Year:2001 | Volume:34 | Issue:4 (70)|Papers Count:9

The goal of neural network engineering (NNE) is to study the advantages and disadvantages of neural networks and also providing methods to increase their performance. One of the problems in NNE is determination of optimal topology of neural networks for solving a given problem. There is no method to determine the optimal topology of multi-layer neural networks for a given problem. Usually, the designer selects a topology for neural networks and then trains it. The selected topology remains fixed during the training period. The performance of neural network depends on its size (number of hidden layers and hidden units). Determination of the optimal topology of neural network is an intractable problem. Therefore, most of algorithms for determination of the topology of neural network are approximate algorithms. These algorithms could be classified into five main groups: pruning algorithms, constructive algorithms, hybrid algorithms, evolutionary algorithms, and learning automata based algorithms. The only learning automata (LA) based algorithms, called survival algorithm, has been proposed by Beigy and Meybodi. This algorithm uses an object migrating learning automata and error back propagation (BP) algorithm and determines the number of hidden units of three layers neural networks, as training proceeds. In this paper, we propose three algorithms which are based on LA and BP. These algorithms determine a near optimal topology with low time complexity and high generalization capability for a given training set. These algorithms have two parts: determination of number of hidden units and determination of the number of hidden weights. One of the proposed algorithms uses the survival algorithm to determine the number of hidden units. A new algorithm based on LA is proposed to determine the number of hidden weights.  This algorithm deletes weights with small effect, which leads to lower time complexity and higher generalization rate. Two other algorithms do not omit the hidden units explicitly; a hidden unit is omitted when all its input weights are deleted. Most of the reported algorithms in the literature for determination of topology of neural networks use hill-climbing method and may stuck at local minima. The proposed algorithms use global search method which results in increasing the probability of escaping from local minima. The proposed algorithms have been tested on several problems such as: recognition of Farsi and English digits. Simulation results show that the produced networks have good performance. The proposed algorithms are compared with Karnin pruning algorithm.

In this paper the effects of grains on near shore hydrodynamic processes are studied by applying physical models. The experiments were conducted in a wave basin with fixed bed. Both regular and irregular waves were used in the tests. Wave induced currents were measured using bi-directional electromagnetic current meters and waves heights were measured using capacitance wave height meters. Twelve tests were conducted on straight beach with no grains and a single grain. The wave height and current patterns are discussed. The significance of wave-current interaction in the near shore region was shown in the results of the experiments, especially in the presence of grains. The data provided by this series of tests could be applied to the calibration and verification of two and three dimensional numerical models.

Modeling serves as an important and effective tool in urban runoff quality management. The models can then be exploited for a variety of purposes including forecasting runoff quality, assessment of the impacts of different pollutant control options, and the implementation of best management methods. The present paper will first review the various methods and models used in urban runoff quality forecasting and will then study the possibilities for using these models in forecasting runoff quality in Iranian urban areas. Finally, a model selected for one such situation will be presented. Urban runoff quality models include a vast variety ranging from simple mathematical relations to comprehensive quality quantity computer models, including methods of constant concentration and unit load, statistical methods, regression equations, rating curves, equations of pollutant buildup and wash-off and all embracing computer models. The survey of the existing models and methods in the present study revealed that due to the present insufficiency of records and data, it would be wiser to begin surface runoff quality modeling for Iranian cities through the application of simpler models and to postpone the adoption of more complex ones such as the all-embracing computer models until more data have been gathered through time. In the present study, the above mentioned methods have been studied as a case study in forecasting the runoff quality of the Si-Q-Se Pol Bridge area in Isfahan and an appropriate model has been proposed for this particular area.

Using DEM (Discrete Element Method), a model is presented for simulation of breakage of two-dimensional polygon-shaped particles. In this model, shapes of the particles generated after breakage are predefined and each uniform (uncracked) particle is replaced with some smaller inter-connected sub-particles. If the bond between these sub-particles break, breakage will happen. With the help of this model, it is possible to study the influence of particle breakage on internal angle of friction and deformability of granular materials. For this purpose, two series of biaxial test simulations (breakage is enabled and disabled) are conducted on assemblies of two-dimensional polygon-shaped particles with different confining pressures and the results are compared. It was found that the simulation results are in good agreement with observations obtained from experimental tests.

