---

In this research, general performance of Radial basis function (RBF) Artificial neural networks in experimental data on effect of the NiO, WO3, TiO2,ZnO and Fe2O3 nanoparticles in different temperatures and mass fractions on the viscosity of crude oil has been studied. The morphology and stability of the nanoparticles has been analyzed by DLS and TEM analysis, the results showed that the average diameter of the nanoparticles is from 10 to 30 nm which defers for different oxide nanoparticles. The general method for calculating the optimum span of the Isotropic Gaussian function with special algorithm for learning RBF networks, has been presented. This study's results declared that the RBF artificial neural networks, because of having strong academic basis and having the ability to filter the noises, has a good performance. With increase in temperature, the ratio of the viscosity of the nanofluids decreases compering to the viscosity of the basefluid. Also with increase in nanoparticles mass fraction the related viscosity increases boldly. For temperatures higher than 50°C, the related viscosity is less than the viscosity of the basefluid.

---

Research subject: Expandable Poly Styrene (EPS) has many applications. This polymer prepared by the radical polymerization. This material has many uses in packaging and insulation industries Some of the properties of this polymer like low mechanical strength caused its applications to be limited. By adding some materials, these properties can be improved. Styrene Butadiene Styrene (SBS) is from the materials that which by adding it to the EPS it can improve its quality.
Research approach: In this research, EPS having different percentages of SBS (0, 0.01, 0.02, 0.03) in different conversion percentages (0.6, 0.63, 0.66, 0.69) has been prepared. Different tests like Impact Test, Modular Melt Flow test, Vicat Softening Temperature test, Tensile at Break test, K-value test, Rochwell Hardness test and Elongation at Break test are done on the prepared polymer. Laboratory gained data has been simulated by Multi-Layer Perceptron (MLP) method of artificial neural networks (ANN) and the simulated data covers the laboratory data perfectly.
Main Results: Investigating the tests show that in constant percentages of SBS in EPS with increase in conversion percentage of EPS, the numerical amount of the tests increases except MFI test (low MFI number means better quality). Increase in SBS percentage in the EPS, increases the properties of polymer. In addition, the results of simulation show that the laboratory data covers the the simulated data perfectly. The data obtained from the results of this reasearch can be used for predicting the data for the points which has not been tested. Adding SBS in different weight percentages of poly styrene in different conversion percentages in order to increase the properties of poly styrene has been used for the first time in this research and the laboratory data results in points which has not been tested has been acquired by applications of ANN.

---

The complex and stochastic nature of the electro-discharge machining (EDM) process has frustrated numerous attempts of physical modeling. In this paper two supervised neural networks, namely back propagation (BP), and radial basis function (RBF) have been used for modeling the process. The networks have three inputs of current (I), voltage (V) and period of pulses (T) as the independent process variables, and two outputs of material removal rate (MRR) and surface roughness (Ra) as performance characteristics. Experimental data, employed for training the networks and capabilities of the models in predicting the machining behavior have been verified. For comparison, quadratic regression model is also applied to estimate the outputs. The outputs obtained from neural and regression models are compared with experimental results, and the amounts of relative errors have been calculated. Based on these verification errors, it is shown that the radial basis function of neural network is superior in this particular case, and has the average errors of 8.11% and 5.73% in predicting MRR and Ra, respectively. Further analysis of machining process under different input conditions has been investigated and comparison results of modeling with theoretical considerations shows a good agreement, which also proves the feasibility and effectiveness of the adopted approach.

