Research On Feature Extraction And Instability Monitoring Of Fluctuation Pressure In Axial Compressor

Stability analysis of axial compressor is a hotspot but also a difficulty in aviation sector in borneand abroad. There are two kinds of instability in compressor, one is rotating stall and the other is surge.The study on the instability characteristics of compressor helps to realize the onset and process ofrotating stall and surge, and promote mechanical investigation. Stall precursor identification ofcompressor instability has been essential since the proposal of active control. It could prevent thecompressor from getting into unstable state when identify the stall inception as early as possible, andit plays a very important role in raising working efficiency and stability of compressor. The details ofthis paper are listed as follows:1. Signal features in the different states of stationary, transitional and instability are analyzed bymeans of wavelet transform。We find a phenomenon that some Modal disturbances become strongerwhen the compressor approaches to instability in low speed and medium speed compressors. And thefrequency of the disturbance is in different range to stall and surge. Then several Instabilitymonitoring methods are introduced, and there merits and demerits、feasibilities are discussed.2. Based on the Gabor wavelet transform, A new characteristic parameter is presented which candescribe the transient energy of the disturbances over a frequency range. This parameter shows amarked increase in the transitional state when it is applied to calculate the testing data. This method issimple and the best situation1.5seconds earlier can detect the inception. Besides, this parameter canindentify the type of instability when choose different analysis range.3. Given the data of compressor aerodynamic stability experiment, A software about featureextraction and time frequency analysis of fluctuation pressure is developed based on the MATLABGUI. This software has lots of practical functions like time-domain feature analysis、featureextraction、 time frequency analysis、 Form Parameter calculation、 signal reconstruction、cross-correlation analysis、stall cell calculation and the display and output of analysis results and so on.These functions can meet presupposed requirements. The interface of the software is simple butgenerous which is convenient for operation. and it raises analyzing efficiency extremity.4. Applied to computer simulation and testing data, the results show that the method and theprogram are valid and effective. Besides, some problems which have to be considered when usingcross-correlation, cross-power spectrum, stall cell calculation are discussed in this paper.