Model for Blade Diagnosis in a Working Rotor Machine Employing the Method of Virtual Elimination of Stochastic Environment

Abstract

This paper presents the bases of a new method of monitoring technical condition of turbomachine blades during their operation. The method utilizes diagnostic models such as a quotient of amplitude amplification and a phase shift of a diagnostic signal y(t) which is a result of blade operation as well as a signal x(t) of blade environment while a blade tip approaches a sensor, amplitude amplification and phase shift of these signals while the blade tip moves away from the sensor.The adopted diagnostic models indirectly take into account the existing environment of a blade, represented by the signal x(t), without the need to measure it [13,15]. Thus, the model is sensitive to the changes in technical condition of blades and practically insensitive to a change in environment. The suggested method may prove very important in diagnostics of rotor blades during turbomachines operation (compressors, turbines etc.).

Parametric and Nonparametric Diagnostic Models for Blades in the Rotating Machinery With Environment Elimination

“…The second parametric diagnostic model bases on cross-power spectral density functions of signals y and x can be written as [2,4,11,12…”

“…The signal y is measurable, signal x can be accepted as any distribution of x=δ(t,τ) [2,4,11,12]. Same as for the model based on the auto power spectral densities, were conducted the same research for model based on cross power spectral densities.…”

“…This model also has been verified in the laboratory. [4,11,12] Above models are complex variable "s" with a form referring to complex transfer functions, which can be transformed to a spectral transfer function variable jω. The real part of the power spectral transfer function P(ω) can be used to determine the impulse g(t) and step h(t) response of the blades in the rotating machinery during its operation.…”

Verification of Diagnostic Thresholds with Elimination of Ambient Factors / Weryfikacja Progów Diagnostycznych z Eliminacją Otoczenia

“…Similarly to the previous model this one is also established with no need to measure signals x(t) received from the ambient environment. To determine the , and signal one has to apply the distribution law in the form of the , function since it is easy to demonstrate that the product of mutual power density for the y signal and the x signal is insensitive to signals x arriving from the ambient environment, thus it sufficiently eliminates impact of the real surroundings for the model [6,7,10,11,12,17]. Such course of proceeding can be easily adopted to analysis of data sets collected from measurements of tribologic diagnostic signals that are used to establish diagnostic thresholds [15,20].…”

“…So far, the ambient environment has been defined by total hours of flying time or, even better, by means of the vibroacoustic signal A/s 2 . An example for elimination of ambient impact to the diagnostic process is the method that is used to diagnose machine blades during operation of rotating machinery, when a blade is monitored during a very short time when the blade is in the close vicinity of a sensor [6,7,10,11,12,14]. The current practice demonstrates that the problems associated with diagnostic of turbine blades during operation of rotating machinery are really sophisticated when merely one measurable signal y(t) is available for diagnostics and that signal is subjected to interferences, which makes the situation even worse.…”