中文摘要：

浅水水声信道受多径传播、时变和空变等特性的影响，由水听器直接观测到的信号大多是信噪比低、多种信号叠加而成的复杂信号，难以准确提取特征信息及信源分离。针对以上问题，提出基于空间滤波器的多源复杂水声信号的预处理方法。该方法通过Hilbert变换将直接观测信号构造成解析信号来抑制虚像，使用波束形成算法来实现对方向角的选择，在指定方位角下得到期望信号。仿真及试验结果表明：信噪比为0dB时，通过空间滤波器对多源复杂水声信号进行预处理，可以有效地抑制噪声，并有效区分处于同一时间段、同一频段的多个源信号，经处理后的信号与源信号的相似度可达0．9853，对实际信号处理也可达到较好的效果，解决时频域很难处理的问题，进而为后续的研究工作提高精度。

英文摘要：

Due to characteristics including multipath propagation, time-varying and space-varying features of shallow water underwater acoustic channel, signals directly detected by hydrophone are mostly complex signals superposed by various signals with low signal to noise ratio. Thus, the feature information and accurate source separation cannot be obtained. To solve these problems, a preprocessing method of multi-source complex underwater acoustic signal based on spatial filter is put forward. The method constructs the signals observed directly into analytic signals through Hilbert transforming to suppress the virtual image and uses beamforming algorithm to select the direction angle and obtain desired signals under the specified azimuth angle. Simulation and test results show that in a simulation test, if the multi-source complex underwater acoustic signals are processed based on spatial filter when the signal-to-noise ratio is 0 dB, the noises can be effectively suppressed and several source signals at a same time period and same frequency band can be effectively distinguished. After the processing, the similarity between the processed signal and source signal can reach 0.985 3. It can also reach good effects in actual signal processing, solve the problems hardly to be solved in time-frequency domain and ensure high precision in follow-up research.