In recent years, with the advancement of industry and technology, we see that systems are becoming more and more complex. This complexity in the industry has required a parallel progress in control systems (controllers), which has multiplied the need for advanced and intelligen t controllers. One of the most important criteria in the design of any control system is the accurate knowledge of the system or, more precisely, the modeling of the system. Considering these two challenges, this paper examines intelligent systems and specifically neural networks from the perspective of classical theory and their application in modeling and control of complex systems. At first, the basic element of neural networks, i.e. neuron, is introduced and its types of models (collective and radial) are also presented. Then the types of neural networks such as multilayer perceptron neural networks (MLP-NN), radial basis neural networks (RBF-NN) and recurrent neural networks (RNN) are described and the appropriate number of layers and also the appropriate number of neurons in the hidden layers in neural networks for different applications and especially for the approximation of nonlinear functions (modeling) are discussed. Then, it is tried to build a bridge between the concepts of neural networks (intelligent world) and the concepts of classical world and analyze neural networks from the perspective of classical theory. It is shown that from the point of view of classical theory, a neural network can be considered as a model structure and its weights and biases as unknown parameters of this structure. In neural networks, learning algorithms are used to determine the unknown parameters of the network (weights and biases). Therefore, learning algorithms of neural networks are studied from the point of view of classical theory and specifically in relation to numerical optimization and a bridge will be built between learning algorithms and numerical optimization methods. Also, the most important neural network learning algorithms are introduced along with their advantages and disadvantages. In the end, two important applications of neural networks, modeling and control, are discussed and for each of these applications, several illustrative examples are presented and simulated to show the effectiveness of neural networks.

The delay in the supply of gasoil has wide Consequences of political, social and economic. Therefore, the gasoil demand forecasting is very important. The use of artificial neural networks in forecasting has many applications. Appropriate design of ANN parameters enhances performance and accuracy of the neuralnetwork model. Trial and error method is used to setting the ANN parameters in the most of studies. Trial and error method is not a reliable solution to achieve the best ANN structure. In the present study, the optimum structure of ANN to forecasting the demand of gasoil in the province of Hormozgan determined by Taguchi experimental design method. Analysis of variance (ANOVA) of the ANN parameters indicates that contribution of the number of neurons in first hidden layer to the changes in the network mean square error (MSE) is about 41% and contribution of the learning algorithm is about 27%. 
Also, the results of this study show that artificial neural networks are designed using Taguchi experimental design method have better performance than other Networks.

Investment in common stocks is one of most beneficent alternatives in capital markets, therefore forecasting of stock price is great significance for shareholders; however the complexity of stock markets. Most studies revealed stock price in Tehran stock exchange has not random walk pattern. Besides, with regard to nonlinear and chaos system in stock market that eclipsed of political, economical and psychological conditions, we can use nonlinear artificial systems such as artificial neural networks (ANN), fuzzy neural networks (FNN) and genetic algorithms (GA’s) for forecasting stock price. Two goals are aimed in this research: 1- Design and present a model for forecasting stock price, using FNN and GA (GFNN). 2- Reduction the forecasting error of GFNN stock price forecasting model, in compare with mere ANN. At first, for examination of mentioned goals, multi layer perceptron (MLP) of ten listing companies in Tehran stock exchange has been designed. Then, after design and implementation of GFNN, the results of two models, have been compared and hypotheses have been investigated by using error evaluating criteria (MSE, R2, NMSE and MAPE). The results of research show, in compare with mere ANN, GFNN has perfect predictions and has higher speed and more strong estimation for stock price forecasting.

