Miski-Oglu A.G., Shlapak N.S., Goroshko O.V.

PHEI " Pryazovskyi State Technical University"

Modern intelligent methods

in managing the organization

Modern information technologies include methods of the solution of tasks, which are based on algorithms and actions to a greater or lesser extent, the intellectual activities of man, his evolution, everyday behavior. Therefore, we can say that there is full compliance with the terms "modern" and "intelligent" with respect to information technology.

Modern information technology in management tasks include the following guidelines [1]:

• Artificial Neural Networks (ANN);

• Fuzzy Logic (NL);

• Genetic algorithms (GA);

• Nonlinear Dynamics (ND).

Let us consider in more detail each of the directions, and its contribution to the solution of urgent problems of management.

Artificial neural networks. ANN is used to solve problems that can not be precisely defined. The word "neuron" used because much of the theory INS comes from the neurobiology, although in reality the study subjects are not considered real network of biological neurons. Modeling brain function - a completely different scientific sphere, but from this area to the theory of ANN has brought some of the biological analogy.

Application of ANN helps draw in information technology the methods of information processing that are typical of highly biological systems, in particular the principle of parallelism. In contrast to the use of computers in traditional von Neumann architecture in ANN and computers constructed on their basis, performs the role of programming learning, which refers to the change of state neurons themselves and relationships between them.

In the field of management to problems that can be solved with the aid of the ANN are the problem of classification and ranking of companies, firms, construction of bank ratings, sales forecasting, change of the exchange rate.

Fuzzy logic. The main idea of fuzzy logic is that of intellectual reasoning, based on the natural language of man, it can’t be described in terms of traditional mathematical formulas.

NL entity proposed L. Zadeh, is reduced to the following points:

• it uses linguistic variables (instead of the usual numeric variables or in addition to);

• simple relationships between variables are described by the non-clear sentences;

• complex relationships are defined fuzzy algorithms;

Of management tasks that can be solved by means of systems NL, you can divide the class of problems of risk management, where the fuzzy input variables required to obtain a quantitative characterization of the output quantity.

It should be noted that to date features the offered the first unit is not properly evaluated by specialists managers. In the conditions of market economy with typical uncertainties in this situation NL, in essence dealing with "blurred" by some interval values is the best suited to describe such phenomena.

Genetic algorithms. In the last thirty years, became interested in the problem, the solution of which is based on the principles of evolution and inheritance of traits. Methods of evolutionary computation (EC) is often used to describe the evolution of programs or functions (genetic programming), finite automata (evolutionary programming) and systems based on the reduction rules (classification systems). Sometimes EC together with fuzzy logic are used to train the neural network, which leads to a new term, "soft computing", combining the GA, NL and ANN.

Genetic algorithms are applied to solve the optimization of tasks using the method of evolution, i.e. through the selection of the set of solutions, a most suitable.

The objectives of management, are addressed through the GA can assignment of the optimal transportation planning, defining the best trading strategy, accommodation capacities.

Nonlinear dynamics. The most interesting, in our opinion, represents the use of practical management methods of nonlinear dynamics. The fact that the ND and the chaos of everyday human experience plays a greater role than precise predictability. It turned out that the conventional deterministic equations describing some economic model, under certain conditions lead to chaotic phenomena. Many financial time series are chaotic, to analyze and evaluate which you want to use special techniques. In these cases, there is a unique opportunity for the management of short-term forecasting.

Mathematical chaos theory, which is one of the areas of non-linear dynamics, can explain complex economic phenomena and to develop a basis of decision-making in such situations.

Approach to solving economic problems is based on the systematic use of fractal analysis.

In the economic system there are often non-linear phenomena of chaos: small changes in initial conditions or any parameters lead to the rapid changes in the dynamic behavior of the observed processes, there are sets of values that tend to process performance under different conditions, structural changes caused by small shifts the parameters lead to fluctuations.

The basis of the nonlinear dynamic approach is the consideration of the internal peculiarities of the system as opposed to statistical methods, in which all factories rely accidental or undetermined. Research in ND show that, even if it is possible to find statistical regularities of control actions on the process, to predict the behavior of the system with reasonable certainty is not possible. Essential to the relevant provisions of some economic theory is the recognition of the endogenous nature of economic cycles, i.e. a deliberate focus on the internal dynamics of the system. Prominent among these is the theory of chaotic systems [2, 3].

Chaos theory and nonlinear dynamics are new concepts, the application of management tasks which would be more well-grounded decisions and develop better strategies.

An example can be cited processes in the capital markets under the influence of objective economic conditions and subjective decisions of market participants, as well as methods of long-term prognosis for the markets of stocks, bonds, currencies.

Modeling tools of information systems. Development, implementation and further successful maintenance of information systems depends on the construction of adequate mathematical models reflecting the specifics of a given system.

The market modeling tools comprises a very wide range of different kinds of applications, because of standard and specialized.

In our opinion, the most appropriate application of this kind is by the system MATLAB (MATrix LABoratory). It was created by those skilled MathWorks company as high level programming language for on the technical computing.

One of the most important advantages of MATLAB is the possibility of expansion to address new scientific and engineering problems. This is achieved above all the creation of a number of expansion packs, covering many new and practically useful directions of computer mathematics [4]. These expansion packs are:

• expansion pack on neural networks Neural Networks Toolbox;

• The package Fuzzy Logic Toolbox;

• Optimization Pack Global Optimization Toolbox.

These packages provide developers with new information systems for a wide range of design, modeling of, training and use of these systems.

Prospects for future research. We have considered above approach to solving problems of management, based on the systematic use of fractal analysis. There are many other paradigms that may prove to be more useful than fractals or chaos [2]. These alternative methods are also closely related to the theory of chaos. They are relatively new development of, which is only now beginning to recognize.

The first of these approaches is called the theory of wavelets [5], it was a generalization of the spectral analysis. This analysis is based on the Fourier transform, to decompose a signal into a number of sine waves, which, when put together, reconstruct the original signal.

Since wavelet theory can work with multi-scale signals, it can be used for analysis of fractal and chaotic time series. This could be a promising area of future research.

The second approach is the concept of "self-organized criticality." It provides significant opportunities in marketing tasks (analysis of capital markets).

Self-organized criticality has been useful in modeling of earthquakes and other natural phenomena, because natural systems tend to come at all times in critical states. In other words, they are far from equilibrium.

   Self-organized criticality promising, because it proposes a physical model for the fractal statistics. It may be a very fruitful area for future research.

Finally, unlike the chaos theory, the theory of self-organized criticality gives hope for predictions. Self-organized systems "weakly chaotic", which means they are on the edge of chaos. Their close trajectories diverge not exponentially, and in accordance with the power by law. This means that weakly chaotic systems do not have the time scope, outside of which the prediction would become impossible, thus allowing for the possibility of long-term prognosis.

List of literature:

1. Krichevsky M.L. Intellectual methods in management. - SPb.: Peter, 2005. – 304 p.

2. Peters E. Chaos and order in the capital markets. – M.: Mir, 2000. - 333 p.

3. Nikulchev E.V., Volovich M.E. Chaos models for the processes of change in the stock price/ / Exponenta Pro. Mathematics in applications. - 2003. - № 1. - P.49-52.

4. Dyakonov V., Kruglov V. Mathematical expansion packs MATLAB. A special handbook. - SPb.: Peter, 2001. – 480 p.

5. Dyakonov V. Signal and image processing. Special reference. - SPb.: Peter, 2002. - 608 p.