水下目标主动声呐回波信号特征研究

发布时间：2019-03-04 19:55

【摘要】：目前在水下安静型小目标主动声纳回波信号特征提取中存在几个关键性问题亟待研究,包括了目标回波信号特征在复杂环境下的稳定性,目标几何声散射与各阶次弹性声散射时序结构的分析,目标固有物理属性与回波信号特征的关系,以及海底混响的抑制等。针对上述问题,本文研究了典型目标模型在混响背景下稳定的目标回波信号特征提取方法,独立的目标声散射成分信号提取方法,以及基于盲分离与时频分布图像处理的海底混响抑制方法,为实现水下安静型小目标主动声呐探测与识别奠定基础。首先,针对混响背景下目标回波信号特征稳定性问题,提出了目标回波信号的瞬时频率特征提取方法,并研究了目标回波与混响的在信号特征空间中的分布特性。将混响视为一类具有独立信号性质和特征的信号,推导了目标几何声散射成分与混响的时频分布特性,根据信号瞬时频率序列的随机性,提取了瞬时频率方差、瞬时频率熵、瞬时频率峰度与WVD-Radon变换半功率宽度四种特征。研究中采用特征选择方法建立了二维信号特征空间,对湖上沉底目标探测实验数据处理结果表明,目标回波与混响在该信号特征空间中具有稳定的分布特性。其次,针对目标回波中几何声散射与弹性声散射混叠的问题,提出了一种目标声散射成分信号分离方法。在声呐发射信号为线性调频脉冲的条件下,将目标声散射成分映射为单频信号成分,推导了目标声散射时序结构与映射结果之间的线性对应关系,从而可以通过窄带滤波分离出特定的单频信号成分。仿真及水池实验数据处理结果表明该方法对提取独立的目标弹性声散射特征的有效性。另外,提出了一种目标声散射成分时延估计处理流程,通过带阻滤波抑制强声散射成分,突出原目标回波中能量较弱的声散射成分,然后采用谱估计获得其频率。水池实验数据处理结果表明该方法可以有效提取出目标回波中较弱的棱角散射成分的时延信息。最后,从两个方面对混响的抑制问题进行了研究。一是针对目标回波与混响在时频分布平面上混叠的问题,研究采用形态滤波方法对目标回波时频分布图像进行处理的混响抑制方法,提出了一种与目标几何声散射时频分布形态特征相匹配的结构元设计方法;二是在基于盲分离的混响抑制中,针对缺乏目标回波与混响分离性能评价指标的问题,采用瞬时频率方差与瞬时频率熵作为指标来判断目标回波在分离结果中的顺序,并衡量不同盲分离算法对目标回波与混响的分离程度。湖上沉底目标探测实验与海上掩埋目标探测实验数据验证了这两点研究内容的有效性。
[Abstract]:At present, there are several key problems in the feature extraction of underwater small target active sonar echo signal, including the stability of target echo signal feature in complex environment. The analysis of geometric acoustic scattering and subelastic acoustic scattering sequence structure, the relationship between the intrinsic physical properties of the target and the echo signal characteristics, and the suppression of seabed reverberation, etc. In order to solve these problems, this paper studies the target echo signal feature extraction method which is stable under the reverberation background and the independent target acoustic scattering component signal extraction method. The method of underwater reverberation suppression based on blind separation and time-frequency distribution image processing lays a foundation for the detection and recognition of underwater small target active sonar. Firstly, aiming at the characteristic stability of target echo signal in reverberation background, a method of extracting instantaneous frequency feature of target echo signal is proposed, and the distribution characteristics of target echo and reverberation in signal characteristic space are studied. Reverberation is regarded as a kind of signal with independent signal properties and characteristics. The time-frequency distribution characteristics of target geometric acoustic scattering component and reverberation are deduced. According to the randomness of signal instantaneous frequency series, the instantaneous frequency variance and instantaneous frequency entropy are extracted. The instantaneous frequency kurtosis and the half-power width of WVD-Radon transform are four characteristics. The characteristic space of the two-dimensional signal is established by using the feature selection method. The experimental data processing results show that the target echo and reverberation have stable distribution characteristics in the characteristic space of the signal, and the experimental results show that the target echo and reverberation have stable distribution characteristics in the characteristic space of the signal. Secondly, in order to solve the problem of mixing geometric acoustic scattering and elastic acoustic scattering in target echo, a new method for signal separation of target acoustic scattering components is proposed. Under the condition that the sonar emission signal is a linear FM pulse, the target acoustic scattering component is mapped to a single frequency signal component, and the linear correspondence between the target acoustic scattering sequence structure and the mapping result is derived. Thus, specific single-frequency signal components can be separated by narrow-band filtering. The simulation and experimental data processing results show that the proposed method is effective for extracting the elastic acoustic scattering characteristics of independent targets. In addition, a processing flow for time delay estimation of target acoustic scattering components is proposed. The strong acoustic scattering components are suppressed by band-stop filtering, and the weak energy acoustic scattering components in the original target echo are highlighted. Then the frequency of the target is obtained by spectral estimation. The experimental data processing results show that the proposed method can effectively extract the time delay information of the weak angular scattering components in the target echo. Finally, the suppression of reverberation is studied from two aspects. Firstly, aiming at the aliasing problem of target echo and reverberation in time-frequency distribution plane, the reverberation suppression method using morphological filtering method to process the time-frequency distribution image of target echo is studied. A structural element design method which matches the morphological characteristics of time-frequency distribution of target geometric acoustic scattering is proposed in this paper. Second, in the reverberation suppression based on blind separation, the instantaneous frequency variance and instantaneous frequency entropy are used as indexes to judge the order of target echo in the separation result, aiming at the lack of target echo and reverberation separation performance evaluation index. The degree of separation of target echo and reverberation by different blind separation algorithms is also measured. The experimental data of underwater target detection and buried target detection on the lake verify the effectiveness of these two research contents.
【学位授予单位】：哈尔滨工程大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TB56

【相似文献】