A method for autonomous extraction of non-stationary low-frequency line spectrum of ship noise

Abstract

The invention discloses an autonomous extracting method for a non-stable low-frequency line spectrum of ship noise. The method comprises the following steps of (1), performing ship noise long-integration power spectrum line spectrum first detection, and obtaining a suspected line spectrum; (2), for aiming at the suspected line spectrum, setting the number of segments in segmented power spectrum analysis and the interval length of the integration domain; (3), for aiming at the suspected line spectrum, setting the integration domain starting point in analysis of each segmented power spectrum; (4), performing ship noise secondary power spectrum estimation according to the preset integration domain starting point and the interval length; and (5), performing continuous spectrum automatic offsetand line spectrum autonomous identification extraction. The method aims at a difficulty of autonomous extraction for the non-stable low-frequency line spectrum of the ship noise; through automatic analysis for a non-stable characteristic of the low-frequency line spectrum of the ship noise, adaptive adjustment is performed on the number of segments in segmented power spectrum analysis, the integration domain starting point and the interval length, thereby improving tolerance of the power spectrum to the line spectrum non-stable characteristic, improving signal-to-noise ratio of the line spectrum, and realizing autonomous extraction to the non-stable low-frequency line spectrum of the ship noise through automatic continuous spectrum offset and autonomous line spectrum identification.

Description

technical field [0001] The invention relates to a method for autonomously extracting non-stationary low-frequency line spectrum of integral domain self-adaptive ship noise, belonging to the technical field of sonar signal processing. Background technique [0002] In the complex underwater sound field environment, the automatic detection and feature extraction of ship noise is the basis for autonomous detection of unmanned platforms, which has important research value. Ship radiated noise generally consists of two components. One is broadband noise, which mainly comes from the cavitation process of seawater near the propeller, and the frequency extends continuously from near zero to tens of kilohertz. The other is a very narrow line spectrum, mainly concentrated in the low frequency band below 1 kilohertz. The line spectrum is mainly generated by the reciprocating motion of the ship's mechanical parts, the periodic water impact of the propeller blades and the resonance of t...

AI Technical Summary

Problems solved by technology

[0006] Purpose of the invention: Aiming at the problems existing in the prior art, the present invention provides a method for autonomously extracting the non-stationary low-frequency line spectrum of ship noise in the integral domain. Automatic analysis of the non-stationary characteristics of the noisy low-frequency line spectrum, adaptively adjust the number of segments, the starting point of the integral domain and the interval length of the segmented power spectrum analysis, improve the tolerance of the power spectrum to the non-stationary characteristics of the line spectrum, and improve the line spectrum signal. Through continuous spectrum automatic cancellation and line spectrum independent identification, the independent extraction of non-stationary low-frequency line spectrum of ship noise is realized

Images