposed artificial neural network model for biofilter operation could be used as a potential alternative for knowledge based models through proper training and testing of the state variables.     

基于贝叶斯正则化改进BP神经网络的页岩气有机碳含量预测模型

页岩气总有机碳(TOC)含量是评价岩性气藏的关键指标,受复杂地质及岩芯采集等多种因素的影响,常规室内测试分析获得的TOC含量的数据有限且结果有失准确。为合理准确预测页岩气TOC含量,本文首先通过对页岩气储层TOC含量测井资料综合分析选取8条测井曲线,并结合主成分分析法(Principal Component Analysis,PCA)提取四个主成分;其次基于贝叶斯正则化(Bayesian Regularization)改进的BP神经网络方法建立页岩气TOC含量预测的BR-BP模型;最后利用该模型对研究区A区页岩气TOC含量进行预测,并与常规的LM-BP神经网络模型的预测结果进行对比。结果表明：BR-BP模型有较强的非线性拟合能力,能够真实地反映出页岩气TOC含量与各测井参数之间的非线性关系,其模型预测结果与实际值基本吻合,与常规的LM-BP神经网络模型相比,其数据敏感性增强,预测精度有所提高,该研究方法具有一定的理论意义和参考价值,为我国TOC含量预测提供了一种新的技术方法和手段。