Research On Ship Target Recognition Technology

Ship target recognition technology is an effective means of the high technologyconditions to defeat the enemy under the water also one of the key technology to urgentachieve today. The technology has civilian and military values, and has received extensiveattention by scholars around the world. This paper studies around the feature extraction andrecognition technology of the ship target. The feasibility and effectiveness of the method havebeen verified through simulations and the real ship target data processing.The main content of this paper include the following aspects:1、The paper studied the generation mechanism of the ship-radiated noise, analyzed thepurpose and steps of the DEMON spectrum analysis technique, and then described theabsolute value low-pass demodulation and square low-pass demodulation methods. Throughthe simulations we can get the conclusion that the square low-pass demodulation is beneficialto spectrum detection in DEMON spectrum line and is able to obtain a more significant linespectrum characteristic. Also we joined the adaptive line spectrum enhancement technologyin DEMON spectrum analysis to improve the line spectrum of ship target detectioncapabilities, as well as the possibility of target recognition. Finally, we explored the shaftfrequency of the ship-radiated noise. By large number of experimental data improve that thealgorithm can extract the ship shaft frequency accurately.2、The paper analyzed the pitch characteristics of the ship-radiated noise, obtain the pitchvalues by singular feature analytical methods of wavelet analysis. It has been confirmed thatthe method can be used as a basis for classification of ship-radiated noise by simulations.3、The higher-order statistics is an effective tool for the study of nonlinear andnon-Gaussian signal. The paper studied the nature and concept of higher-order statistics andderived conversion the relationship of moments and cumulants.4、The paper discussed the basic principles of the Hilbert-Huang transform, and gave thespecific steps and flow chart of the empirical mode decomposition algorithms. Intrinsic ModeFunctions of the signal have been separated through the simulations. Finally, the paperdiscussed a new algorithm, through the simulations the mode mixing have been weakenedobviously.5、The paper introduced the basic concepts of the artificial neural network, and gave thedefinition of the BP neural network classifier. Firstly get the Intrinsic Mode Functionscomponent of the ship-radiated noise through Hilbert-Huang transform, and then conduct feature extraction to instantaneous frequency and higher-order statistics characteristics,finally execute classification and identification through BP neural network. Satisfactoryresults were achieved by simulations. The correct recognition rate was about80%. The paperLaid a good foundation for the follow-up work.