Research And Application Of Empirical Mode Decomposition And The Instantaneous Frequency Filtering Algorithm

Traditional Fourier analysis has worked very well in stationary and linear signal processing, but it can not do the same thing in nonstationary and nonlinear signal processing. Based on the knowledge of stochastic process, some kinds of signal analysis methods have developed in recently. The Empirical Mode Decomposition and the Hilbert-Huang Transform analysis was the hotspot in the last ten years.Firstly, this paper introduces the Hilbert-Huang Transform analysis, list out the basic steps of Empirical Mode Decomposition(EMD), discuss the essence idea in this analysis method.Secondly, studied the problem of EMD time-frequency analytic method, including modes mixing, end effect, curve spline, sifting stop condition, Intrinsic Mode Functions(IMF) orthogonalization and the affection of sampling rate to the result of EMD, while introduced a new formula to describe the effect of sampling rate to the first IMF; combined some new optimization algorithms, produced the improved real time EMD time-frequency analytic method; studied some kind of filtering method making use of the EMD and IMF, and based on the physical essence of instantaneous frequency, proposed a new filtering method that can filter any band of the instantaneous frequency from the Hilbert-spectrum directly at any time. Simulation results show the validity and superiority.Thirdly, summarized the fault diagnosis system of Air Cycle Machine in Boeing 747. Introduced the design of software mainly, such as demand analysis, flowing graph and coding. At the same time illuminated the process of data acquisition and the way to get the data to the software.Finally, the filtering method based on the instantaneous frequency is used to filtering color noise in the ACM vibration signal and a simulative signal which is very like the real one. So demonstrate the efficiency and superiority of this new filtering method.