As critical air-dropped acoustic sensors for underwater target detection, sonobuoys are frequently compromised by severe hydrodynamic self-noise induced by sea-surface wave excitation, which masks target signals and degrades detection performance. While structural optimizations have traditionally been employed, effective signal-processing-based noise suppression remains challenging because the noise is non-stationary and physically coupled with buoy motion. To address the limited physical interpretability of conventional decomposition methods, this study proposes a physically guided self-noise suppression framework: VMD Constrained by DCCA Correlation (VMD-DCCA). The main contribution is the incorporation of the Detrended Cross-Correlation Analysis (DCCA) coefficient between the sonobuoy’s vertical velocity and the acoustic data as a correlation-dependent constraint within the Variational Mode Decomposition (VMD) optimization process. This motion prior allows more targeted isolation of motion-induced components than standard data-driven decomposition. Simulation and controlled water-tank results show that VMD-DCCA outperforms EEMD and standard VMD, achieving an SNR improvement of approximately 15 dB at an input SNR of -9 dB. The reconstructed signal also preserves visible narrowband spectral lines in the time-frequency representation. These results demonstrate the potential of the proposed method for controlled or post-processing sonobuoy self-noise reduction, while validation under irregular open-ocean conditions remains necessary.
Loading....