INFORMATIVITY'S ANALYSIS OF WAVELETT AND GRAPHIC REPRESENTATIONS OF VIBRATION SIGNAL IN THE BEARINGS DIAGNOSTICS PROBLEM

А.А. Obleuhov, D.S. Shalaev, T.P. Gryzlova

The new systems of diagnostic features based on the structural analysis of vibration signals are presented. In one of the system vibration signals representations are building, afterwards the conceptual analysis of the image is making. Representative elements characteristics of the image are the basic system of features. The second system formed on the set of signal complexity functions characteristics and Haar Wavelet-decomposition. Where the analysis of decomposition results based on complexity function is performed. The results of features informativity estimation based on the analysis of wavelet and graphic representation in bearing diagnostics of GTE problems and in bearings reliability tests are presented. Features with much more informativity than informativity of features in well-known methods of bearing diagnostics are selected. The informativity of the founded features allows building linear classifiers with simple decision rules.

Key words: features informativity analysis, representative elements, systems of the conceptual description of images, Wavelet-coefficients, complexity function, bearing diagnostics of GTE.