Feature Extraction Method of Underwater Acoustic Signals Using VMD and Reverse Weighted Permutation Entropy

The underwater acoustic signals are rife with various non-linear and non-stationary elements. Denoising is vital in processing these signals. In this paper, we propose a denoising method for underwater acoustic signals based on an enhanced variational mode decomposition (VMD) and reverse weighted permutation entropy (RWPE). A significant challenge in the VMD algorithm is the incorrect selection of the mode component K and the penalty factor alpha. This can lead to partial decomposition and mode interference. To tackle this issue, we employed the whale optimization algorithm (WOA). This algorithm refines the penalty factor and adjusts the mode component count in the VMD. Consequently, the optimal parameter combination for K and alpha is identified. Using this optimal set, the VMD processes the acoustic signal to yield several intrinsic mode functions (IMFs). The RWPE then identifies the IMFs embodying significant signal features. These selected IMFs are utilized for reconstruction, culminating in the denoising of the underwater acoustic signal. Comparative analyses indicate that our refined VMD method outperforms other denoising techniques.