Research On Modulation Feature Extraction Methods Of Passive Targets Divided Frequency Segments

The passive sonar identifies the targets depends on the noises radiated from the targets, using line-spectrum radiated from the targets to classify and identify the targets is an important research in the underwater acoustic signal processing field. Feature extraction is the key link of the targets classification, on account of the sea environment complexity and the underwater acoustic channel particularity, it's a puzzle to extract a kind of target essence feature and underwater distant detection effective feature.Modulation feature can reflect the ship's physical parameters, for example, the number of the screw propeller's vane and rotate speed of the principal axis and so on, is the fixed feature of the ship radiated noises. Using DEMON identifies the underwater acoustic get more recognition, but it still exists many shortages in application at present, for example, demodulation spectrum quality is not good and feature extract automated is difficulty, influences the application of DEMON.This paper is to extract feature to measure the frequency of the line-spectrum and shaft-frequency based on wavelet transform and high distinguishable ZFFT. For the sake of improving the demodulation quality, presents a kind of improved DEMON method: Firstly, divides the data frequency segments in wavelet packet transform, amalgamates the DEMON spectrum that demodulated from the different frequency segments, and gets distinct line-spectrum. Secondly, purifies the line-spectrum makes the DEMON line-spectrum clearly. Lastly, extracts the modulation feature of the targets to provide important message for identifying the targets.