时变海洋声速剖面移动声源反演方法研究

发布时间：2018-06-28 02:41

本文选题：声速剖面 + 反演　；参考：《浙江大学》2015年硕士论文

【摘要】：海洋水体声速通常可由仪器直接测量得到,然而只能获得局部点及线的声速数据,无法获得局部海域的观测数据。利用声传播模型和观测数据相结合的声学反演技术,为高效、快速的获取感兴趣的区域海洋声速场数据提供了一种有效的方法。针对声速剖面所具有的空时演化特性,本文首先根据序贯滤波的思想展开了相关算法的研究。从改善集合卡尔曼滤波(Ensemble Kalman Filter, EnKF)集合多样性的角度,引入了马尔科夫链蒙特卡罗(Markov Chain Monte Carlo, MCMC)方法,提出了MCMC-EnKF方法,并将其应用于声速剖面反演中。在不同模型下的数值仿真和实验数据处理结果表明,在同等计算量的情况下,MCMC-EnKF算法可有效的提高EnKF算法的反演性能,且在时变性较强的声速剖面反演问题中优势更为明显。针对固定声源反演问题中观测范围与分辨率所存在的矛盾,研究了移动源声速剖面反演方法,通过移动声源的灵活性以增加局部声压的采样密度从而改善反演精度的目的。由于移动源的反演模型增加了声速剖面反演的维度,本文探索了高自由度的反演方法,将MCMC的改良算法差分演化自适应抽样(Differential Evolution Adaptive Metropolis, DREAM-zs)应用于声速剖面的反演。仿真分析验证了该方法在处理高自由度声速剖面反演问题中的可行性。利用部分海试数据进一步验证MCMC-EnKF和DREAM-zs算法的有效性,并将两者的实验结果与经典的EnKF进行比较。另外,利用湖试数据对移动源声速剖面反演的方法进行了初步验证。
[Abstract]:The velocity of sound in ocean water can be measured directly by instrument, but only the data of local points and lines can be obtained, and the observation data of local sea area can not be obtained. The acoustic inversion technique, which combines acoustic propagation model with observed data, provides an effective method for the efficient and rapid acquisition of regional ocean acoustic velocity field data of interest. In view of the space-time evolution characteristics of the acoustic profile, this paper firstly studies the related algorithms according to the idea of sequential filtering. In order to improve the diversity of Ensemble Kalman filter (EnKF) sets, the Markov chain Monte Carlo (MCMC) method is introduced. The MCMC-EnKF method is proposed and applied to acoustic profile inversion. The results of numerical simulation and experimental data processing in different models show that the MCMC-EnKF algorithm can effectively improve the inversion performance of the EnKF algorithm under the condition of the same computational complexity, and the superiority of the MCMC-EnKF algorithm is more obvious in the time-varying acoustic profile inversion problem. In view of the contradiction between the observation range and the resolution in the fixed sound source inversion problem, the method of moving source acoustic velocity profile inversion is studied. The purpose of improving the inversion accuracy is to increase the sampling density of local sound pressure by using the flexibility of moving sound source. Since the inversion model of moving source increases the dimension of acoustic profile inversion, a high degree of freedom inversion method is explored in this paper. The improved MCMC algorithm, differential Evolution Adaptive Metropolis (DREAM-zs), is applied to the inversion of sound velocity profile. The simulation results demonstrate the feasibility of this method in dealing with the problem of high degree of freedom acoustic velocity profile inversion. The validity of MCMC-EnKF and DREAM-zs algorithm is further verified by some sea trial data, and the experimental results are compared with the classical EnKF. In addition, the method of acoustic velocity inversion of moving source is preliminarily verified by lake test data.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P733.2

【相似文献】