LAUSR

Abstract This paper investigates the detection of the internal leakage in the hydraulic inlet guide vane (IGV) system at the gas turbine compressor. Based on methods of ensemble empirical mode decomposition (EEMD) and Hilbert transform (HT), we propose the internal leakage detection method. Firstly, EEMD is applied to decompose the pressure signal at chamber A of the cylinder into intrinsic mode functions (IMFs). Secondly, the instantaneous amplitude of the first IMF (IMF1) is obtained through HT. Finally, the mean of IMF1 instantaneous amplitude at a time period is adopted for the detection of the internal leakage fault and its levels. After setting the proper number of ensemble size and threshold value of the white noise amplitude to reduce the root mean-squared deviation for the multi-mode distribution of the IMFs, the developed method can identify the internal leakage in the IGV cylinder during experiments under the time-varying cylinder position and load at a built hydraulic IGV emulator. The empirical mode decomposition (EMD) and discrete wavelet transform (DWT) based methods are applied on the same pressure signals, respectively. The results show that the developed scheme owns the best performance comparing with the EMD and DWT based approaches in our case.