Journal of Engineering Geology

---

(Paper pages 179-192) Artificial Neural Network (ANN), has many abilities which have increade its application in different fields of engineering and geosciences. In this paper, the application of ANN in geological engineering(prediction of landslide hazard) in Talesh area, north of Iran, is evaluated. The results are shown that, the system is able to process input data by selecting effective parameters of landslide and give the landslide hazard potential as a ANN output. By considering the landslide hazard zonation map of the area and by using the ANN system, it becomes clear that, the Talesh area is a landslide hazard prone area. The most effective factors of slope instability of the area, are land use and land cover conditions, ground water and surface water effects, river erosion and tectonics activities.

---

(Paper pages 649-676) Engineering characteristics of alluvium and cemented materials of the slopes around the Milad Tower, and the results of slopes stability analyses under static condition is presented in this paper. Also in the paper, the feasibility of developing and using artificial neural networks (ANNS) for slope stability prediction is investigated. According to the geometry of slopes and strength and deformation properties of alluviums, factor of safety is calculated in 2D and 3D by PLAXIS7.2 and PLAXIS 3D Tunnel codes, respectively, and the results are also compared. In addition, stability of slopes is investigated through the use of MLP artificial neural networks (ANNs), which developed in MATLAB environment. The database used for development of the model comprises a series of 252 factor of safety for different slopes conditions (2D, 3D, flatted and 18 inclined from horizon at top of cut). The optimal ANN architecture (hidden nodes, transfer functions and training) is obtained by a trial-and-error approach in accordance to error indexes and real data. The input data for slope stability estimation consist of values of geotechnical and geometrical input parameters. As an output, the network estimates the factor of safety (FoS). The results indicate that the ANN model is able to accurately predict the FoS of the slopes.

---

Tunnel boring machines (TBM) are widely used in excavating urban tunnels. These kinds of machines have different types based on supporting faces and tunnel walls. One type of these machines, is the Earth Pressure Balance (EPB) type that was used in excavating the Line 1 Tunnel of Tabriz Metro. Different parameters such as geological conditions, rock mass properties, dip and machine specifications affect the efficiency of the machine. One method of predicting the efficiency of these machines is to estimate their penetration rates. In this study the value of TBM penetration rates are predicted by an artificial neural network. Predicting of this parameter is so effective for conducting in high risk regions by understanding the time of facing to these regions. The main result of this study is to forecast the penetration rate with an acceptable accuracy and to determine the effective parameters through sensitivity analysis measured by an artificial neural network.

---

The Keshvari watershed is located at south east of Khorramabad city in Lorestan province. This area is one part of the folded Zagros zone based on structural geology classification. By consider the type of geological formations, topographic conditions and its area, this watershed is very unstable and capable for occurring landslide. In this study, artificial neural network (ANN) with structure of multi-layer percepteron and Back Propagation learning algorithm used for zonation of landslide risk. The results of ANN showed the final structure of 9-11-1 for zonation of landslide risk in Keshvari watershed. According this zonation, 23.81, 7.53, 6.49, 18.68 and 43.47 percent of area are located in very low, low, moderate, high and very high risk classes, respectively.

---

Prediction of location of future earthquakes with event probability is useful in reduction of earthquake hazard. Determination of predicted locations has attracted more attention to design, seismic rehabilitation and reliability of structures in these sites. Many theories were proposed in the prediction of time of occurrence of earthquake. There is not a method for prediction time of future earthquakes. Many studies have been done in the prediction of magnitude of earthquakes, but there are not any investigations on prediction of earthquake hazard zonation. In this study, the locations that have probability of the event of future earthquake have been predicted by artificial neural networks in Qum and Semnan. Neural networks used in this study can extract to complicate properties of patterns by receipting the interval patterns. Furthermore, the map of earthquake hazard zonation has been drawn. Properties of occurred earthquake were collected since 1903. The most probable event of earthquake in Qum has been predicted 31.6% in center, and 28.9% in north of Semnan

