Yaser Bageri; Esfandiar Abbas Novinpour; A Nadiri; Keiwan Naderi
Abstract
Most of the country's geographically area is located in dry and semi-dry zone with low rainfall. The growing population, the limitation of water resources and the prevalence of groundwater resources in most parts of the country requirement to accurate prediction of the amount of these resources due to ...
Read More
Most of the country's geographically area is located in dry and semi-dry zone with low rainfall. The growing population, the limitation of water resources and the prevalence of groundwater resources in most parts of the country requirement to accurate prediction of the amount of these resources due to the importance of these resources in optimal planning and management. In this research, in order to estimate the fluctuations of groundwater level in the Baruq aquifer, the artificial intelligence models including fuzzy, support vector machine and neural network models were used by the data of depth from 7 piezometers with long-term data of 14 years, as well as changes in temperature and precipitation in this period. Despite the inherent abilities of each models in predicting groundwater level, the heterogeneity of the study area prevented the high efficiency of these models. Therefore, SOM-AI modeling combined the self-organized maps (SOM) classification method and each model that is increased the efficiency of each composite model in different parts of the aquifer by dividing the study area into homogeneous regions. The results showed that the proposed method can be an effective method in the modeling of heterogeneous and even multi-layered aquifers.
N. Alidadi; A. Mahdavian
Abstract
When seismic waves pass through alluvial layers, the seismic wave amplitude increases significantly in some periods, which is known as site amplification. In this case, it can be analyzed with an analytical model of the surface response spectrum Estimates of the input response spectrum. This behavior ...
Read More
When seismic waves pass through alluvial layers, the seismic wave amplitude increases significantly in some periods, which is known as site amplification. In this case, it can be analyzed with an analytical model of the surface response spectrum Estimates of the input response spectrum. This behavior is essential in assessing the seismic performance of the structures and vital arteries. In this paper, we investigate this effect on different layers of sand with different thicknesses for Urmia city, a metropolitan area in the North West of Iran and an earthquake prone region. 120 geotechnical boreholes were drilled in sandy layers of different thicknesses in the central and northern parts of the city. The analysis simulated artificial acceleration based on seismic hazard analysis results. By observing the spectral accelerations for different periods at the earth’s surface the spectral acceleration magnitudes of the seams were shown for different periods. Using a neural network and genetic algorithm, these coefficients were modelled. Using the evolutionary algorithm of gene expression programming, the mathematical relation was expressed in terms of sand layer thickness and different periods. However, the results obtained from the artificial neural network using the correlation coefficient and root mean square yielded more accurate results than the gene expression programming. In conclusion, the results show that by increasing the thickness of the sand layers, the amplification ratio also increases considerably for some periods. Using the modeling results, we can estimate the amplification of the sandy soils of Urmia city with different thicknesses for variables up to 4 seconds.
M.S Mirkamali; H.R Ramazi; M.R Bakhtiari; H Ramesh
Abstract
This study has focused on identifying fault systems in the HormuzStrait area using compilation of seismic attributes and artificial neural networks. Faults and fractures play an important role in creating areas of high porosity and permeability. In addition, they cut off the cap and reservoir rocks along ...
Read More
This study has focused on identifying fault systems in the HormuzStrait area using compilation of seismic attributes and artificial neural networks. Faults and fractures play an important role in creating areas of high porosity and permeability. In addition, they cut off the cap and reservoir rocks along fluid migration pathways. Intense tectonic activities and salt tectonics have resulted in complex structures in the Strait of Hormuz area. Therefore, precise identification of faults and fracture zones and their extensions has special importance in increasing petroleum production from traps. In order to identify the geometry and kinematics of faults in the Mishan and Aghajari Formations and in the units under the base-Guri unconformity in the HormuzStrait area (eastern part of the Persian Gulf), we have used structural imaging and visualization techniques of seismic interpretation. The structural imaging of the fault zones was obtained by this technique based on the integration of input attributes in an artificial neural network system and creating new attributes. First, a set of advanced attributes were introduced as input for the artificial neural network system to train and compile the calculated attributes on fault and non-fault interpreted points. As a powerful exploration tool, finally, the fault cube was obtained to precisely identify fault systems and better detect faults and fractures in quantitative modeling of the area. As a result of integrated attributes, the high correlation between the faults within the fault cube provides more accurate and reliable tracking of fault extensions. Therefore, three types of fault systems were identified in study area, which are thought to be results of the extensional and compressional tectonics of the Oman Orogeny, vertical tectonic movements of the Zagros Orogeny, and syn-sedimentary salt movements.
