Noise-Robust DEMON Spectrum Extraction for Ship-Radiated Noise via Adaptive Decomposition and Sparse Reconstruction

To improve the quality of detection of envelope modulation on noise (DEMON) spectra extracted from ship-radiated noise under noisy conditions, a noise-robust DEMON spectrum extraction method incorporating adaptive decomposition and sparse reconstruction is proposed. Ensemble empirical mode decomposition (EEMD) is employed to adaptively select the effective frequency band, while the diagonal slice of the third-order cumulant is utilized to suppress Gaussian noise and enhance modulation characteristics. Subsequently, sparse Bayesian learning (SBL) is introduced to reconstruct the DEMON spectrum and improve the spectral resolution. The effectiveness of the proposed method is validated through numerical simulations and lake-trial experiments. The results show that the proposed method produces cleaner spectral backgrounds and improves the clarity of the shaft frequency and its harmonics compared with the conventional EMD-based method, especially under low modulation-depth conditions. These results demonstrate that the proposed method provides an effective solution for robust DEMON spectrum extraction from ship-radiated noise in complex underwater acoustic environments.