---

This paper presented the feasibility of developing and using artificial neural networks (ANNs) for modeling the monotonic large scale triaxial tests over angular, rounded rockfill and materials contained various percentages of fines as a construction material in some dams in Iran. The deviator stress/excess pore water pressure versus axial strain behaviors were firstly simulated by employing the ANNs. Reasonable agreements between the simulation results and the tests results were observed, indicating that the ANN is capable of capturing the behavior of gravely materials. The database used for development of the models comprises a series of 52 rows of pattern of strain-controlled triaxial tests for different conditions. A feed forward model using multi-layer perceptron (MLP), for predicting undrained behavior of gravely soils was developed in MATLAB environment and the optimal ANN architecture (hidden nodes, transfer functions and training) is obtained by a trial-and-error approach in accordance to error indexes and real data. The results indicate that the ANNs models are able to accurately predict the behavior of gravely soil in CU monotonic condition. Then, the ability of ANNs to prediction of the maximum internal friction angle, maximum and residual deviator stresses and the excess pore water pressures at the corresponding strain level were investigated. Meanwhile, the artificial neural network generalization capability was also used to check the effects of items not tested, such as density and percentage smaller of 0.2 mm.

---

 In this research, prediction of concrete strength containing different aggregates using Non-destructive (Ultrasonic) testing through Artificial Neural Networks was carried out. For this purpose, aggregates with different properties were selected from the quarries, and then their destructive and nondestructive properties were obtained in laboratory. The significance of this research, using different aggregates with physical, mechanical and chemical properties also used two different test methods, such as Non-destructive static and dynamic testing, which are respectively uniaxial compressive strength and compressive wave velocity. Thus, this model includes various types of samples and the prediction model includes static and dynamic tests. The results showed that the use of artificial neural networks not only increases the accuracy, but also it reduces costs and time.

---

Ground-Penetrating Radar (GPR) is a non-destructive and high-resolution geophysical method which uses high-frequency electromagnetic (EM) wave reflection off buried objects to detect them. In current research this method has been used to identify geometrical parameters of buried cylindrical targets such as tunnel structures. To achieve this aim, relationships between the geometrical parameters of cylindrical targets with the parameters of GPR hyperbolic response have been determined using two intelligent pattern recognition methods known as artificial neural network and template matching. To this goal GPR responses of synthetic cylindrical objects produced by 2D finite-difference method have been used as templates in the neural network and template matching algorithms. The structure of applied neural network has been designed based on extracting discriminant and unique features (eigenvalues and the norm of eigenvalues) from the GPR images and predicting all geometrical parameters of the targets, simultaneously. Also the template matching operation carried out using two diverse similarity approaches, spatial domain convolution and normalized cross correlation in 2D wave number domain. The results of the research show that both two employed intelligent methods can be applied for in situ, real-time, accurate and automatic interpretation of real GPR radargrams, however in general the neural network method has led to less error and better estimation than template matching to predict the geometrical parameters of the cylindrical tar

---

Landslides are natural hazards that make a lot of economical and life losses every year. Landslide hazard zonation maps can help to reduce these damages. Taleghan watershed is one the susceptible basin to landslide that has been studied. In this paper, landslide hazard zonation of the study area is performed at a scale of 1:50,000. To achieve this aim, layers information such as landslides distribution, slope, aspect, geology (lithology), distance from the faults and distance from rivers using artificial neural network-based Radial Basis Function (RBF) and perceptron neural network (MLP), has been studied. Principal of RBF method is similar to perceptron neural network (MLP), which its ability somewhat has been identified up to now and there are several structural differences between these two neural networks. The final results showed that the maps obtained from both methods are acceptable but the MLP method has a higher accuracy than the RBF method.

