Abstract
According to nonlinear characteristics of vibration signals measured on the turbine used in the aircraft environment control system (ECS), the ensemble empirical mode decomposition (EEMD) together with fractal dimension analysis is investigated in the paper to extract characteristic quantities for the goal of fault diagnosis of turbine bearings. Firstly, in order to filter noise signal vibration and advance signal‐to‐noise signals under different statements of bearings, including normal bearing, inner ring fault, outer ring fault, and cage fault, are decomposed by EEMD. Then correlation dimension of those signals phase is calculated, contrasted, and analyzed after space reconstruction. The experimental result shows that the correlation dimension, as nonlinear geometric invariants, can be used as the characteristic quantity of ECS turbine bearing on running state. Moreover, this method can accurately and effectively identify the running state of the bearing.
Highlights
In 1998, Huang [11] et al proposed a data-driven adaptive decomposition, empirical mode decomposition (EMD) method, which greatly improved the analysis effect of the nonlinear and nonstationary signal
In practical applications, there is a general problem of model mixing that is a single IMF containing the characteristic signal with maximum frequency difference or a signal with similar frequencies decomposed into different IMF in the method. e main reason for the model aliasing is that the abnormal events in the signal have an adverse effect on the selection of the extreme points which results in the uneven distribution of the extreme points to bring about the phenomenon of “overshoot” and “undershoot” in EMD process
Wu [12] et al put forward the method of ensemble empirical mode decomposition (EEMD). e method is to add the Gauss white noise of the finite amplitude to the decomposed signal and use the characteristic of Gauss white noise in the time frequency domain to smoothen the abnormal events, so as to reduce the adverse effects of abnormal events on the extreme point selection in the EMD process to achieve the purpose of the distribution of the extreme value point of the uniform signal
Summary
In 1998, Huang [11] et al proposed a data-driven adaptive decomposition, empirical mode decomposition (EMD) method, which greatly improved the analysis effect of the nonlinear and nonstationary signal. E EEMD decomposition steps are summarized as follows: (1) Gauss white noise ni(t) with a mean value of 0 and a constant amplitude standard deviation is added N times in the original signal x(t), respectively: xi(t) x(t) + ni(t), (1) En, the simulation signal was decomposed by EMD and EEMD, respectively, to get a series of IMF components as shown in Figures 2 and 3.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have