集合卡尔曼滤波在时变声速剖面追踪中的性能分析

对集合卡尔曼滤波在时变海洋环境下的声速剖面追踪性能进行了分析。将南海实验背景下普林斯顿海洋模型预报的声速剖面正交分解为3阶系数组成的状态-空间形式,其状态转移方程建模为3阶自回归过程;基于卡尔曼反馈理论,利用适合于水平非均匀模型RAM仿真的观测声压场对系统状态进行校正,实现声速剖面的动态追踪。在水平均匀、水平非均匀和海底参数失配环境下的仿真结果均能较好地实现对声速剖面的追踪,验证了算法的可行性。同时对不同信噪比、粒子数、阵元数和海底参数失配等情况下的分析表明,观测信息量的增加可以有效抑制观测误差和模型误差的影响,相关结论得到了实验数据的验证。     

浅海声速剖面与移动声源的跟踪定位

在水平非均匀分布的浅海环境中,针对移动声源跟踪时,声速剖面的变化会对声场产生影响,提出了一种利用集合卡尔曼滤波算法的声速剖面跟踪反演和移动声源跟踪定位的方法。首先,将声速剖面进行距离和深度的参数化表示,从而将对声速剖面的跟踪转化为对声速剖面前3阶经验正交函数系数的跟踪;其次,通过将声源状态信息和声速剖面信息表示为状态变量,而将垂直线列阵接收到的声场信息作为测量值建立状态-测量模型,然后利用集合卡尔曼滤波方法对模型状态变量进行跟踪。仿真结果得出:声速剖面跟踪反演的均方根误差和移动声源跟踪定位的绝对误差都非常小,对声源的跟踪定位精度很高。并且通过增加集合样本数、增加接收信号信噪比以及增加接收阵元数目都可以提高跟踪定位结果精度。最后,利用东海实验数据对本方法进行了验证。     

浅海复杂环境下等效声速剖面的构建方法*

针对浅海复杂环境声速剖面水平变化情况下的声传播损失预报及目标定位问题,提出了一种基于遗传算法的等效声速剖面重构算法。首先,将声速剖面进行时间和空间上的分解,从而将声速剖面抽象为对声速剖面前三阶正交函数系数的反演;其次,利用遗传算法,以先验声速剖面集为基础,进行参数反演。仿真结果表明,在浅海复杂条件下,传播损失预报受声速剖面及海洋参数的影响,不能直接运用接收或发射位置处的声速剖面进行传播损失预报,否则会对预报结果造成误差。通过构建具有声传播累积效应的等效声速剖面可以提高匹配场定位精度,完成目标定位,且在构建等效声速剖面时,接收位置处即本地声速剖面所占权重较大。     

浅海声速剖面的匹配波束反演方法

为了快速反演获得浅海声速剖面,利用匹配波束构造了反演代价函数,在此基础上研究了浅海声速剖面的匹配波束反演方法(MBI)。并运用东中国海的实验数据进行实际反演,验证了MBI的可行性和鲁棒性。利用垂直线列阵接收的29个不同位置的爆炸声信号进行了声速剖面反演,MBI反演的结果与实验中直接测量的声速剖面一致,且反演声速剖面的均方根误差小于2 m/s。研究结果表明,MBI比常规匹配场反演(MFI)获得浅海声速剖面更快速、更准确,并且对底质参数的失配具有较高的鲁棒性。     

基于多域卷积神经网络与自回归模型的空中小目标自适应跟踪方法

针对星空背景下卫星跟踪中运动小目标与伪目标交会造成的跟踪漂移问题,提出一种基于多域卷积神经网络(MDNet)与自回归(AR)模型的空中小目标自适应跟踪方法。对用MDNet采集到的图像序列第1帧的正样本进行bounding-box回归模型训练;再训练用最小信息准则和最小二乘法确定阶数和参数的AR模型,估计目标运动轨迹并预测目标位置;最后,将该目标位置作为MDNet的采样中心,约束采样候选区域,用bounding-box回归模型调整目标位置。实验用8种跟踪方法测试了8组场景复杂的视频序列,结果表明,本文方法的成功率及平均覆盖率均显著高于其他7种典型算法,具有较高的精确性和稳健性。     