---

Maximum surface settlement (MSS) is an important parameter for the design and operation of earth pressure balance (EPB) shields that should determine before operate tunneling. Artificial intelligence (AI) methods are accepted as a technology that offers an alternative way to tackle highly complex problems that can’t be modeled in mathematics. They can learn from examples and they are able to handle incomplete data and noisy. The adaptive network–based fuzzy inference system (ANFIS) and hybrid artificial neural network (ANN) with biogeography-based optimization algorithm (ANN-BBO) are kinds of AI systems that were used in this study to build a prediction model for the MSS caused by EPB shield tunneling. Two ANFIS models were implemented, ANFIS-subtractive clustering method (ANFIS-SCM) and ANFIS-fuzzy c–means clustering method (ANFIS-FCM). The estimation abilities offered using three models were presented by using field data of achieved from Bangkok Subway Project in Thailand. In these models, depth, distance from shaft, ground water level from tunnel invert, average face pressure, average penetrate rate, pitching angle, tail void grouting pressure and percent tail void grout filling were utilized as the input parameters, while the MSS was the output parameter. To compare the performance of models for MSS prediction, the coefficient of correlation (R²) and mean square error (MSE) of the models were calculated, indicating the good performance of the ANFIS-SCM model.

---

In urban areas, it is essential to protect the existing adjacent structures and underground facilities from the damage due to tunneling. In order to minimize the risk, a tunnel engineer needs to be able to make reliable prediction of ground deformations induced by tunneling. Numerous investigations have been conducted in recent years to predict the settlement associated with tunneling; the selection of appropriate method depends on the complexity of the problems. This research intends to develop a method based on Artificial Neural Network (ANN) for the prediction of tunnelling-induced surface settlement. Surface settlements above a tunnel due to tunnel construction are predicted with the help of input variables that have direct physical significance. The data used in running the network models have been obtained from line 2 of Mashhad subway tunnel project. In order to predict the tunnelling-induced surface settlement, a Multi-Layer Perceptron (MLP) analysis is used. A three-layer, feed-forward, back-propagation neural network, with a topology of 7-24-1 was found to be optimum. For optimum ANN architecture, the correlation factor and the minimum of Mean Squared Error are 0.963 and 2.41E-04, respectively. The results showed that an appropriately trained neural network could reliably predict tunnelling-induced surface settlement.

---

This paper presents a feed-forward back-propagation neural network model to predict the retained tensile strength and design chart to estimate the strength reduction factors of nonwoven geotextiles due to the installation process. A database of 34 full-scale field tests was utilized to train, validate and test the developed neural network and regression model. The results show that the predicted retained tensile strength using the trained neural network is in good agreement with the results of the test. The predictions obtained from the neural network are much better than the regression model as the maximum percentage of error for training data is less than 0.87% and 18.92%, for neural network and regression model, respectively. Based on the developed neural network, a design chart has been established. As a whole, installation damage reduction factors of the geotextile increases in the aftermath of the compaction process under lower as-received grab tensile strength, higher imposed stress over the geotextiles, larger particle size of the backfill, higher relative density of the backfill and weaker subgrades.

 