---

Aims: Soil organic carbon (SOC) is contemplated as a crucial proxy to manage soil quality, conserve natural resources, monitoring CO2 and preventing soil erosion within the landscape, regional, and global scale. Therefore, the main aims of this study were to (1) determine the impact of terrain derivatives on the SOC distribution and (2) compare the different algorithms of topographic wetness index (TWI) calculation for SOC estimation in a small-scale loess hillslope of Toshan area, Golestan province, Iran. (3) Comparison between multiple linear regression (MLR) and artificial neural networks (ANN) methods for SOC prediction.
Materials & Methods: total of 135 soil samples were taken in different slope positions, i.e., shoulder (SH), backslope (BS), footslope (FS), and toeslope (TS). Primary and secondary terrain derivatives were calculated using digital elevation model (DEM) with a spatial resolution of 10 m × 10 m. To SOC estimation (dependent variable) was applied two models, i.e., MLR and ANN with terrain derivatives as the independent variables.
Findings: The results showed significant differences using Duncan’s test in where TS position had the higher mean value of SOC (25.90 g kg⁻¹) compared to SH (5.00 g kg⁻¹) and BS (12.70 g kg⁻¹) positions. The present study also revealed which SOC was more correlated with TWIMFD (Multiple-Flow-Direction) and TWIBFD (Biflow-Direction) than TWISFD (Single Flow Direction). The MLR and ANN models were validated by additional samples (25 points) that can be explain 65% and 76% of the total variability of SOC, respectively, in the study area.
Conclusion: These results indicated that the use of terrain derivatives is a beneficial method for SOC estimation. In general, an accurate understanding of TWIMFD is needed to better estimate SOC to evaluate soil and ecosystem related effects on global warming of as this hilly region at a larger scale in a future study.

---

---

In today's competitive world, attention to the improvement and development of employees has become increasingly important. One of the procedures that organizations can use to improve their employees is implementation of assessment and development centers. But it should be noted that the implementation of improvement programs for each participants in the assessment center is time-consuming and costly. Accordingly, the aim of this study is reducing costs of assessment center implementation by segmenting the participants in assessment center and providing improvement programs for each sector. Sampling method is targeted and based on criteria and the sample consists of 75 employees. In this research, self-organizing maps have been used for segmentation of employees and the data were analyzed by Viscovery Profiler software. According to the findings, four sectors identified in the assessment center and they have been named: Talented, improvable employee (with a focus on competence), improvable employee (with focus on character) and insusceptible experts. Results of this segmentation can be used in five areas, including training, promotion, solve organizational challenges, recruitment and retention policies and rewards. Based on the results, for the areas that identified in the assessment center, suggestions are offered.

---

Stock market timing is a very difficult task because of the complexity of the market. Since there are various factors affecting the market and therefore it is not a simple task to predict future stock price and its trend. This paper aims to apply advanced tools and algorithms such as the artificial neural networks (ANN) to model nonlinear processes and predict future stock price and its trend. More specifically, this study explores the abilities of the ANN to enhance the effectiveness of the technical analysis indicators to predict stock trend signals. Using a sample of 50 companies in the Tehran Stock Exchange (TSE), the results indicate that the ANN is capable to predict the direction of the short term movement in the future stock price. After considering the transaction costs, the results confirm that there is not significant difference among the returns gained from the ANN method, buy and hold strategy, and the most profitable technical indicators in the market when the trend is increasing. While, the ANN model yields higher returns compared to buy and hold strategy in the market when the trend is decreasing. Nevertheless, in the case of decreasing trend, the finding confirms the trend indicators (moving averages) achieve the highest returns.

---

The evolution of financial data shows a high degree of volatility of the series, coupled with increasing difficulties of forecasting financial variables. Some alternative forecasting methods, based on the literature review, have been developed, which can be particularly useful in the analysis of financial time series. Despite of the numerous time series forecasting models, the accuracy of time series forecasting is fundamental to many decision processes. Selecting an efficient technique in unique situations is very difficult task for forecasters. Many researchers have integrated linear and nonlinear methods in order to yield more accurate results. In practice, it is difficult to determine the time series under study are generated from a linear or nonlinear underlying process while many aspects of economic behavior may not be pure linear or nonlinear. Although both ARIMA and Artificial Neural Networks (ANNs) models have the flexibility in modeling a variety of problems, none of which is universally the best model used indiscriminately in every forecasting situation. In this paper, based on the foundations of ARIMA and ANNs models, a hybrid method is proposed to forecast exchange rate. Empirical results indicate that integrating linear and nonlinear ARIMA and Artificial Neural Networks (ANNs) models can be an effective way to improve forecasting accuracy achieved by either of the above linear and nonlinear models used separately.

