Two-level Signal Processing of Vibroacoustical Signals for Condition Monitoring of Complex Rotary System

In this paper we investigate signal processing of vibroacoustical diagnostics of the initial crack-like damage of the blades in aviation gas-turbine engines. We propose to use the Fractal analysis of the Time-Frequency Spectra contour images of vibroacoustical signals radiated by gas-turbine engine, as an additional level of signal processing to increase the informative of the damage features. The box-counting method is used to calculate Minkowski dimension. The efficiency of the box-counting method is demonstrated by the rate of change of the damage feature (fractal dimension). To conveniently compare the obtained results, they are listed in the table and shown on the diagram.