Ship Radiation Noise Analysis Based On Optimal Matching Wavelet

Ship radiated noise signal is one of the important indicators to ships or other underwater targets and due to long propagation distance, low attenuation and rich information of its line-spectrum,which has become the focus of the study on ship radiated noise. In fact, owing to the complex noise from ocean, ship radiated noise signal can be received weakly, so it is more and more difficult to detect and extract line-spectrum..Therefore line-spectrum recognition problem is worth deeply researching.In recent years, wavelet transform has found a wide range of applications due to its time-frequency localization and multiresolution, particularly for detecting singular points. According to features of ship radiated noise in the frequency domain, this paper is proceeded to select optimum wavelet base to further analyze power spectrum of the signal in low-SNR based on wavelet transform used in irregularity examination.Bior5.5wavelet is the optimal base in terms of the three wavelet base selection rules. By comparing with different wavelets which would be used for the detection of the power spectral singularity of ship radiated noise, it demonstrates the optimality of the bior5.5.Not only its approximate orthogonality guarantees the accuracy of the reconstructed signal, but also it balances the support between the compact and regularity.In order to accurately depict the true line spectrum in low-SNR signal, the ship radiated noise is simulated by three conditions of low frequency, mid-band and high-band for line-spectrum analysis by multiple perspectives under different noisy background.At low frequency, a signal matching method based on WT is proposed in this paper. By comparing the different wavelets processing experiments, only can the bior5.5wavelet accurately extract the characteristics of the ship radiated noise, which indicates the feasibility and effectiveness of the method. At mid-band, based on summaring the requirements of threshold function used in wavelet denoising, a new thresholding function is proposed and applied for the ship radiated noise analysis, to improve the signal matching analysis method based on wavelet transform, realized the line-spectrum extraction.At high-band, wavelet packet transform is presented in this analysis method.A large number of experiments show the effectiveness of the three methods, feasibility and their suitability.This article firstly analyse the characteristics of ship radiated noise signal based on wavelet from the frequency domain and get some important preliminary results. It proves that the bior5.5wavelet has the high resolution of the ship radiated noise signal.