---

“Selecting right Person for the right job” in the organizations seems to be the most important managerial issue. Traditionally, this is realized through a “simple Job- Person Match”, which is an “Individual view”. This methodology, by some trends and paradigm shift in human resource management (i.e. Team Working), should be evolved to a “Group viwe”. In this study, “interpersonal interactions” is added to the traditional one. First, the suggested model for personnel selection and placement is formulated into a Quadratic mathematical form and then a Hopfield neural network has been used to solve it by using Matlab software, a well known and validated software for neural nets. The results show, there is a significant difference between the mean of solutions by traditional view (based on individual level) and the mean of solutions by the new suggensted one (based on group level): t(19) = - 10.966, P-Value=0.000.

---

  The mechanical losses which occur in agricultural products are the damage's which imposed on country economy. It is important to investigate the bruising phenomena, as an index of mechanical losses for loss reduction and optimization of harvest and postharvest machinery. In the current study, by means of an impact pendulum apparatus and by conducting a series of impact tests, the effects of temperature (0, 10, 20 and 30⁰C), variety (Golden Delicious and Red Delicious), padding surface (corrugated carton, rubber and galvanized iron) and kinetic energy (300, 600 and 900 mJ) were investigated on rate of apple bruise. Statistical results showed that the effect of temperature, variety, padding surface and impact energy were significant on the mean value of bruise volume at 1% of statistical level. By increasing temperature, the bruised volume was decreased, whereas it increased by increase of energy level in both varieties. While, the Golden Delicious had more strength than Red Delicious. Also, the maximum rate of bruised volume was related to Red Delicious in contacting to galvanized iron and the minimum rate was related to Golden Delicious in contacting to corrugated carton. Prediction of bruised volume was provided using artificial neural network based on four factors of: temperature, impact energy, padding surface and variety. Multilayer perceptron of neural networks were used for prediction of bruised volume. In comparison with other topologies, algorithm Levenberg-Marquardt had better performance with structure 1-26-4 and logsigmoid transfer function in hidden layer. Based on this algorithm, the mean of prediction accuracy in training, evaluation and testing process was equal to 92.48, 88.94 and 87.72 percent, respectively. In addition, the correlation coefficient (R) was calculated equal to 0.975 for linear regression between predicted model and experimental data.  

---

Fine-blanking is an effective and economical shearing process which offers a precise and clean cutting edge finish, eliminates unnecessary secondary operations and increases quality. Fine-blanking process utilizes triple-action tools: a punch, a stripper with an indented V-ring and a Counter punch (ejector) to generate a highly compressive stress state. The deformation is more violent and localized than that of any other metal forming operations. Therefore it is difficult to fully understand the mechanism of the process. This study investigates the effect of V-ring indenter, clearance of die, Force of holder and Counter punch, etc on state of stress, quality and accuracy of production. Some parameters have both positive and negative effect on quality of production and the life of the tool. Utilizing V-Ring indenter in Die will increase quality of production and life of the tool. Also Artificial Neural Networks was used to simulate Fine-Blanking process. It has been shown that booth of FEM and ANN is suitable for simulating and forecast of effect of the parameters on production.

