NON-PARAMETRIC ALGORITHM FOR CONSTRUCTING CLASSIFICATION STRUCTURE OF THE LEAKAGE TEST CHARACTERISTICS

The problems of correct and effective choice of the methods and means used for leakage monitoring by penetrating substances  (gases) on the basis of certain criteria are considered. Methods of  mathematical statistics are used for classification analysis of the  main characteristics of the leakage test methods and approaches in  the framework of general theoretical research to prioritize them in  compliance with current GOST 51780–2001. In other words, an  attempt has been made to search for and identify the effective ways  of ordering the classification structure from the main characteristics  of the methods and means of leak testing necessary for the  researcher at the preparatory stage of testing. This is partly due to  the lack of clear recommendations in this standard regarding the  significance of these characteristics for leakage control in real conditions and possibility of its implementation. The procedure rests on the proposed nonparametric algorithm which includes  classification of the characteristics based on pairwise clustering. The  expert evaluation method is used for effective addressing the  problem of nonparametric analysis. Solutions of multivariate analysis are given for the problem of inverse reduction of characteristics to a  small number of groups formed on the basis of generalizing features  using discriminative analysis on the basis of nonparametric criteria to check statistically significant difference. Comparatively high posterior probability of attributing the classification features to the groups distributed among six clusters is determined as a result of studying  characteristics pertaining mainly to the leakage test facilities (as the  most numerous ones compared to test methods). We also consider and assess the possibility of introducing some clarifications to refine the current standard.

1. RF State Standard GOST R 51780–2001. Non-destructive testing. Methods and means of leak testing. Order and selection criteria. GOSSTANDART of Russia. — Moscow: Izd. standartov, 2002.—4p.[inRussian].

2. Aivazyan S. A., Enyukov I. S., Meshalkin L. D. Applied statistics: Research of dependences. Reference book — Moscow: Finansy i statistika, 1985. — 487 p. [in Russian].

3. Shtremel’ M. A., Kudrya A. V., Ivashchenko A. V. Nonparametric discriminant analysis in quality management tasks / Zavod. Lab. Diagn. Mater. 2006. Vol. 72. N 5. P.53–62.

4. Lapko A. V., Lapko V. A., Sokolov M. I., Chentsov S. V. Nonparametric classification system. — Novosibirsk: Nauka, 2000. — 240 p. [in Russian].

5. Orlov A. I. Applied statistics. Textbook. — Moscow: Йkzamen, 2004. — 656 p. [in Russian].

6. Petrov P.K.Mathematico-statistical processing and graphic representation of the results of pedagogical research using information technology: a tutorial. — Izhevsk: Udmurtskii universitet, 2013. — 179 p. [in Russian].

7. Stukach O. V. The program complex STATISTICA in the decision of tasks of quality management: a tutorial. — Tomsk: Izd. Tomsk. Politekhn. Univ., 2011. — 163 p. [in Russian].

8. Sidorenko E. V. Mathematical methods of processing in psychology. — St. Petersburg: JSC “Rech’”, 2002. — 350 p. [in Russian].

9. Ermolaev O. Yu. Mathematical statistics for psychologists: Textbook. 2nd edition. — Moscow: Flinta, 2003. — 336 p. [in Russian].

10. Gromova N. M., Gromova N. I. The foundations of economic forecasting. Tutorial. — Moscow, 2006. — 80 p. [in Russian].

11. Buyal’skaya A. K., Kleimenova N. A., Gorevaya M. I. Marketing Research: textbook for universities. — Novosibirsk: Sibprint, 2010. — 404 p. [in Russian].

12. Durant B., Odell P. Cluster analysis / E. Z. Demidenko (transl.), B. Ya. Boyarskii (ed.). — Moscow: Statistika, 1977. — 128 p. [Russian translation].

13. Borovikov V. The art of data analysis on the computer: for professionals. 2nd edition. — St. Petersburg: Piter, 2003. — 688 p. [in Russian].

14. Kulaichev A. P. Methods and tools for data analysis in Windows STADIA. 4th edition. — Moscow: Informatika i komp’yutery, 2002. — 341 p. [in Russian].

15. Simchera V. M. Methods of multivariate analysis of statistical data: a tutorial. — Moscow: Finansy i statistika, 2008. — 400 p. [in Russian].

16. Kim J.-O., Mueller Ch.. W., Klecka W. R., et al. Factoril, discriminant and clattery analysis / I. S. Enyukov (ed.). — Moscow: Finansy i statistika, 1989. — 215 p. [in Russian].