Today, the most common method to estimate centrifugal pump unit condition is vibration analysis which analyze vibration signals on the unit’s surface. In that case spectral analysis of signals is used, in which harmonic components of the spectrum are placed in correspondence with particular defects. However, usage of spectrum analysis requires precise information on the type and construction of the unit, its operation practice which might be imprecise or even absent. That might lead to condition estimation errors. Thus, a task of investigating new diagnostic methods using algorithms invariant to type and construction of the unit is urgent.
The paper considers using in diagnostics of centrifugal pump unit one of non-linear dynamic methods - fractal dimension, in which Hurst exponent of fractal dimension is used as a diagnostic feature. The main advantage of the feature is invariance to the level of analyzed signal, and, consequently, independency of power and dimension characteristics of the unit.
The paper reveals the results of researches on using the Hurst exponent of fractal dimension for determining unit’s cavitation carries out by means of the COMPACS®-RPG bench system for pump units testing. It is shown that the exponent’s response is accurate, and, in some cases, even more accurate than the main cavitation spectral feature – level of blade harmonics.
Thus, usage of the Hurst exponent of fractal dimension allows to improve reliability of output from an expert system for pump units condition diagnostic.
Kostyukov V.N., Boychenko S.N., Pavlenkov D.V. Usage of fractal dimension coefficient for vibration analysis of centrifugal pump units’ condition // XX Russian scientific conference on NDT and technical diagnostics: proceedings – Moscow: Spectr Editorial, 2014. – p.352-353.
Download the publication
You can order the translation by e-mail: email@example.com