A Abbaszadeh shahri; R Hosseini; F Rezaei; K Mehdizadeh; N Panaei
Abstract
Artificial Neural Network methods (ANN) are computational methods, which capable to predict a specific log or classify different data. Unlike the digital computers, which require the completely definite and distinguished rules, the ANN methods do not need a pure mathematical model; rather like the human ...
Read More
Artificial Neural Network methods (ANN) are computational methods, which capable to predict a specific log or classify different data. Unlike the digital computers, which require the completely definite and distinguished rules, the ANN methods do not need a pure mathematical model; rather like the human brain has the ability to learn by recognized and determined examples. The target of the present paper is to establish and prove the Petrophysical Analysis as powerful approach in prediction and diagnosis of rock reservoir porosity by use of petrophysical logs, in which by a high accuracy suggested Petrophysical Analysis based solution the porosity can be estimated using conventional logging data. On the basis of the available petrophysical data, the proposed method was examined in one of the southwest oil field of Iran. The obtained results of network analysis conditioning to reliability to data with different tests such as regression, root mean square and SPLine showed that the amount of network error in terms of available data in engineering range with a high acceptable safety factor could be used to predict and estimate porosity. This method with ability of cost reduction and viability can help and provide a large variety in this field for further extended research.
M. forutan; A. mansourian; M. Zareinejad; M. R. Sahebi
Abstract
Drilling in mine deposits has proven to be complicated, costly and time consuming process, hence it has identified the determining of optimum boreholes as a crucial issue in detailed studies. Because of some complexity in formation of mineral deposits, decreasing in risk and expenses of drilling ...
Read More
Drilling in mine deposits has proven to be complicated, costly and time consuming process, hence it has identified the determining of optimum boreholes as a crucial issue in detailed studies. Because of some complexity in formation of mineral deposits, decreasing in risk and expenses of drilling may be continued by considering the wrapped condition of mineral deposits formation, followed by the integration of effective mineralization factors. By considering that in traditional methods of combination of mineralization’s factors like overlay and index overlay, is based on expert’s knowledge and expert knowledge is related to data accuracy, therefore, the accuracy of these methods could be remarkably decreased by large amount of data and noise. In order to solve these problems, utilization of flexible methods and powerful tools in data processing is obviously needed, especially in case of noise presence. Artificial Neural Networks are appropriate tools in large amount of data management and pattern recognition of noisy data, because of nonlinear, parallel and flexible architecture. So ANNs has been used in determining of proper position of boreholes. Neural Networks have various structures regarding their activation function and number of hidden layer and neurons in each layer. Consequently it is necessary to examine the performance of all these structures in determining the optimum position of boreholes.This paper represents a study on utilization GIS and four different Neural Networks namely: Multilayer peceptrons, Radial Basis Function, Generalized Neural Network and Probability Neural Network, for determining the position of boreholes of porphyry copper exploration in Chahfirouzeh region using cross correlation method. First, the mineralization factors are explained based on conceptual model of porphyry copper and predictor maps are produced, then, the training vectors are derived. After that, the networks are trained by geology, geochemistry and geophysics data layers. At the end, performances of the networks are compared. Implementation of Artificial Neural Networks reveals that two Neural Networks, GRNN and RBF, have the highest accuracy (approximately 80 to 83 %). Eventually, a potential map is produced by the best method.
P. Tahmasbi; A. Hezarkhani
Abstract
In the present research, comparative evaluation of various learning algorithms in neural network modeling was performed for ore grade estimation in Sonjail porphyry copper deposit. The main goal of the following investigation would be optimizing the network architecture and to present an architectural ...
