Improved Hilbert–Huang transform based weak signal detection methodology and its application on incipient fault diagnosis and ECG signal analysis

• Developed a novel weak signal detecting methodology based on the improved HHT method.
• Restrained end effects of EMD by embedding wavelet analysis in the sifting process.
• Achieving the best results among the improved EMD methods and stopping criterions.
• Verified the validity by processing weak mechanical signal and biomedical signal.
• High computational efficiency whilst achieving robustness against low SNR.

In the present study, a weak signal detection methodology based on the improved Hilbert–Huang transform (HHT) was proposed. Aiming to restrain the end effects of empirical mode decomposition (EMD), wavelet analysis was embedded in iteration procedures of HHT to remove iterative errors as well as noise signal in the sifting process. Meanwhile, a new stopping criterion based on correlation analysis was proposed to remove undesirable intrinsic mode functions (IMFs). Results of analyzing synthetic signal, incipient rotor imbalance fault of Bently test-rig and weak electrocardiogram (ECG) signal show that the improved HHT combined with wavelet analysis have excellent weak signal detecting performance whilst achieving robustness against low signal-to-noise ratio (SNR). Furthermore, comparative studies of the proposed method, the classical EMD method, and other four generally acknowledged improved EMD methods, as well as a widely used stopping criterion demonstrate that the proposed method significantly reduces end effects and removes undesirable IMFs.