---

Introduction
The discussion of modeling the interaction of soil-pile groups due to a large number of parameters involved in is one of the complex topics and it has been one of the interests to researchers in recent years and has been dealt with in various ways. In recent years, the artificial neural network method has been used in many issues related to geotechnical engineering, including issues related to piles.. In this study, firstly it was tried to explain the importance of soil - structure interaction in calculating the dynamic response of bridges. Then, the effect of different effective parameters in calculating the interaction stiffness of the pile - soil group using artificial neural network was studied.  For this purpose, firstly, Sadr Bridge ( The intersection of Modarress and Kaveh Boulevard because the presence of tallest piers ) in the transverse direction, considering and without considering of the effect of soil - structure interaction was analyzed. The analysis was carried out in which the substructure soil was replaced with a set of springs and dashpots along the piles. Considering the fact that many factors are involved in determining the equivalent stiffness of springs, in the second stage, the effect of different factors on the stiffness of spring equations using artificial neural network was investigated. Finally, the artificial neural network method was used as a suitable method in order to estimate the equivalent stiffness values, the equivalent stiffness of the pile - soil group was introduced for different input values. equivalent stiffness of the substructure soil using the artificial neural network ,has not been used by researchers yet, so estimation of the optimal length and diameter of piles used in constructions and estimating the seismic performance of the bridge system after its implementation could be effective .
Material and methods
In this paper, spring-dashpot method is proposed to the non-uniform analysis of single-pier bridges which led to a 5-degree freedom model in the case of Sadr Bridge. This study also endeavors to investigate the SSI effect in dynamic analysis of bridges. This method is based on the traditional spring-dashpot method but in this method, non-linear stiffness is used along the piles, instead of linear stiffness and upgraded shape functions and coefficients are applied to make more precise mass, stiffness and damping matrices. Then the seismic responses of Sadr Bridge are compared in different conditions including or excluding the SSI effects. Considering the fact that in the present study to calculate the stiffness of the soil-pile group at depth, due to the effect of soil - structure interaction, the recommended method by API is used, the study of neural network analysis was used and the effect of different parameters used to determine the complexity of the soil-pile group system has been evaluated. The multi-layer feeder network, which has the most application in engineering issues, has an input layer, an output layer and one or more layers of hidden content, has been used for this purpose.  The best model of the neural network with a topology of 1-20-6 was provided using the hyperbolic sigmoid activation function, and the Levenberg Marquardt model and the training cycle 84, which had the least error mean square and the best regression coefficient. The effect of internal friction angle, soil density, pile diameter and the resistance per unit length has been evaluated with this method.
Results and discussion
[8] ارائه شده است صورت می پذیرد In this study, the importance of considering the effect of soil - structure interaction on the dynamic response of the Sadr Bridge was studied. Dynamic stiffness of the soil around the pile group was calculated based on the equivalent linear method and using the p-y springs. So, the effect of substructure soil was considered in dynamic analysis of the system . The artificial neural network was used to predict the stiffness of the soil - pile group, based on various input parameters and the stiffness sensitivity analysis of the calculated output values was conducted. In hard soils, the stiffness of the pile - soil group increases with increasing the diameter of the pile in the range of 1 to 1.5 m in diameter. However, in the range of 0.5 to 1 m in diameter, the diameter of the pile does not have much effect on the stiffness of the system and also stiffness decreases in the range of 1.5 to 2 m in diameter by increasing the pile diameter. Soil specific weight and angle of internal friction can change the system stiffness but the effect of the soil specific density is much greater on the stiffness of the soil-pile group system. Generally, the specific density in the range of 1000 to 2300 (kg/m³) will increase the stiffness of the system. In general, the ultimate strength of the soil among 100 to 550 (kN/m) affects the system stiffness. This effect within the ultimate strength between 100 and 220 (kN/m) causes increasing in the interaction stiffness value of the system and in the range of 220 to 550 (kN/m) causes reducing the stiffness of the system . The ultimate strength values ​​in a unit of length outside of the above range have little effect on the system interference stiffness. Despite the fact that the problem of calculating the soil - pile interaction stiffness is a direct solution, the use of the proposed neural network model can help in predicting optimal values ​​of diameter and length of the pile to achieve maximum soil- pile stiffness and especially for long bridges it will has a significant impact on reducing cost and seismic design of the bridge.
Conclusion
The results of this study are as follows:
The results showed that considering the interaction effect, although it increases the relative displacement of the deck, reduces the maximum base shear and moment. This suggests that considering the maximum base shear and moment in the interaction conditions may not lead to a seismic design for certainty, although closer to reality.
Artificial neural network is an efficient way and new method to predict the stiffness of the soil-pile group system based on different input values that have not been used yet. So that with the physical and mechanical properties of the soil as well as the geometric properties of the piles, it is possible to predict the interaction stiffness values with the proper precision.
According to the results and diagrams obtained from the neural network model, which are mainly sinusoidal, the optimal values ​​of the interaction stiffness can be obtained by obtaining the pile diameter, specific gravity, the internal soil friction soil to achieve optimal interaction strength. It is also possible for each site to estimate the depth of the piles in order to achieve optimal hardness. 
./files/site1/files/142/4Extended_Abstracts.pdf