非等声速信道下的循环神经网络机动目标跟踪模型

针对水下非等声速信道声线弯曲导致传统滤波跟踪轨迹偏差的问题,提出一种循环神经网络的目标跟踪模型。该模型在缺乏声速剖面信息的情况下,通过数据驱动迭代训练,学习输入观测值与输出状态值之间的映射关系,实现目标位置和瞬态特征变化的精确获知。蒙特卡洛仿真实验结果表明,本文模型在非等声速信道下复杂机动场景中相较传统单模型滤波算法以及交互式多模型算法,水平距离跟踪精度分别提升4.06%,1.57%,深度估计精度分别提升0.87%,0.85%。本文模型相较于传统滤波方法具有更高的跟踪精度,并且能够在失配声速分布信道下进行迁移学习,提升模型在失配声场环境下的泛化性。     

一种声速剖面展开的正交基函数获取方法

为了评估经验正交函数(Empirical Orthogonal Functions)在声速剖面的声学反演应用中的可靠性和适用性,降低声速剖面基函数的获取对样本数据的依赖性,提出了一种声速剖面展开的正交基函数的获取方法.该方法通过分析引起声速剖面变化的海洋动力学过程,利用动力学方程求解得到的水动力简正波(Hydrodynamic Normal Modes)来给出声速剖面展开的基函数.HNMs方法获得的声速剖面正交基函数与EOFs方法的结果非常吻合,且对样本数据的依赖性小。HNMs方法物理意义清晰,能够给出决定声速剖面正交基函数的物理参数,使评估经验正交函数的准确性和适用性成为可能。     

考虑内阻时变特性的电池状态估计方法

针对锂离子电池欧姆内阻随温度变化情况,提出了一种考虑内阻时变特性的两步无迹H∞滤波锂电池状态估计方法。首先,对锂电池和内阻抗进行分开建模,在电池Thevenin模型的基础上构建内阻抗预测模型,实时修正模型参量;接着,将无迹变换嵌入到扩展H∞滤波中,降低测量噪声对估计精度的扰动,从而提高电池荷电状态的估计精度。最后,在实验室环境下对电池进行充放电实验,分别针对降温和升温情况下的内阻值及电池端电压的估计进行了详细的实验分析,同传统方法相比,本文方法具有较高的估计精度。     

浅海时变声速环境下的自适应匹配场定位算法实现*

针对浅海环境下声速剖面失配引起的匹配场处理器失配问题,提出了一种自适应匹配场定位算法在声速剖面时变环境下的实现方式。将先验声速剖面集简化为经验正交函数表示,结合蒙特卡洛方法与环境扰动约束算法对当下时刻的目标声源进行匹配场定位。本文以某次试验获取的连续20小时的声速剖面数据为研究对象,通过仿真试验对该算法进行验证,结果表明：在先验声速剖面集的半小时之后,利用自适应算法的距离和深度定位成功率较常规匹配场算法有较大提升,其中,深度正确定位概率相对较低。     

激光跟踪测量系统自校正方法研究

建立了测量系统运动学模型。针对以往在球坐标激光跟踪系统中参考点只能借助于外部更高精度的仪器来校正的缺点,提出了平面约束自校正方法,并对系统进行了自校正研究。仿真结果表明了该方法的有效性。     

Tracking of time-evolving sound speed profiles with an auto-regressive state-space model

An approach for time-evolving sound speed profiles tracking in shallow water is discussed.The inversion of time-evolving sound speed profiles is modeled as a state-space estimation problem,which includes a state equation for predicting the time-evolving sound speed profile and a measurement equation for incorporating local acoustic measurements.In the paper,auto-regression(AR) method is introduced to obtain a high-order AR evolution model of the sound speed field time variations,and the ensemble Kalman filter is utilized to track the sound speed field.To validate the approach,the accuracy in sound speed estimation is analyzed via a numerical implementation using the ASIAEX experimental environment and the sound velocity measurement data.Compared with traditional approaches based on the state evolution represented as a random walk,simulation results show the proposed AR method can effectively reduce the tracking errors of sound speed,and still keep good tracking performance at low signal-to-noise ratios.     

