Compressor surge poses significant risks to aerodynamic performance and structural safety, motivating the development of accurate early warning techniques based on signal analysis. This study identifies a previously under-recognized ultra-low-frequency surge precursor in a 1.5-stage axial compressor. Its under-recognition in prior studies is attributed to its close spectral proximity to the dominant high-amplitude surge signal, whose strong energy masks this weak low-frequency signature in conventional time–frequency representations. To resolve this faint precursor, a high-resolution signal-processing framework is established by integrating the fractional adaptive superlet transform—which has rarely been applied in compressor aerodynamic instability research—with complete ensemble empirical mode decomposition with adaptive noise. This combined approach provides the enhanced low-frequency resolution and robust mode separation required to isolate the precursor from the adjacent surge component. Using this framework, the precursor is extracted and characterized. It exhibits a characteristic frequency lower than the surge frequency, non-axisymmetric behavior, and a spiral upstream-propagating pattern, which cannot be identified using traditional Fourier- or wavelet-based techniques. Further analysis locates this low-frequency component to the vicinity of the outlet guide vane exit, which appears 20–100 rotor revolutions prior to surge onset and is closely associated with surge initiation. The results demonstrate that the proposed signal-processing techniques can reveal hidden instability signatures that are difficult to detect with conventional methods, and offer new possibilities for early surge detection, as demonstrated by the identification of a low-frequency surge precursor in the tested 1.5-stage axial compressor.
Liu et al. (Mon,) studied this question.