Eng During last 365 days Approved articles: 1926,   Articles in work: 331 Declined articles: 648 

Komartsova L.G., Lavrenkov Yu.N., Antipova O.V. Comprehensive approach to the study of complex systems

Published in journal "Software systems and computational methods", 2013-4 in rubric "Knowledge Base, Intelligent Systems, Expert Systems, Decision Support Systems", pages 330-334.

Resume: the main techniques for evaluate proposed variants of solutions in design and studying complex systems are simulation methods and methods of experiments scheduling. The accuracy of decision making may be increased with the use of neural network (NN) to summarize the results of simulation experiments. Predicted with such method the solutions are then tested on the simulation model. The use of genetic algorithm makes it possible to find the corresponding genetic operators that provide faster convergence for each specific task of simulation. A database of experiments allows choosing a plan of experiment according to the type of parameters of simulation model. One of the main problems that must be solved in implementation of this technique is the choice of a suitable type of a neural network that will provide high percentage of recognition and low training time. A three-layer feedforward network with a combined learning algorithm based on the use of genetic algorithm and simulated annealing algorithm was proved to be ths best neural network architecture for solving the given.

Keywords: design, research, simulation methods, experiment scheduling methods, decision making, neural network, genetic algorithm, genetic operators, a three-layer network, simulated annealing algorithm

DOI: 10.7256/2305-6061.2013.4. 10551

This article can be downloaded freely in PDF format for reading. Download article

1. Komartsova L.G. Metodika imitatsionnogo modelirovaniya vychislitel'nykh sistem
s ispol'zovaniem neyronnoy seti dlya obobshcheniya rezul'tatov. Izvestiya vuzov.-ser.
2. Komartsova L.G. Research of a Genetic Algorithms Designer // Proc. of IEEE Int. Conf. on
Artificial Intelligence Systems.-Los Alamitos, California.-2002.-P.395-397.
3. Komartsova L.G., Maksimov A.V. Neyrokomp'yutery (uchebnoe posobie).-M.: Iz-vo
MGTU im. N.E. Baumana.-2002.-320 s.
4. Miller W., Glanz F., Kraft L. CMAC: An associative neural network alternative to back
propagation // Proc. of the IEEE.-1990.-V.78.-N 10.-P.1561-1567.
5. Aved'yan E.D. Assotsiativnaya neyronnaya set' SMAS // Sbornik statey "Problemy
postroeniya i obucheniya neyronnykh setey".-M.: Mashinostroenie.-1999.

Correct link to this article:
just copy this link to clipboard