---

Runoff estimation is one of the main challenges encountered in water and watershed management. Spatial and temporal changes of factors which influence runoff due to het-erogeneity of the basins explain the complicacy of relations. Artificial Neural Network (ANN) is one of the intelligence techniques which is flexible and doesn’t call for any much physically complex processes. These networks can recognize the relation between input and output. In this study ANN model was employed for runoff estimation in Plaszjan Riv-er basin in the central part of Iran. The models used are Multiple Perceptron (MLP) and Recurrent Neural Network (RNN). Inputs include data obtained from 5 rain gauges as well as from 2 temperature recording gauges, the output of the model being the monthly flow in Eskandari Hydrometric Station. Preprocessing of the data as well as the sensitivity analysis of the model were carried out. Different topologies of Neural Networks were cre-ated with change in input layers, nodes as well as in the hidden layer. The best architec-ture was found as 7.4.1. Recurrent Neural Network led to better results than Multilayer Perceptron Network. Also results indicated that ANN is an appropriate technique for monthly runoff estimation in the selected basin with these networks being also of the ca-pability to show basin response to rainfall events.

---

In this research work, a comprehensive combustion analysis has been conducted to evaluate the performance of a low speed diesel engine (M8/1 Lister) using biodiesel fuel. Waste vegetable cooking oil as an alternative fuel. Biodiesel obtained from waste vegetable cooking oil (WCO) as an alternative fuel. The properties of biodiesel produced from WCO was measured based on ASTM standards. In order to compare brake power, torques , brake specific fuel consumption (BSFC) and concentration of the UHC and CO emissions of the engine, it has been tested under same load of Dynamometer(5 levels) and biodiesel fuel blends (levels)) at constant engine speed(750 rpm). The results were found to be very comparable. An artificial neural network (ANN) was developed based on the collected data of this work. Multi layer perceptron network (MLP) was used for nonlinear mapping between the input and the output parameters. Different activation functions and several rules were used to assess the percentage error between the desired and the predicted values. The results showed that the training algorithm of Back Propagation was sufficient in predicting the engine torque, brake power, specific fuel consumption and exhaust gas components for different engine loads and different fuel blends ratios.

---

Market segmentation by artificial neural networks has no deep root in the history. Generally, this ever developing approach has started since several years ago, and developed to other marketing areas. Now, beside statistical techniques, it is considered as one of the most popular methods in Custamer classification. In Due to the necessity of recognizing the target market for a specific company, a need for the usage of an effective approach for customers grouping was recognized, in the Present research, and finally cluster analysis with SOM neural networks, was selected, and used for customers clustering. Firstly, beneficent criteria for market segmentation were identified, and then a proper, questionnaire was designed. After gathering the questionnaires and collecting the data, using artificial neural networks, the customers were clustered, and the obtained, results were analyzed. At the end, the Findings of this method were compared with those of the traditional methods for clusteringusing K-means.

---

In this research, the data relating to global land/oceans temperature anomalies and annual mean precipitation of Tabriz station were used for the period of 1951-2005. The main methodologies used in this research include the Pearson correlation coefficient method, analysis of trend component of time series, simple linear and polynomial regression (as a semi-linear model) and Artificial Neural Networks methods. The results of applying Pearson analysis indicated a significant negative and an inverse correlation between global land/oceans temperature anomalies and annual precipitation in Tabriz station. This is an indicative of increase in precipitation and occurrence of wet years during the negative global temperature anomalies and, on contrary, precipitation reduction and occurrence of droughts during the positive global temperature anomalies. The analysis of long term trend components of time series showed that the annual mean precipitation of Tabriz has a decreasing trend towards the length of the period, but annual global land/oceans temperature anomalies has an increasing trend towards the length of the period. Also we simulated the relationships between annual precipitation in Tabriz station and global warming using Artificial Neural Networks. Applying of different methods recognized artificial neural network as a better and more accurate simulation model compared to the other models applied in this research, i.e. simple regression model, and semiـ linear polynomial regression with the power of 6 models. Different artificial neural network methods were used to demonstrate this relation, among which the Multi Layer Perceptron (MLP) with three hidden layers analysis with back propagation learning algorithm showed excellent capability in predicting the correlation between the series.

