【作者】 李旭；

【导师】 韩宁； 吉志海；

【作者基本信息】 东南大学 ， 电子信息（专业学位）， 2023， 硕士

【摘要】 目标特征提取是水下目标分类识别的重要前提,而水声信道中的多径效应会导致回波信号的时频域混叠,从而严重影响目标特征提取的准确性。着眼于主动声呐系统,水下多径信号中未经多次反射的直达回波信号包含较为纯净的目标信息,如目标速度、尺度和材料等;若能够从多径信号中准确获得直达回波信号,将为目标特征提取提供更好的信号条件。本文以从多径信号中准确获得目标直达回波为研究目标,选取时间、频率分辨率较好的线性调频(Linear Frequency Modulation,LFM)信号作为发射信号,采用具有较好能量聚焦性能的分数阶傅里叶变换(Fractional Fourier Transform,FRFT)作为研究手段,通过分析FRFT中LFM信号的分辨性能,设计发射信号参数并提高多分量LFM信号的分辨率,以更好地辨别和分离水声信道中的多径信号,从而获得包含较纯净目标信息的直达回波。本文主要工作内容如下:(1)分析LFM信号参数和FRFT的量纲归一化因子对分数域(包含分数阶和分数域频率)中分辨多分量LFM信号的影响,给出主动声呐发射信号参数和量纲归一化因子的设计依据。分数域中LFM信号尖峰对应的分数域频率与分数阶是区分多分量LFM信号的主要依据,本文通过分析LFM信号在分数域频率中的主瓣和分数阶上的最佳旋转角,设计LFM信号参数和FRFT的量纲归一化因子,改善多分量LFM信号在分数域中的分辨性能以及FRFT的计算分辨率,并通过仿真验证了上述结论。以上研究有助于更好地区分多分量LFM信号。(2)依据水下声学多径(Underwater Acoustic Multipath,UWAM)信道模型和目标回波亮点模型,提出了基于FRFT的水下多径信号抑制和直达回波提取方法。在UWAM信道中,首先对单亮点目标模型的水下多径信号进行分析,在LFM信号参数和FRFT量纲归一化因子满足设计准则时,分析了多径信号在分数域的分布特征。然后在多亮点模型下,分析亮点之间的距离对多径信号在分数域中分布特征的影响,并给出了分数域中抑制多径信号和完整提取目标直达回波信号的理论依据。最后基于LFM信号在分数域中的主瓣宽度,使用不同窗长的窄带滤波器恢复原信号,并通过分析恢复信号与原信号的均方误差和相关系数,得出了窄带滤波器窗长设计原则。(3)在水池中进行实验,验证UWAM信道中多径信号在分数域中的分布特征并依据该特征提取目标直达回波信号。首先,搭建水池实验系统,选取合适的发射信号参数;然后,对接收信号进行滤波降噪,并利用匹配滤波处理在接收信号中对目标回波信号进行定位;最后,通过选择合适窗长的窄带滤波器,从分数域中提取直达回波,并依据均方误差和相关系数对直达回波信号的完整性进行分析,验证了本文研究内容的有效性。更多还原

【Abstract】 Target feature extraction is an important prerequisite for underwater target classification and identification,and the multipath effect in the underwater acoustic channel will lead to the time-frequency domain aliasing of the echo signals,which seriously affects the accuracy of target feature extraction.Focusing on the active sonar system,the direct arrival echo signal without multiple reflections in the underwater multipath signal contains relatively pure target information,such as target speed,scale and material.If the direct arrival echo signal can be accurately obtained from the multipath signals,it will provide better signal conditions for target feature extraction.In this thesis,the direct arrival echo of the target is accurately obtained from the multipath signals as the research goal,the Linear Frequency Modulation(LFM)signal with good time and frequency resolution is selected as the transmitted signal,and the Fractional Fourier Transform(FRFT)with good energy focusing performance is used as the research method.By analyzing the resolution performance of LFM signals in FRFT,designing the transmitted signal parameters and improving the resolution of multicomponent LFM signals to better identify and separate multipath signals in underwater acoustic channels,the direct arrival echo containing purer target information are obtained.The main contents of this thesis are as follows:(1)The influence of LFM signal parameters and FRFT dimensional normalization factors on resolved multi-component LFM signals in the fractional domain(including fractional order and fractional domain frequency)is analyzed,and the design basis of active sonar emission signal parameters and dimensional normalization factors is proposed.The fractional domain frequency and fractional order corresponding to the LFM signal speak in the fractional domain are the main basis for distinguishing multicomponent LFM signals.In this thesis,the optimal rotation angle of LFM signal in the main lobe and fractional order in the fractional domain frequency is analyzed,the LFM signal parameters and the dimensional normalization factor of FRFT are designed,the resolution performance of multi-component LFM signal in the fractional domain and the calculation resolution of FRFT are improved,and the above conclusions are verified by simulation.The above research is helpful for better identification of multi-component LFM signals.(2)According to the Underwater Acoustic Multipath(UWAM)channel model and target echo highlight model,an underwater multipath signal suppression and direct echo extraction method based on FRFT method is proposed.In the UWAM channel,the underwater multipath signals of the single-highlight target model is analyzed firstly,and the distribution characteristics of the multipath signals in the fractional domain are analyzed when the LFM signal parameters and FRFT dimensional normalization factor meet the design criteria.Then,under the multi-highlight model,the influence of the distance between the highlight on the distribution characteristics of the multipath signals in the fractional domain is analyzed,and the theoretical basis for suppressing the multipath signals and completely extracting the target direct arrival echo signal in the score domain is proposed.Finally,based on the main lobe width of the LFM signal in the fractional domain,the narrowband filter with different window lengths is used to restore the original LFM signal,and the window length design principle of narrowband filter is obtained by analyzing the mean squared error and correlation coefficient between the recovered signal and the original LFM signal.(3)Experiments are carried out in the pool to verify the distribution characteristics of the multipath signal in the fractional domain in the UWAM channel and extract the target direct arrival echo signal according to this feature.Firstly,the pool experiment system is built and the appropriate transmitted signal parameters are selected;Then,the received signals are filtered and noise reduction is carried out,and the matching filtering processing is used to locate the target echo signals in the received signals.Finally,by selecting a narrowband filter with appropriate window length,extracting direct arrival echo from the fractional domain,and analyzing the integrity of the direct arrival echo signal according to the mean squared error and correlation coefficients,the validity of the research content in this thesis is verified.更多还原