Waveform Fusion And Background Normalization For Low Frequency Active Sonar

Due to the heavy reverberation in shallow water, it is much harder to detect the target accurately with the low frequency active sonar. Most of the components of the reverberation are seabed back scatters. This thesis is targeted to the waveform selection and fusion, background normalization and the separation between reverberation and the target echo, by which the detection performance of the active sonar will be apparently increased.From perspective of waveform selection, the range resolution, the velocity resolution and the performance in suppressing reverberation are studied with transmitted CW, LFM and PTFM respectively. The resolution has close relationship with the ambiguity function. Through deep analysis on different ambiguity functions, conclusions are as following: CW is of better velocity resolution; LFM has better range resolution with the identical duration of CW; PTFM covers both advantages. In the heavy reverberation environment, it is an effective way to improve the signal to reverberation ratio by the waveform selection. Through the analysis on the Doppler distribution model of the reverberation, the reverberation can be suppressed by CW in theory when the target moves with a high speed; the broadband signal can reduce the spectrum power of the reverberation, therefore, the detection performance is definitely improved.Different signals have different advantages and disadvantages. However, the environment is previously unknown in, usual, the exact motion of the target is often hard to be accurately obtained. Therefore, an appropriate transmitted waveform selection is pretty difficult. Provided that different waveforms are fused and the waveform fusion detection method is applied, the advantages of different transmitted waveforms may be adopted. By this mean, the detection performance will be upgraded. Both the theory and the experiment validate that the fused. scheme can suppress the frequency selective fading, CW and LFM can complement with each other in detection by which the performance naturally is increased.The normalization is a post process in the active sonar detection. It is oriented to constant false alarm rate detection. Based on the analysis of the basic theory, the instantaneous characteristic and the stationary state of the normalization, it is obtained that the reasonable number of samples to estimate the background may decrease the detection loss. Two dimensions-time and beam-normalization approach is adopted. The means, mid-values, and weighted value algorithms and performance parameters are compared. By simulation and sea experiment, it is proved that fluctuation of data can be effectively reduced by normalization. The display plane background is more smoothing, the target echo is much clearer, the detection performance is much better, and the false alarm probability is much lower.Along the line of the normalization, a new method is explored to separate the target echo from background. Although the transmitted waveforms are different, the reverberation, background noise and the target echo still retain their corresponding characteristics in the outputs of match filter. In the display plane, the reverberation is still heavy; the background noise is slight, but covers most areas; the target echo is also slight and only small spots may contain the target echo. Based on those characteristics, two rectangle blocks with the same center and various sizes are explored. The smaller block corresponds to the target echo, and the size is identical to the target echo in the display plane; the other is related to the reverberation. Two mean values are computed. The first is the mean value of the smaller block, and the second is the mean value of the area in the other block without the small block. In the display plane, different ratios of two mean values correspond to different colors. By this way, the display is much more effective than before in sea experiments.This thesis is targeted to the waveform selection and fusion, background normalization, separation between the reverberation and the target echo, by which the detection performance of the active sonar will be improved and the false alarm probability is lowered.