强非线性时间演化声速剖面的序贯反演

受海面波浪起伏、降雨和内波等海洋动力学过程的影响,浅水声速剖面的时间演化具有高度非线性,针对该问题提出使用改进的粒子滤波方法进行声速剖面序贯反演.该方法通过建立声速剖面的经验正交模型(EOF)以及描述声速剖面时间演化特征的状态空间模型,将声速剖面反演问题建模为状态跟踪问题,利用不敏粒子滤波(UPF:Uncented Particle Filter)算法进行声速剖面序贯反演。仿真试验通过实测声速剖面数据和先验地声参数信息产生接收声场数据,再利用模拟声场数据估计声速剖面的时间变化.结果表明,相比于集合卡尔曼滤波(EnKF:Ensemble Kalman Filter),在计算效率等同的情形下,该方法可以在状态参数的时间跳变点保持良好的跟踪性能,一定程度上克服了现有反演算法在跳变点发散的问题,可以有效提高声速剖面反演精度,尤其在声速剖面时变性较强时具有显著优势.     

Determination of equivalent sound speed profiles for ray tracing in near-ground sound propagation

The determination of appropriate sound speed profiles in the modeling of near-ground propagation using a ray tracing method is investigated using a ray tracing model which is capable of performing axisymmetric calculations of the sound field around an isolated source. Eigenrays are traced using an iterative procedure which integrates the trajectory equations for each ray launched from the source at a specific direction. The calculation of sound energy losses is made by introducing appropriate coefficients to the equations representing the effect of ground and atmospheric absorption and the interaction with the atmospheric turbulence. The model is validated against analytical and numerical predictions of other methodologies for simple cases, as well as against measurements for nonrefractive atmospheric environments. A systematic investigation for near-ground propagation in downward and upward refractive atmosphere is made using experimental data. Guidelines for the suitable simulation of the wind velocity profile are derived by correlating predictions with measurements.     

A model for the vertical sound speed and absorption profiles in Titan's atmosphere based on Cassini-Huygens data

Measurements of thermodynamic quantities in Titan's atmosphere during the descent of Huygens in 2005 are used to predict the vertical profiles for the speed and intrinsic attenuation (or absorption) of sound. The calculations are done using one author's previous model modified to accommodate non-ideal equations of state. The vertical temperature profile places the tropopause about 40 km above the surface. In the model, a binary nitrogen-methane composition is assumed for Titan's atmosphere, quantified by the methane fraction measured by the gas chromatograph/mass spectrometer (GCMS) onboard Huygens. To more accurately constrain the acoustic wave number, the variation of thermophysical properties (specific heats, viscosity, and thermal conductivity) with altitude is included via data extracted from the NIST Chemistry WebBook [URL webbook.nist.gov, National Institute of Standards and Technology Chemistry WebBook (Last accessed 10/20/2011)]. The predicted speed of sound profile fits well inside the spread of the data recorded by Huygens' active acoustic sensor. In the N(2)-dominated atmosphere, the sound waves have negligible relaxational dispersion and mostly classical (thermo-viscous) absorption. The cold and dense environment of Titan can sustain acoustic waves over large distances with relatively small transmission losses, as evidenced by the small absorption. A ray-tracing program is used to assess the bounds imposed by the zonal wind-measured by the Doppler Wind Experiment on Huygens-on long-range propagation.     

Tracking-positioning of sound speed profiles and moving acoustic source in shallow water

An ensemble Kalman filter(EnKF) approach is proposed to perform sequential tracking of water column sound speed profile(SSP) using a moving acoustic source. First,the SSPs are discretized in depth and range, and are expressed by the empirical orthogonal functions(EOFs). Second, the acoustic source state information and the first three orders of EOF coefficients are expressed as the state variable, and the acoustic field information received by the vertical line array are the measured values. Successively, the state variables and measured values are used to establish the state-measure model. Last, the EnKF is utilized to track the state variables. The simulation results show that the root mean square error of SSP and the absolute error of source are all small, and thus the acoustic source tracking-positioning has high accuracy. Moreover, increasing the number of sample collection, the signal-to-noise ratio and the number of receiving elements can improve the tracking-positioning results. The method is verified using the experimental data of the East China Sea.     

Causality and the speed of sound

A usual causal requirement on a viable theory of matter is that the speed of sound be at most the speed of light. In view of various recent papers querying this limit, the question is revisited here. We point to various issues confronting theories that violate the usual constraint.