Today, stock investment has become an important mean of national finance. Apparently, it is significant for investors to estimate the stock price and select the trading chance accurately in advance, which will bring high return to stockholders. In the past, long-term trading processes and many technical analysis methods for stock market were put forward. However, stock market is a nonlinear system, due to the political, economical and psychological impact factors. Thus, it is difficult for us to use traditional analysis tools to make stock transaction decision. With the developmenting of nonlinear methods such as neural networks and fuzzy neural networks, we can now use these methods for stock price forecasting. In this research, we presented three scenarios: 1) stock price forecasting with classical methods approach, 2) stock price forecasting with artificial intelligence methods approach, and 3) stock price forecasting with hybrid model. Therefore, first, we designed classical models such as exponential smoothing, trend analysis, and ARIMA, then artificial intelligent models such as neural networks and fuzzy neural networks were designed, next the third scenario i.e., hybrid model, was presented. Finally, the scenarios were measured. The experimental result showed that the hybrid model had more accuracy than the classical or artificial intelligent models. The hybrid model enjoyed also such properties as fast convergence, high precision and strong function approximation ability, there¬fore, it is suitable for real stock price forecasting.

Decision making by managers of different organizations is not something merely quantitative but it needs synthesizing and processing quantitative data and qualitative knowledge. With regard to the nature of the problem, dynamic behavior of managers, and complexities of strategic decision making, artificial neuron networks may be a good help for decision makers. These networks are helpful since they copy and represent the natural ones. The value of neural network technology lies in its application for recognizing patterns, learning, classification, generalization, summarizing and interpreting incomplete inputs. Neural networks are potential factors for providing some human characteristics which are used in problem solving. Simulation using logical and analytical techniques and decision support systems or even expert systems is difficult. The aim of this paper is, as well as presenting new findings, to describe the application of artificial neural networks in strategic decision making by managers. In this paper, the artificial neural networks are discussed to use a decision model within comparative strategic decision making in management. It describes the way networks are applied and accepted within strategic decision making. In chapter one the literature of artificial neuron networks is reviewed. In the second chapter, the principal idea of strategic decision making framework and, on the third, how to make a decision model based on artificial neuron networks are discussed. On the whole, the main purpose of the paper is to understand the human natural neuron networks and to reconstrnct this function artificially to use it in strategic decision making. In summary, artificial neuron networks learn from experience. They are used for making complex and repetitive decisions. They focus on the known patterns and are able to decide and predict on the basis of the previous instances. In future, decision making techniques of artificial network will be applied for all strategic cases and for every strategic case a unique artificial network will be planned and different decisions will be made.

In construction project management, success in the controlling of projects’ final cost .Also duration is a critical issue that can help managers. The main purpose of the management of construction projects is to ensure that the projects can be performed according to the defined budget. Cost and duration overrun may lead to profits cut and sometimes even can cause project failure. To solve this problem, this study employs the earned value technique and the neural networks to form models EAC-ANN and EAC (t) -ANN for final cost and duration prediction. For input variables in neural network, two sets of variables have been employed, including the earned values variables and composite ones. Composite variables also include the earned value variables and environmental ones. Then the performance of the earned value techniques in predicting final cost and duration is considered. Moreover, the influence of each data set in neural network models on the cost and duration prediction performance is compared separately to yield a better model. The results show that the neural networks outperform earned value management technique. Furthermore, using a neural network based on the composite data can lead to better performance than the one based on the earned value data.

The use of artificial neural networks (ANN) in forecasting has many applications. Appropriate design of ANN parameters enhances the performance and accuracy of neural network models.  Most studies use a trial and error approach in setting the value of ANN parameters. Other methods used to determine the best structure of a neural network only use a single evaluation criterion to determine the appropriate structure. In this study, the authors provide a new method to design the network structure. In this method, we use a combination of Taguchi experimental design and TOPSIS methods, to determine he optimal ANN structure, taking into account three evaluation criteria simultaneously. The  estimated demand for gasoline in the Hormozgan province produced using this method, confirms its efficiency and effectiveness. Analysis of variance (ANOVA) of the ANN variables indicates that contribution of the number of neurons in the first hidden layer to the changes in the network performance is about 54% while the contribution of the learning algorithm is about 27%.