The Study Of Target Echo Extraction In Active Sonar Signal Processing Based On Blind Source Separation And Array Extension

Underwater target echo extraction is an important issue in active sonar signal processing. Due to the strong interference of Seabed reverberation and the limits to physical array aperture, the underwater target detection will be a difficult task. In this paper, the feasibility of blind source separation algorithm for underwater target echo extraction in active sonar data to eliminate the reverberation was analyzed, and the second-order statistics-based blind separation algorithm -- SOBI algorithm was introducted.In this thesis, the simulation of the SOBI algorithm and experimental data processing were made. It is shown that using only receive datas of three neighboring array elements for target echo blind source separation, the extracted signal had a better matched filter results and a clearer time-frequency characteristics than beamforming did. And SRR was increased by the algorithm.For the limit to array aperture, the four-order cumulant array aperture extension was descripted in this article. For the fourth-order cumulant array extension did not give a time-domain waveform of the shortcomings, a time-domain signal recovery method for virtual array was put forward in this paper, which was deduced through formula. Simulation results show that: with time-domain signal recovered, the main beam is more narrower and the sidelobes was lower than the original array. And the array's spatial resolution was improved. And a virtual array of time-domain waveforms was given, which provides the basis for the following signal processing. The validity of this method was also verified by Lake test data processing.