水下目标矢量声场建模与基于EMD的信号检测

发布时间：2018-08-24 19:53

【摘要】：随着人类对海洋资源的深入开发以及对探测方面日益增长的民用和军用需求,传统的信号检测方法已经不能满足这些要求。为了提高信号处理的性能,本文采用一种基于经验模式分解的改进方法-总体经验模式分解方法来进行信号处理,仿真研究了该算法的性能,并使用基于总体经验模式分解方法和基于小波包分解方法对信号进行检测,对比性能,最后使用这种方法进行实际数据的处理。论文的主要内容如下:首先,分别对点源和偶极子源在自由场和波导中传播进行建模仿真后,得到标、矢量信息,可以得出声波在波导中传播时,在波导汇聚区传播的声波比其在自由场中传播的衰减更小。其次,对小波包分解算法的理论基础进行了研究,推导了小波包的频率顺序,使用两种小波包阈值对加噪信号进行去噪处理,仿真后比较了两种阈值的均方根误差(RMSE),选取自适应阈值的均方根误差比固定阈值的均方根误差更小,具有更好的去噪效果。将熵理论与小波包理论相结合,应用到信号检测中,得到基于小波包能量熵算法的检测性能,提供了性能对比。然后,研究了 EMD方法的理论基础,对EMD方法中存在的模态混叠问题进行了分析。为了抑制分解间断信号时EMD算法的模态混叠问题,本论文采用总体经验模式分解(EEMD)方法来进行解决。由于EEMD可以通过添加高斯白噪声并利用EMD的二元滤波器组的特性,通过仿真对EEMD算法中的相关参数进行了分析、选定,可以很好的克服间断信号EMD过程中的模态混叠问题,然后将这种方法应用到信号检测中,与信号的EEMD能量熵相结合,并使用浮动门限对信号进行检测。与基于小波包能量熵方法进行比较后发现基于EEMD能量熵的检测方法比基于小波包的方法检测性能更好。最后将这种方法对试验数据进行检测,采用合适的门限后,基于此种方法的漏检情况比基于小波包方法要少很多,且基于EEMD能量熵方法的检测性能比基于小波包能量熵的检测性能要更好,此种方法对试验数据处理是有效的,可以为水下信号的检测提供技术支持。
[Abstract]:With the further development of marine resources and the increasing demand for civil and military exploration, the traditional signal detection methods can not meet these requirements. In order to improve the performance of signal processing, an improved method based on empirical mode decomposition, the global empirical mode decomposition, is adopted to process the signal. The performance of the algorithm is simulated. The method based on total empirical mode decomposition and wavelet packet decomposition is used to detect and compare the performance of the signal. Finally, the method is used to process the actual data. The main contents of the thesis are as follows: firstly, after modeling and simulating the propagation of point source and dipole source in the free field and waveguide, the scalar and vector information can be obtained, and the sound wave propagating in the waveguide can be obtained. The attenuation of sound waves propagating in the convergence region of the waveguide is less than that in the free field. Secondly, the theoretical basis of wavelet packet decomposition algorithm is studied, the frequency sequence of wavelet packet is deduced, and two kinds of wavelet packet threshold are used to Denoise the noised signal. Simulation results show that the root-mean-square error (RMS) of the two thresholds selected by (RMSE), is smaller than that of the fixed threshold, and the root-mean-square error of the adaptive threshold (RMSE), is better than that of the fixed threshold. The entropy theory is combined with the wavelet packet theory and applied to signal detection. The detection performance based on the wavelet packet energy entropy algorithm is obtained and the performance comparison is provided. Then, the theoretical basis of EMD method is studied, and the modal aliasing problem in EMD method is analyzed. In order to suppress the modal aliasing problem of EMD algorithm when discontinuous signal is decomposed, this paper uses the (EEMD) method of total empirical mode decomposition to solve the problem. Because EEMD can add Gao Si white noise and make use of the characteristic of binary filter bank of EMD, the related parameters in EEMD algorithm are analyzed and selected by simulation, which can overcome the problem of modal aliasing in the process of discontinuous signal EMD. Then this method is applied to signal detection, combined with the EEMD energy entropy of the signal, and the floating threshold is used to detect the signal. Compared with the energy entropy method based on wavelet packet, it is found that the detection method based on EEMD energy entropy is better than that based on wavelet packet. Finally, this method is used to detect the experimental data. After adopting the appropriate threshold, the leakage rate based on this method is much less than that based on wavelet packet method. The detection performance based on EEMD energy entropy method is better than that based on wavelet packet energy entropy. This method is effective for experimental data processing and can provide technical support for underwater signal detection.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P714;TB566

【参考文献】