Vibro-forecasting of fault development in hydropower units
Автор
Kukharchuk, V. V.
Katsyv, S. Sh.
Hraniak, V. F.
Kucheruk, V. Yu.
Madiarov, V. G.
Bykovsky, S. A.
Khassenov, A. K.
Karabekova, D. Zh.
Кухарчук, В. В.
Кацив, С. Ш.
Граняк, В. Ф.
Кучерук, В. Ю.
Мадьяров, В. Г.
Биковський, І. А.
Хасенов, А. К.
Карабекова, Д. Ж.
Дата
2018Metadata
Показати повну інформаціюCollections
- Наукові роботи каф. ІРТС [779]
Анотації
Proposed in the paper are the mathematical models and algorithms for vibroforecasting of fault development
in hydropower units, such models and algorithms being based on the spectral analysis of vibration signals.
With real-world vibroacoustic signals being of purely non stationary nature, i.e. their spectr a varying with
time, the authors proposed to carry out spectral analysis using discrete wavelet transformation resulting in acquisition of three-dimensional amplitude-frequency-temporal spectrum in the form of wavelet coefficient matrix. It should be noted that, due to an exceptional complexity of the hydropower unit as a dynamic
hydroelectromechanical system and practical impossibility of mathematical description of dependence between the vibroacoustic signal and all factors that cause vibration, it makes sense to treat the hydropower unit
as the «black box», that is to simulate its external functioning rather than its structure. That is why, for the
purposes of forecasting of fault development in hydropower units, generation of three-layer artificial neurallike network is stipulated. Such fore casting is based on analysis of trends in wavelet coefficients in each of
frequency bands of vibroacoustic signal’s amplitude-frequency-temporal spectrum. Further on, stipulated is
the breakdown of these trends into vibration signals’ components (background, hydrodynamic,
electrodynamicetc.) that correspond to particular factors that cause vibration. Processing of these components
allows determining there sulting prognostic conclusion with regard to fault development in hydropower units
as avariety of values representing the probability levels of different vibration factors. Besides, the paper contains an example of such prognostic conclusion obtained on the basis of historical values of vibroacoustic
signals obrained from vibration monitoring subsystem of DnestrHPP-2.
URI:
http://ir.lib.vntu.edu.ua//handle/123456789/35038