Using fractal analysis of the time-frequency spectra of vibroacoustical signals for diagnostic of gas-turbine engines
Journal Title: Вісник НТУУ КПІ. Серія Радіотехніка, Радіоапаратобудування - Year 2018, Vol 0, Issue 74
Abstract
The article is devoted to the improvement of signal processing methods of complex vibroacoustical signals for the diagnosis of initial crack-like damage in the blades of aircraft gas-turbine engines during operation. The low-frequency vibrational and acoustic noise in the range 0-10 kHz is used as diagnostic information, which is emitted by the engine during operation. Initial crack-like damage in the blade does not cause an increase in the overall level of vibroacoustical signals or their components. When the occurrence and initial propagation of damages change the signal structure, new components appear that are characterized by low energy capacity. The following signal processing methods are used in order to abstraction such components: time-frequency analysis, polyspectral (high-order spectral) analysis, scale-time analysis. However, the results of such signal processing are often quite complex for interpreting, comparing and deciding about the technical condition of the testing object. We propose an additional level of processing of diagnostic information, based on the methods of fractal analysis in order to increase the diagnostic value of the time-frequency spectra. The results of physical modeling and frequency-time analysis of vibroacoustical signals are presented. For this purpose, experimental studies of the forced vibrations of the physical model (turbine imitator) of the turbine are carried out under steady-state and non-steady-state vibration excitations. Two technical conditions of the turbine imitator are investigated: defect-free and the presence of an initial crack-like damage in one blade. We use the a time-frequency analysis based on Wigner-Wille pseudo-distribution to signal processing of vibroacoustical signals, which are emitted by a rotating turbine imitator during different excitation modes. The results of the time-frequency analysis are presented in the form of two-dimensional contour images characterizing the dependence of spectral estimates on the normalized frequency and time. At the second signal processing level, we determine fractal box-counting dimension (Minkowski dimension). Minkowski dimension is an integral numerical index that characterizes the geometry of the contour image, and allows to discriminate the turbine imitator conditions during operation at the different modes of vibrational excitation. We propose to use the Minkowski dimension as a diagnostic feature of a crack of the turbine blade. It is established that the Minkowski dimension is more sensitive to the occurrence of damage, in the case of its determination, not for full images, but for separate parts of images in different frequency ranges.
Authors and Affiliations
N. Buraou, S. Ignatovych, O. Pazdrii
Keywords
Related Articles
- EP ID EP392101
- DOI 10.20535/RADAP.2018.74.73-83
- Views 61
- Downloads 0
Сriteria of crystal-like structures approaching by impedance delta-inhomogeneities lattices
Introduction. Crystal-like structures (CS) have unique, similar to crystals, band spectral characteristics and form the basis of new various signal processing devices. In this paper the criteria of CS approaching by δ-in...
Radiometric modulation measuring device of intensity of optical radiation
Introduction. The paper proposed a block diagram of radiometric measuring the intensity of optical radiation from modulation conversion parameter input . To assess the intensity of use periodically comparing the referenc...
Photometric Absorbance Spectrum Analyzer of Slightly Transparent Biomaterials
Introduction. The measurement of optical radiation, especially the intensity of the light flux passing through biological tissues or other slightly transparent (cloudy) materials and the environments is a challenging pro...
Influence of the Matrix Structure of the Modulator and Detector on the Optical Spectrum Analyzer Output Signal
In this article, we investigate the physical and mathematical model of a coherent optical spectrum analyzer (COSA), which uses a matrix light modulator and a matrix detector as input and output devices. This model allows...
Segmentation of anatomical heart structures in Ensite Verismo software for radiofrequency ablation of arrhythmogenic tissues
Review of the problem. The relevance of the research is the need of a high-quality imaging of the left atrium anatomy for the effectiveness of radiofraquency ablation of arrhythmogenic tissue of the myocardium in real ti...