Yur T. Models and methods of spectral analysis of gas-turbine engine elements technical state

Object of the research is the process of diagnostics of air gas-turbine engine units (i.e. airscrew drive reduction gear of turboprop engine and fan stage of two-spool gas turbine engine). Research objective is synthesis of models, methods and computer programs for early diagnostics of engine elements and units defects initiation and propagation. Research methods are methods of wavelet analysis and fast Fourier transform - for frequency characteristic components identification; methods of mathematical statistics - for signals characterization and wavelet coefficients distribution analysis. Equipment - the personal computer. Theoretical and practical results of research - methods and models for early diagnostics of engine elements and units defects initiation and propagation are developed and formed for the first time. The investigation allowed improving performance of wear feature extraction. In the process of examination and testing methods and models demonstrated their efficiency and functionability on solving particular problems of diagnostics and processing of non- stationary signals. Scientific novelty. Method of eliminating edge effects in the results of wavelet transform performed with complex Morlet wavelet is proposed for the first time. The method differs from the existent one in use of the analytical expression developed by author to compute region of reliability for wavelet coefficients that allowed reducing distortion in the results of wavelet analysis. Method of high frequency random vibration envelope analysis obtained further development by means of complex Morlet based wavelet filter banks and cross-spectral analysis. Such an improvement gave the ability to extract more characteristic frequency components of the signal which include harmonic and impulsive components. This allows getting for engine elements more accurate indication of deviation from normal state. Statistical method of vibration signals analysis got further development by use of S-discriminant which allowed to create method of rapid engine technical state diagnostics. Models of technical state diagnosis for gas turbine engine elements are improved with wavelet analysis methods and S-discriminant which allowed detecting wear on earlier stages. Results of dissertational work are introduced on State Enterprise "Ivchenko-Progress" for analysis of dynamic vibrations of engine units and elements, in developing and refinement of air and earth gas turbine engines and during operation at customer. Results are adopted at the following courses: "Decision theory", "Decision support systems" and "High end databases and data mining" at Zaporizhzhya national technical university and "Mathematical modelling" at Zaporizhzhya national university. Scientific theoretical and practical results can be used at educational institutions, at organizations that research spectral characteristics of non-stationary signals and diagnose state of elements of rotor equipment, at enterprises that develop tools of vibration-based diagnostics, at teaching classes training engineers in the sphere of computer information processing system and control systems.