Research On Background Equalization And Target Detection Technology For Active Sonar

For active sonar,reverberation is the main interference in addition to the marine environmental noise.Unlike marine environmental noise,reverberation is caused by the transmission of signals through interfacial reflection and the scattering of tiny particles in water.It is similar to the emission signal in the time domain,coherent with the emission signal in the frequency domain,and often stronger than the echo signal.Therefore,it has been a long standing problem for active sonar.Suppression of reverberation interference has always been a difficult problem in underwater acoustic signal processing.The active sonar generally assumes the Gaussian white noise hypothesis.Actually,the reverberation exhibits non-uniformity,non-Gaussian,non-white and non-stationary.Limited by these,there is no optimal detector in the reverberation.This paper aims to build an asymptotically optimized detector in the context of reverberation,and mainly carries out the following work:First,aiming at the various heterogeneity of reverberation,the mixed Gaussian autoregressive model(GMAR)is used to model it,and the expectation maximization algorithm(EM)and the weighted least squares algorithm(WLSE)are used to estimate the model parameters,and then the Gaussian filter and AR pre-whitening filter are built based on the model parameters to eliminate the heterogeneity of the reverberation background to be incorporated into the traditional detection framework.Then,in order to break through the threshold of the traditional active sonar detection technology and achieve the detection of echo signals at lower SNR,the paper adopts a set of combined filter methods,mainly including conventional matched filter,t0 filter,pre-filter and adaptive line spectrum enhancement.The self-adaptive matched filter algorithm can separate echo signals from reverberation,random fluctuation and other interferences,making the detection performance of active sonar improve by about 6-8dB compared to the traditional optimal filter-matching filter.Then,the paper proposes a reverberation suppression and target detection technique based on target echoes.The difference between the target echo and the reverberation is that the former has a certain relative stability on the time axis and the frequency axis.This article uses this feature to strengthen the target echo,and at the same time weaken the random interference to achieve the purification of the sonar image.The technique has a strong inhibitory effect on reverberation and significantly improves the ability to detect weak targets,especially weak targets in a strong short-range reverberation background.Finally,the above methods are used to process the lake test data in order to detect the weak echo target in the strong reverberation background.The reverberation is well suppressed in three dimensions: time,space and frequency,and the background is well purified.The technique achieves the effect of only retaining the short-range weak target in the sonar image,and significantly improves the detection ability of the weak target in the reverberation background.The detection performance is improved at least 10 dB compared with the existing detection method.