Study On Fault Feature Extraction Method Of Aeroengine Early Early Rub Impact Fault

As the heart of an aircraft, the health of aircraft engine will directly affect the performance, reliability and economy of aircraft. As the aircraft engine structure is becoming more and more compact, such fault as rubbing becomes a common fault of aircraft engine, and its influence is very serious. If the fault sign could be found at the early stage of rubbing fault, and the diagnosis given at time, it is of great significance for the safety operation of the aircraft engine.This paper is to aim at early rubbing fault of the aircraft engine, and the simulation experiment of the typical rubbing fault is carried out via the rotating machinery laboratory bench, in addition, a software package called Beijing Oriental DASP6.18 system is applied for data acquisition and signal processing. This paper firstly gives a brief analysis of failure mechanism, early rubbing fault characteristics as well as several typical early rubbing faults;some methods such as wavelet transform, wavelet packet decomposition, the EMD decomposition and harmonic wavelet packet method are applied for characteristics extraction of some kinds of typical early rubbing faults; Based on a comparison between the results of four characteristic extraction methods by MATLAB simulation, the wavelet method can decompose all early rubbing fault signals, its effect is not obvious for the high frequency parts,but significant for low frequency parts decomposition, as a result the wavelet method can't extract the characteristics of early rubbing faults completely; The method of wavelet packet is no sense for the low frequency parts, however can extract the fault features of high frequency parts effectively. On the other hand, fault characteristics can also be extracted via spectrum analysis method from the data obtained from the wavelet packet decomposition; The EMD method can decompose data into different IMF envelope groups, and obtain the data information of different groups of IMF; Spectrum analysis result from Harmonic wavelet packet algorithm for dealing with state data is similar with the one from wavelet packet decomposition, but effect of extraction for fault characteristics of high frequency part based on a harmonic wavelet packet method is less than the one from wavelet packet method.This paper carries out a comparative study on the four different characteristics extraction methods, and summarizes the advantages and disadvantages of them. As a result, it is concluded that wavelet packet method is more suitable for characteristic extraction of early rubbing fault.