Underwater sound target identification method based on vessel radiated noise MFCC character

Abstract

The embodiment of the invention discloses an underwater sound target identification method based on a vessel radiated noise MFCC character. The method comprises the following steps: conducting angular domain framing operation on vessel target training radiated noise data; extracting the targeted MFCC character and differential MFCC character of vessel target radiated noise data for training and testing; obtaining an MFCC integrated character of the vessel target radiated noise data; training a multitask cutting prior hidden markov model on the MFCC integrated character of each frame of the vessel target training radiated noise data; calculating the conditional likelihood function value of the multitask cutting prior hidden markov model wherein the MFCC integrated character of the vessel target radiated noise data for testing corresponds to each frame of the vessel target radiated noise data for training; taking the categorical attribute of the frame of the vessel target training radiated noise data corresponding to the maximum conditional likelihood function value as the categorical attribute of vessel target testing radiated noise data. The method is used for improving the performance of underwater sound target identification.

Description

technical field [0001] The invention belongs to the technical field of underwater acoustics, in particular to an underwater acoustic target recognition method based on MFCC fusion features of ship radiation noise. Background technique [0002] Underwater acoustic automatic target recognition technology can provide information such as the attributes and categories of target ships. It is an important part of modern sonar systems and a key technology to realize the intelligence of other underwater equipment systems. [0003] Ship radiated noise refers to the noise generated by the mechanical operation and ship motion on the ship and radiated into the water. It is mainly composed of mechanical noise, hydrodynamic noise and propeller noise, and contains rich target physical information. Among them, the mechanical noise is the noise sound wave caused by the vibration of the ship's hull and radiated into the sea during the movement of the main engine, propulsion device and auxiliar...

AI Technical Summary

Problems solved by technology

[0007] (2) The difference in radiation noise data caused by the relative position and occlusion of the same target at different observation angles is not considered. Only one model is established for a ship target. The model is relatively rough and the recognition performance is not good.

[0008] (3) The recognition performance of the model is not considered when the number of samples available for training is small. When the number of samples used for training is relatively small, the recognition performance is greatly reduced

[0009] (4) There is a manual model selection problem, that is, the parameter set size of the model is determined manually, and the correctness of the given parameter set size determines the final recognition performance

[0010] Therefore, in view of the above-mentioned defects in the current existing technology, it is necessary to conduct research to provide a solution to solve the defects in the existing technology and avoid poor recognition performance of underwater acoustic targets. Identify issues with faster performance degradation

Images