Analysis of Chaotic Characteristics of Ship Radiated Noise Signals with Different Data Lengths

Abstract

Passive identification of maritime targets is an important research direction of sonar signal processing. The problem of passive sonar detection of maritime targets is very complex, and factors such as the yearly decrease of the radiation noise level of ships and the difficulty of sample acquisition have restricted the development of passive sonar target recognition. How to extract the effective feature parameters of maritime target signals and realize the target type quickly, accurately and robustly is the new demand of passive sonar target recognition in the background of new technology. The variability of the marine environment and the uniqueness of the hydroacoustic channel make the identification and feature extraction of naval radiation noise a major problem in hydroacoustic signal processing. Traditional signal analysis and processing is based on stochastic modeling of statistical signal processing. In recent years, the rapid development of nonlinear dynamics has shown that many apparently complex irregular and disordered motions can be modeled as deterministic nonlinear dynamical systems. For the exploitation of the ocean, hydroacoustic signal processing is the basis for underwater target identification, and the extraction of the eigenvalues of ship radiation noise is an important part of hydroacoustic signal processing. It has been shown by previous studies that ship radiated noise is a nonlinear signal, and the propagation behavior becomes complex and disordered with time extension.This paper analyzes the chaotic characteristics of ship radiated noise on the basis of time series chaotic dynamics analysis, and then experimentally studies the results under different data parameters. The following is the work done:(1)The modeling about chaos is performed in phase space, so the phase space reconstruction of time series is the basis for studying nonlinear dynamical systems. Phase space reconstruction is the mapping of our initially obtained one-dimensional time series to a higher dimensional space. In this paper, based on Taken's theorem, the phase space reconstruction is performed on the ship's radiation noise. The important parameters time delay and embedding dimension that can perform phase space reconstruction are calculated.(2)The actual ship radiation noise signal is collected as the research object, and two chaotic features that can be effectively used for target identification are calculated, which are the correlation dimension and the largest Lyapunov exponent.(3)The data lengths of 5,000, 10,000 and 15,000 points were analyzed at sampling frequencies of 1kHz, 2kHz, 5kHz and 10kHz, respectively.(4)The distribution characteristics of these two chaotic features are analyzed under different parameter conditions. The experiments on the extraction of chaotic features of the actual ship noise signals show that different data sample parameters have different effects on different chaotic features.The effect of data sample parameters on the correlation dimension and the largest Lyapunov exponent in this paper is different. For the correlation dimension, the data sample points have a greater impact than the sampling frequency, while for the largest Lyapunov exponent, the sampling frequency has a greater impact than the data sample points. The significance of this paper is that it is more meaningful for us to focus on the study of stable chaotic feature quantities. For the influence of different data parameters on different features, we need to select the data sample parameters corresponding to the stable chaotic feature quantities to estimate the feature values, so that the estimated feature values are more informative. Feature extraction of underwater target signals has very important theoretical significance in modern hydroacoustic signal processing, and has important application value for detection, tracking and identification of hydroacoustic signals.