The peer-reviewed scientific and technology journal. ISSN: 0204-3327

# Analisys of safety system pump condition based on their testing results

### UDC: 621.039; 62-932.4

The method for analyzing the condition of emergency system pumps based on their periodical testing results is presented in the paper. The method and the algorithms are based on the presentation of the testing results in the space of the principal components. Such an approach enables one to show the pump condition in a convenient form. The parameter variation measured from the beginning of the test until the steady state condition is achieved, i.e. the dynamic section of the curve for each parameter, is used for the analysis.

Comparing the behavior curves of different technological parameters as a time function of a particular pump for different tests one can see that some sections of these curves do not change from test to test. This simply means that these sections are not informative relative to extraction of the information concerning the defect formation. These sections should be classified as some kind of «noise» and should be excluded as providing little information. On the contrary, the sections with abnormal behavior of technological parameters are more informative, and we take these sections for further analysis.

As a measure of the system uncertainty entropy H(X) is used. This new parameter is defined by the relationship where pi – is the probability of the i-th state of the system; N – is the total number of the states of the system.

The entropy enables us to describe the probabilistic spread in the measured data. The entropy has the maximum value if all the states of the system are equiprobable. We can use this feature of the entropy to choose the more informative time intervals of the dynamic behavior of the technological parameters. The smaller the entropy, the more probable certain states of the system are. So, the most informative are those time sections, which have the maximum entropy value, i.e. the time sections for which the maximum spread in the measured data is observed. Using this approach a matrix is constructed based on the time intervals with maximum entropy, the so-called matrix of informative criteria.

To describe the condition of the pump using the different technological parameters are measured in the course of the tests we have to normalize the values of the parameters to the root-mean-square deviations of the parameters. The normalized data are then used for the transformation of the original data matrix on the basis of the most informative criteria with a statistical method known as the Karhunen-Loeve transform, which is also known as a principal components method.

The approach is applied to processing the testing results of the emergency system pumps of the Kalinin NPP (Russia). Interesting results are obtained.

### References

1. Iserman R. Process fault detection based on modeling end estimation methods – a survey. Automatica. 1984, v.20, no. 4, pp. 387-404.
2. Basseville M. Detecting changes in signal and systems – a survey. Automatica. 1988, v. 24, no. 3, pp. 309-326.
3. Frank P.M. Fault diagnosis in dynamic systems using analytical and knowledge-based redundancy – a survey and some new results. Automatica.1990, v. 26, no. 3, pp. 459-474.
4. Reisen C., Marshall E. Evaluating operator support system in realistic conditions at HAMMLAB. Nuclear Engineering International. 1988, v. 33, no. 402, pp. 39-41.
5. Transactions. ANS.1982, v. 41, pp. 524-528.
6. Abagyan A.A., Dmitriev V.M, Klebanov L.A., Kroshilin A.E., Larin E.P., Morozov S.K. Monitoring and diagnostics systems for nuclear power plant operating regimes. Atomnaya energiya. 1987, v. 63, pp. 311-315 (in Russian).
7. Vapnik V., Chervoninkis A. Pattern Recognition Theory. Moscow. Nauka Publ., 1974 (in Russian).
8. Tao Gu, Tou J.T. A new criterion for optimal classification. Pattern Recognition, 1982, no. 2, pp. 1063-1065.
9. Leskin S.T. Algorithm development for abnormality detection of NPP equipment conditions based on technological testing results. Izvestya vusov. Yadernaya energetika. 1997, no. 4, pp. 4-12 (in Russian).
10. Herbert M.R. A review of on-line diagnostic aids for nuclear power plant operators. Nucl. Energy. 1984, v. 23, no. 4, pp. 259-264.
11. Pavelko V.I. A review of application of expert system methodology in nuclear power engineering. Atomnaya energiya. 1990, v. 11, pp. 1-8 (in Russian).
12. Weiss S., Reagan W., Roe J. Experience with operator aids for nuclear power plants in the USA. In: Proc. Intern. Conf. on Man-Machine Interface in Nuclear Industry. Tokyo, 15-19.02.1988, Vienna, 1988, pp. 323-329.
13. Urig Robert E. Potential application of nuclear networks to nuclear power plants. Proc. Amer. Power Conf. Vol.53. Pt.2 53rd. Ann. Meet., Chicago, III., Apr. 29-May 1. 1991, pp. 946-951.
14. Fault diagnosis in dynamic systems. Theory and applications. Edited by Patton R., Frank P., Clarc R. Prentice Hall Inc., Englewood Cliffs, NY, 1989.
15. Willsky A.S. A Survey of design methods for failure detection in dynamic systems. Automatica. 1976, v. 12, pp. 601-611.
16. Demidovich B.P., Maron I.A. The basic principles of computational mathematics. Moscow. Nauka Publ., 1960, 664 p. (in Russian).
17. Ventzel E.S. Theory of probability. Moscow. Vysshaya Shkola Publ., 1999 (in Russian).
18. Fukunaga K. Introduction to statistical pattern recognition, Academic Press, New York and London, 1972.
19. Tu J, Gonsales R. Pattern Recognition Principles. Moscow. Mir Publ., 1978 (in Russian).
20. Leskin S.T., Slobodchuk V.I., Shelegov A.S. Analysis of VVER-1000 main circulation pump condition under operation. Izvestiya vusov. Yadernaya energetika. 2016, no. 4, pp. 12-22 (in Russian).