ANALYSIS AND DATA PROCESSING SYSTEMS

This paper considers the design stage of the software development process to carry out cluster analysis and classification based on artificial neural networks with a preliminary reduction of attribute space dimension also based on artificial neural networks and additionally on factor analysis. The design stage is one of the most important software development stages. Most modern software systems contain a great variety of cooperating program modules which in its turn increases the complexity of the design process. At this stage software developers use various methods. This article describes a well-known method of system modeling based on the Petri net which has been widely used in various subject areas. A Petri net with 48 vertexes was created for the developed software complex.

The realized Petri net is analyzed. The analysis provides insights into the behavior of the developed program.

The Petri net analysis was performed in the CPN Tools software environment (version 4.0.1). This environment has automated analysis tools and generates reports on deadlock states and deadlock state transitions. The analysis conducted in this environment, inherently solves one of the main problems of the Petri nets theory, namely the reachability problem.

High-quality analysis contributes to the adjustments in the software architecture at the design stage.

The analysis results are the estimation of reachable markings, identification of deadlock states that lead to a system’s shutdown, looping and detection of scripts that are not participating in the system. One of the key results is the reachability graph. The analysis has shown that there are no deadlock transitions, which means that the software architecture as been designed correctly.

1. Kotov V.E. Seti Petri [Petri nets]. Moscow, Nauka Publ., 1984. 160 p.

2. Reisig W. Petri nets: an introduction. Berlin, New York, Springer-Verlag, 1985. 172 p.

3. Kharakhinov V.A., Sosinskaya S.S. Programmnyi kompleks "Analiz eksperimental'nykh dannykh na osnove neironnykh setei". The certificate on official registration of the computer program. No. 2017618294, 2017. (In Russian, unpublished).

4. Haykin S. Neural networks: a comprehensive foundation. 2nd ed. Upper Saddle River, NJ, Prentice Hall, 1999 (Russ. ed.: Khaikin S. Neironnye seti: polnyi kurs. 2nd ed. Moscow, Villiams Publ., 2006. 1104 p.).

5. Harman H.H. Modern factor analysis. 2d ed., rev. Chicago, University of Chicago Press, 1967 (Russ. ed.: Kharman G. Sovremennyi faktornyi analiz. Moscow, Statistika Publ., 1972. 484 p.).

6. Lawley D.N., Maxwell A.E. Factor analysis as a statistical method. London, Butterworths, 1963 (Russ. ed.: Louli D., Mansvell A. Faktornyi analiz kak statisticheskii metod. Moscow, Mir Publ., 1967. 141 p.).

7. Hinton G.E., Zemel R.S. Autoencoders, minimum description length and Helmholtz free energy. Advances in Neural Information Processing Systems, 1994, vol. 6, pp. 3–10.

8. Markov A.V., Voevoda A.A. Analiz setei Petri pri pomoshchi derev'ev dostizhimosti [Analysis of Petri nets by trees reachable]. Sbornik nauchnykh trudov Novosibirskogo gosudarstvennogo tekhnicheskogo universiteta – Transaction of scientific papers of the Novosibirsk state technical university, 2013, no. 1, pp. 78–95.

9. CPN Tools: website. Available at: http://cpntools.org/ (accessed 11.12.2018).

10. Peterson J.L. Petri net theory and the modeling of systems. Englewood Cliffs, NJ, Prentice-Hall, 1981 (Russ. ed.: Piterson Dzh. Teoriya setei Petri i modelirovanie sistem. Moscow, Mir Publ., 1984. 264 p.).

11. Romannikov D.O., Markov A.V. Ob ispol'zovanii programmnogo paketa CPN Tools dlya analiza setei Petri [The review of works describing sharing UML diagrams and Petri nets at a design stage of software]. Sbornik nauchnykh trudov Novosibirskogo gosudarstvennogo tekhnicheskogo universiteta – Transaction of scientific papers of the Novosibirsk state technical university, 2012, no. 2, pp. 105–116.

12. Girault C., Valk R. Petri nets for systems engineering: a guide to modeling, verification and applications. New York, Springer, 2003. 628 p.

13. Kharakhinov V.A., Sosinskaya S.S. Issledovanie sposobov klasterizatsii detalei mashinostroeniya na osnove neironnykh setei [The visualization methods for cluster analysis results of mechanical engineering components based on neural network]. Programmnaya inzheneriya – Software Engineering, 2017, no. 4, pp. 170–176.

14. Kharakhinov V.A., Sosinskaya S.S. Vliyanie sokrashcheniya razmernosti prostranstva priznakov na rezul'taty klassifikatsii list'ev razlichnykh vidov rastenii [The effect of dimension reducing on classification results of leaves of various plant species]. Programmnaya inzheneriya – Software Engineering, 2018, no. 2, pp. 82–90.

15. Kharakhinov V.A. Geneticheskii algoritm kak al'ternativa obucheniya sloya Kokhonena [The genetic algorithm as the alternative method for training Kohonen layer]. Informatsionnye tekhnologii – Information Technologies, 2018, no. 10, pp. 642–648.