Research on Ship-Radiated Noise Denoising Using Secondary Variational Mode Decomposition and Correlation Coefficient

Area of Science:

• Oceanography
• Signal Processing
• Marine Engineering

Background:

• Ship-radiated noise (SN) contains vital ship characteristics but is often corrupted by noise.
• Effective denoising algorithms are crucial for accurate SN signal analysis and ship identification.
• Existing methods like Empirical Mode Decomposition (EMD) and standard Variational Mode Decomposition (VMD) have limitations in complex noise environments.

Purpose of the Study:

• To propose a hybrid secondary denoising algorithm combining Variational Mode Decomposition (VMD) with a correlation coefficient (CC) for ship-radiated noise (SN).
• To evaluate the denoising performance of the proposed algorithm against various existing methods.
• To demonstrate the algorithm's effectiveness in improving ship feature extraction and classification accuracy.

Main Methods:

• A secondary VMD (2VMD) approach is employed to decompose signals into intrinsic mode functions (IMFs).
• Correlation coefficients (CC) are calculated between IMFs and the original signal to identify and remove noise-corrupted IMFs.
• A secondary denoising process is performed by repeating decomposition, screening, and reconstruction based on CC thresholds.

Main Results:

• The proposed 2VMD combined with CC algorithm demonstrated superior denoising performance compared to EMD, EEMD, VMD, and 3VMD methods.
• The algorithm's effectiveness was validated across different signal-to-noise ratios and VMD decomposition times.
• Application to SN signals significantly improved feature extraction and ship recognition rates.

Conclusions:

• The hybrid secondary VMD and CC algorithm offers a robust and effective solution for denoising ship-radiated noise (SN).
• This advanced denoising technique facilitates more accurate feature extraction, leading to improved ship classification.
• The study highlights the potential of VMD-based hybrid methods in marine acoustics and surveillance.