Study On The Method Of Target Echo Detection Under Reverberation Background

For active sonar system, the reverberation is the dominating interference in shallow water. The reverberation is dependent on the transmitted signal greatly and it is non-stationary. In this paper, study and analyze the method of detection in reverberation, emphasize to study high-order statistic analysis to solution detection in reverberation. First, divide up the reverberation data to reasonable partition then we think the neighboring two sections of reverberation data may be partial steady, calculates the AR model's the parameter using high-order statistic, and design filter to pre-whited reverberation, it was proved that it's more effective than the pre-whited algorithms based of correlation or non-whitening algorithms in non-Gauss reverberation environment by analysis of the ROC. Second, analysis the difference between reverberation and echo, extracts the bispectrum and higher-order statistics feature, then BP neural network is used for target detection, and test it by lake trial data, find it's effective than matching filter.