---

Sensible vibration of steel beams in long spans is undesirable issue in the buildings. These beams may be vibrated during people passage, although the strength calculations of this beams to be performed, accurately and drift control index based on buildings codes to be considered. Iranian Steel Buildings Code has offered a formula for controlling of vibration of beams in building frames with pin connections in serviceability phase. However, this code has not presented criteria for beams include fixed connections. Since these beams have the considerable portion of building frames, their vibration control needs special attentions. The presented equations for determination of beams frequency are complicated and have been not used for control of buildings floor vibration. In this paper, the mentioned formula in forenamed codes has been discussed. The dynamic analysis, finite element method (FEM) and artificial neural networks (ANN) techniques have been adopted to constitute the frequency equations of the fix ends and cantilever steel beams. Comparison of resulted frequency from presented equations and ANN showed that the error is low. Furthermore, it is suggested that use proposed equations for determination of frequency of moment connection beams.

---

Determining the bearing capacity of piles is an important issue that always Geotechnical engineers focus on. Effect of factors such as environmental dissonance of soil which contains a pile, pile implementation, pile gender and its shape make correct estimation of bearing capacity difficult. Pile load testing as a reliable method could be used in various stages of analysis, design and implementation of piles to determine the axial bearing capacity of piles. On the other hand, pile load testing, despite high accuracy, imposes high cost and long duration for development projects and it causes limitations in this experiment. Thus acceptance of numerical analysis at geotechnical studies is increasing. In this study serious models of multi-layer perception neural network, one of the most commonly used neural networks, was used. In all models four parameters are used as input data which are length and diameter of the pile, the coefficient of elasticity and internal friction angle of soil and the bearing capacity of piles is used as output data. Models have reasonable success in predicting the bearing capacity of piles. To increase the accuracy of predicting bearing capacity, for the network training stage the real tests that has been done at the geotechnical studies of dry dock area Hormozgan by POR Consulting Engineers were used. According to (Because we) need of more data for training and testing network, several tests on pile bearing capacity, in smaller dimensions were performed in the laboratory. To perform these tests the device of pile bearing capacity, made in university of Tarbiat Modarres, was used. Models based on neural networks, unlike traditional models of behavior don’t explain effect of input parameters on output parameters. In this study, by the sensitivity analysis on the optimal structure of introduced models in each stage it has been somewhat trying to answer this question.

---

Determining the bearing capacity of piles is an important issue that always Geotechnical engineers focus on. Effect of factors such as environmental dissonance of soil which contains a pile, pile implementation, pile gender and its shape make correct estimation of bearing capacity difficult. Pile load testing as a reliable method could be used in various stages of analysis, design and implementation of piles to determine theaxial bearing capacity of piles. On the other hand, pile load testing, despite high accuracy, imposes high cost and long duration for development projects and it causes limitations in this experiment. Thus acceptance of numerical analysis at geotechnical studies is increasing. The modeling using artificial neural networks is the method that is based on previous data and don’t need to simplify and improve the high reliability coefficient. In this study serious models of multi-layer perceptron neural network, one of the most commonly used neural networks, was used. Network design and factors influencing its behavior in this issue has been studied as a summary. In this study, artificial neural networks are used for prediction of bearing capacity of driven steel piles in sandy soil, in all models four parameters are used as input data which are length and diameter of the pile, the coefficient of elasticity and internal friction angle of soil and the bearing capacity of piles is used as output data. Models have reasonable success in predicting the bearing capacity of piles. In order to evaluation of networks, the different indices such as RMSE, MAE, MAXAE and SDAE were used. To increase the accuracy of predicting bearing capacity, for the network training stage the real tests that has been done at the geotechnical studies of dry dock area hormozgan by POR Consulting Engineers were used.Acording to (Because we) need of more data for training and testing network, several tests on pile bearing capacity, in smaller dimensions were performed in the laboratory. The sixty tests have been performed on piles with various length (35, 40, 45 and 50 cm), various diameters (20, 25 and 32 mm) and different relative compacted sandy beds (50, 60, 70, 75 and 80%). To perform these tests the device of pile bearing capacity, made in university of TarbiatModarres, was used. Models based on neural networks, unlike traditional models of behavior don’t explain effect of input parameters on output parameters. In this study, by the sensitivity analysis on the optimal structure of introduced models in each stage it has been somewhat trying to response this question. .

