A Leakage Signal Identification Method Based on EEMD and Frequency Band Energy Ratios

Infrasound-based pipeline leak detection represents a highly precise and versatile method. However, practical applications in long-distance gas pipelines face challenges due to the presence of operational noise, which often masks the faint infrasound signals generated by leaks. To tackle this issue, this study adopts Empirical Mode Decomposition (EEMD) to process the non-stationary infrasound signals associated with leaks. Subsequently, leveraging the distinct autocorrelation characteristics between leak and noise signals, the signal is reconstructed to determine the leak presence based on energy distribution within the reference frequency band. Analysis of field data reveals significant disparities in frequency band energy distribution between leak and noise signals in the infrasound range, thereby enhancing the accuracy of leak detection. This approach provides pivotal insights for advancing infrasound-based leak detection systems.