Read More
In the present research, comparative evaluation of various learning algorithms in neural network modeling was performed for ore grade estimation in Sonjail porphyry copper deposit. The main goal of the following investigation would be optimizing the network architecture and to present an architectural optimization trend to better performing the copper grade estimation within the region. Therefore, 12 algorithms were investigated back propagation learning algorithms. Based on this research it is merged that by applying the LM and BFG algorithms, there would be the best performance. The reasons why the other algorithms have the same performance would be presented within the paper as well. The input parameters are coordinates and the outputs are the copper grades for each specified point. To obtain the optimal structure, a network with different layers has been applied, which it has acquired 12 neurons within one layer. To investigate the data normal shapes, various normal shape has been acquired in the [0 1], which could merged the best results. Finally to get the best network optimizations several transfer functions have been applied, and the sigmoid transfer function illustrated least error when the transfer function is selected. Considering the optimal conditions, the R2 value has merged 0.946 for network which could be the result showing that the optimal network architecture causes estimation improvement.
Ahmad Zamani; M. Nedaei
Abstract
One of the basic discussions in geosciences is construction of different tectonic zoning maps. In conventional tectonic zoning, not only the great amounts of subjective judgment are involved but also accurate interpretation of high-dimensional data is so difficult and out of human capability. To alleviate ...
Read More
One of the basic discussions in geosciences is construction of different tectonic zoning maps. In conventional tectonic zoning, not only the great amounts of subjective judgment are involved but also accurate interpretation of high-dimensional data is so difficult and out of human capability. To alleviate these deficiencies, quantitative scientific methods in data mining domain can be applied as an effective and useful tool to construct the new numerical maps in geosciences. In this paper self-organizing map (SOM) neural network that is one of the common methods in data mining has been applied for numerical tectonic zoning of Iran. SOM is an unsupervised artificial neural network particularly adept at pattern recognition and clustering of high-dimensional data. Visualization of high-dimensional data in two-dimensional topological-preserving feature map is another specific capability of SOM that represent both homogeneity within and similarity between clusters. Although there are some similarities between SOM's numerical maps constructed here and the conventional maps but SOM method is more powerful for identification and interpretation of different zones than conventional methods. Utilizing SOM method enables us not only to evaluate the degree of homogeneity in each zone, but also to separate regions zone that experience similar geological evolutionary despite of their geographical locations. For instance Lut and Gavkhuni zones show more homogeneity than Makran and Azerbayejan zones also Kopeh-Dagh and Zagros are located at different regions, they have similar features. The results obtained here represent separation between Makran from EastIranianRanges and Western Azerbaijan from AlborzRanges, too. It is important to recognize that the SOM's results are based purely on the geophysical, geological and seismic features presented previously. So correspondences and differences between the SOM's zones and a given zone based on conventional method must receive careful thought.
M. J. Mohammadzadeh; H. Aghababaei; A. Naseri
Abstract
The amount of total organic carbon (TOC) is one of the most important parameter in evaluating hydrocarbon source rock. This parameter is not only used for hydrocarbon geochemical studies but also plays an important role in evaluating the extension of hydrocarbon source rock. ...
Read More
The amount of total organic carbon (TOC) is one of the most important parameter in evaluating hydrocarbon source rock. This parameter is not only used for hydrocarbon geochemical studies but also plays an important role in evaluating the extension of hydrocarbon source rock. As the increase in TOC may indicate the presence of source rock, the depletion of TOC reveals no extension of source rock in a certain depth. Therefore the need for a powerful tool in this aspect is essential. One of the linear methods for solving such problem is artificial neural network, a biologically inspired computing method which has an ability to learn; self adjusted and are trained, capable of classification, image processing and different problem analysis, with an attempt to estimate. This paper presents the features and framework for application of neural network in estimating TOC for hydrocarbon source rock in Binak oil field, Bushehr province, using well log data. The results of this study reveal that Multi-Layer Perception (MLP) is the optimum network which was used for TOC estimation. MLP topology was a hidden layer with 6 nodes, back propagation momentum learning algorithm and tangent activation function. After training is completed, the estimated error calculated as 0.0013, and then the network performance was tested upon training and testing data. Ultimately the predicted TOC values were compared with the actual one which showed a reliable network performance (R=0.9956). Finally the sensitivity analysis was attempted on effective parameters and based on neutron porosity parameter (NPHI) found to be as the most sensitive, and the sonic travel time (DT), the least sensitive parameters in estimating TOC.