In order to study of efficiency and the overall volumetric mass transfer coefficients, several series of experiments are carried out by a ten-stage progress horizontal mixer-settler. According to the experiments, we can conclude. that in general an increase of impeller speed results to an increase of extraction efficiency. Moreover, an increase of impeller speed results to an increase of the overall volumetric mass transfer coefficients, but after a specific speed the overall volumetric mass transfer coefficients decrease. Also, an increase of raffinate volumetric flow rate causes a decrease of extraction efficiency and results to an increase of the overall volumetric mass transfer coefficients. On the other hand, an increase of extract (solvent) volumetric flow rate causes an increase of extraction efficiency and results to an increase of the overall volumetric mass transfer coefficients. Also, the obtained results show that the resistance against mass transfer in both phases is considerably high and none of them can be neglected The hold-up of each stage can also affect the stage efficiency and overall volumetric mass transfer coefficients. The value of the overall volumetric mass transfer coefficient for each stage as compared to other stages doesn't have many differences and therefore an average overall volumetric mass transfer coefficient can be defined for designing and computation.

In this study ,the environmental condition of the Zayandehrood river was theoretically and experimentally investigated. Therefore, a dynamic mathematical model with multiple pollution sources was formulated to predict the parameters such as BOD, DO, and temperature of water in the different locations of the river. Also, the water samples were collected from 120 kilometers of the river starting from the south Isfahan municipal wastewater treatment plant and the BOD, DO, and temperature of the samples were measured in the four seasons of the year 1377.The comparison of the obtained experimental data with the results of computer simulation revealed the extent of accuracy of the model. Using the mathematical simulation, it is possible to predict the Zayandehrood river environmental conditions in future with respect to industrial development and yearly rainfall. Thus by utilizing the model's theoretical results, one is able to make the proper decisions in order to achieve a sustainable development in the Isfahan province.  

In order to investigate the effects of drop size and its falling distance on drop interface coalescence time, the so-called coalescence cell had been designed and built. The selected chemical materials were as follows: Water-Toluene, Water normal Heptanes and 60% aqueous Glycerol-Toluene. In all experiments, the oil phase was as continuous phase. Before each run, the two immiscible phases were mutually saturated by mixing and then, separated from each other. It was indicated that, the increasing of drop diameter or drop falling distance increases drop interface coalescence time, separately. These factors also cause the subsidence of multistep coalescence phenomena in water-Toluene system, which has usually occurred severely. Added to these, the vertical or horizontal fluctuations of drop on interface, causes the delaytion of drop coalescence.

Carbides are new ceramic materials which are growingly used as substitute for metals in modern technology. Among these, . silicon carbides is more under attention. Machining of this material is very difficult due to its high hardness and resistance to wear. Among traditional methods, grinding is the most effective one. Nonetheless, specific and expensive wheels, limits on technology parameters and shapes and low machining rate are still problems. Electro-discharge machining is the best no conventional method but it shows high instability and tendency to arcing compared with machining of steels. This paper aims to investigate the causes and possible guidelines for instability reduction or elimination. When machining silicon carbide under different pulse-times and number of transistors, experimental results show that, arc discharges have the highest percentage of 60 among the other pulse types. Open circuit pulses have the least percentage of 20 and normal discharges .percentage is close to open circuit value. SEM pictures show existence of pure silicon on the machined surface which can cause instability due to its high electrical resistance. Heat generation within silicon carbide body especially near the discharge surface, may cause development of a thin dielectric vapor layer on the, surface which can be a reason for a large and fast growing plasma channel and as a result' are formation.  

Tungsten carbide is one of semented carbide materials, that are known as metal carbides. Tungsten and carbon are the elements of WC, W2C and W5C3 that is stable only at higher than 2535°C. Among above mentioned three combinations of Tungsten, the stability of mono Tungsten carbide (WC) that briefly named Tungsten carbide is better, so, it is more important. The plasticity of Tungsten carbide is high and in comparison with steel its modulus of elasticity is three times bigger. Its hardness is bigger than that of steel, even in very stream conditions, so that it doesn't make any noticeable changes at high temperature. Because of its high resistance to wear and good shock ness, it is used in various kinds of carbide dies and cutting tools. Although the machining of Tungsten carbide with conventional methods is almost successful in comparison with machining of Silicon carbide, but because of complexity of the shape of work piece and high needed accuracy in machining, the conventional machining methods cannot achieve them. Among non-conventional machining methods EDM is an especial case in machining of cemented 'carbides. With consideration of instability of ED machining of Tungsten carbide process, this paper tries to give some instructions that leads to more stable machining process. About the machining of different grades of Tungsten - carbide, open circuit pulses are more than 50% and Arcing are less than 15% with different pulse durations. In these materials, increasing the pulse duration, increases machining instability. In nearly all grades of Tungsten- carbide, the size of the grains on the machined surfaces become bigger and show the gathering much more resolidified materials. The presence of cobalt nearly on everywhere of the machined surface patterns, may be because of resolidification of cobalt and recombination with Tungsten or Tungsten - carbide on the surface of the work piece. The elimination of the cobalt and Tungsten together from arcing area, make holes which are suitable places for carbon to remain and grow and afterwards make arcing.