---

The prerequisite in the majority of control processes is modeling. The model used to design a controller must be both accurate and real-time. Utilizing prevalent approaches of modeling, namely modeling based on (numerically) solving the equations governing the fluid in the combustion chamber, is too time-consuming and not suitable for a control purpose. This paper is to model combustion in an SI engine by means of neural networks and present an accurate and fast-response model for combustion. Obviously, any training procedure of neural networks does involve empirical data acquisition. On the other hand, engine testing is highly expensive, and testing data tables available (in industry) are not sufficient to train neural networks. In this paper, first with the aid of a CFD software, a one-dimensional model of an engine is constructed, and then calibrated using factual experimental data at hand. Afterwards, acquiring data required is performed via the validated CFD model. As a matter of fact, because of not having access to necessary experimental coefficients, calibration is an extremely complicated and time-consuming process. It will be attempted to accomplish and spell out the calibration of the engine model in the GT-Power software, in a scientific practice. After a brief survey on the methods employed in designing the neural networks, modeling of the combustion chamber will be stated. Eventually, the response of the constructed NN model will be compared to the results gained from the GT-Power software, and the great accuracy of the NN model will be indicated.

---

Collapsible soils are soils that compact and collapse after they get wet. The soil particles are originally loosely packed and barely touch each other before moisture soaks into the ground. As water is added to the soil in quantity and moves downward, the water wets the contacts between soil particles and allows them to slip past each other to become more tightly packed. Water also affects clay between other soil particles so that it first expands, and then collapses like a house of cards. Another term for collapsible soils is "hydrocompactive soils" because they compact after water is added. The amount of collapse depends on how loosely the particles are packed originally and the thickness of the soil that becomes wetted. Collapsible soils consist of loose, dry, low-density materials that collapse and compact under the addition of water or excessive loading. These soils are distributed throughout the southwestern United States, specifically in areas of young alluvial fans, debris flow sediments, and loess (wind-blown sediment) deposits. Soil collapse occurs when the land surface is saturated at depths greater than those reached by typical rain events. This saturation eliminates the clay bonds holding the soil grains together. Similar to expansive soils, collapsible soils result in structural damage such as cracking of the foundation, floors, and walls in response to settlement. Collapsible soils may be suspected in undeveloped areas that have young, accumulating sandy and silty soils in dry areas. The soils may be confirmed to be collapsible through engineering testing. These tests include study of seismic waves through the soils, rates of drilling through the soils (blow counts), and testing undisturbed soil samples obtained by careful drilling for compaction after wetting. In this study, the ability of Artificial Neural Networks (ANN) has been investigated to determine the collapse potential of soils. Therefore, different samples of collapsible soil have been collected from an area (Zahedan plain). General tests were performed on the samples in the laboratory and 130 samples of collapsible soil from different depths and locations were recorded in the database. The collapse potential tests (One-dimensional collapse test) was carried out on the samples and with the aim of further investigations, the grain size distribution, specific gravity, atterberg limits and strength properties of the samples were performed. In the later stages, the collapsible samples data were prepared for the artificial neural networks input. After the network training process and the subsequent learning, some network models have been selected under experiments, which include six inputs and one output. According to the predicted results, it was indicated that the correlation between experimental and predicted data by the ANN is 95%. Furthermore, the results show that artificial neural networks can predict collapse potential of soils, also the calculations and required tests will be reduced due to their simple use and inexpensive tests.