基于单矢量传感器的飞行器姿态角研究

针对惯导部件使用不便的缺点,提出基于极化电磁波进行飞行器姿态感知的方法.飞行器通过单电磁矢量传感器接收来自基站的极化电磁波信号,通过基于四阶累积量的ESPRIT算法进行信号参数估计,最终估计飞行器姿态角(俯仰角、航向角、横滚角).     

基于测控电磁波极化特征信息的飞行器姿态估计

该文应用正交解耦的姿态参数信号模型描述极化电磁波的3 维波结构,将最大似然估计和MUSIC 算法推广应用于部分极化波参数估计,利用极化电磁波传递信号基站标定的姿态基准信息。飞行器通过单电磁矢量传感器接收并处理信号,确定电磁波结构向量,计算得到地理坐标系下的飞行器姿态。此方法有别于需多点测量的三角计算法,只需一个地面基站信号和运动平台上单一接收点,就可实现姿态感知,可替代姿态导航仪和航向导航仪。该文介绍了运动平台姿态估计算法,为工程应用提供技术支持。      

基于单电磁矢量传感器的MUSIC联合参数估计方法

在有电磁波信息链的条件下,飞行器通过单电磁矢量传感器接收来自基站的完全极化电磁波信号,并对该信号进行处理获得电磁波的波达角和极化角.本文提出了一种新的基于电磁矢量传感器的MUSIC极化-空城联合谱估计方法,此方法将极化空域导向矢量分解成矩阵旋转秉积的形式,并将其作进一步的演化成矩阵生成元的形式.最后用仿真验证了算法的有效性.     

从三正交偶极子天线接收信号中估计飞行器姿态参数

来自基站的极化电磁波包含的波达方向和极化状态信息,能建立飞行器姿态与大地坐标系的联系,因此利用机载三正交偶极子天线接收信号可以估计飞行器的姿态。该文研究了在只有一个基站信号和飞行器上单一接收点 的条件下,如何实现飞行器的姿态感知,并推导了飞行器姿态估计算法。该文首先研究了估计波达角和极化椭圆倾斜角的算法,并分析了参数估计的性能,然后利用这些参数推算出飞行器的俯仰角、航向和横滚角,从而获得对飞 行器进行控制的必要姿态信息。最后通过仿真实验验证了该方法的有效性。      

空间相机像移补偿计算中飞行器大姿态角使用方法

提出了在大姿态角情况下飞行器姿态参数的使用方法以实现像移速度的精确计算。根据飞行器轨道坐标系转换到飞行器坐标系的转换矩阵,推导了在大姿态角存在时轨道坐标系下的姿态角速度与飞行器坐标系下的姿态角和姿态角速度的关系式。通过计算得出,在侧摆角为30时对飞行器坐标系绕飞行器轨道坐标系的转动角速度S1、S2和S3的影响最大误差分别为1.175%、50%和13.223%;在前后摆角为30时对S1、S2和S3的影响最大误差分别为63.397%、0.1745%和63.397%。根据空间相机像移速度计算精度要求比较高,确定了在一定的姿态角情况下飞行器姿态参数使用方法。提出的方法简单,易于实现,适用于空间相机像移补偿的研究。     

基于信号极化三维平稳性的飞行器姿态估计

以电磁波为参照的飞行器姿态测量研究,可弥补空间参照物缺乏,丰富姿态测量手段。利用3个正交电(磁)场传感器接收电磁波三维电(磁)信号,测量运动平台姿态。该方法根据姿态参数估计的CRB(克拉美—罗界)是否为有限值,判断该环境电磁波信号能否用于运动平台姿态测量;根据基准姿态与实时姿态互相关矩阵变化,估计得到实时天线坐标系到基准天线坐标系的转换矩阵,据此将实时姿态坐标系下的运动平台姿态,转换到基准姿态坐标系下,再转换到地理坐标系下,完成运动平台姿态感知。电磁波信号可以为独立或相干信号,波达方向和极化状态任意,可不受多径效应困扰。仿真试验表明该方法稳定有效。     

矢量传感器阵列MIMO雷达高精度二维DOA与极化联合估计

该文采用矢量传感器配置下的十字型阵列MIMO雷达系统,提出一种新的2维高精度DOA与极化参数联合估计算法。首先根据MIMO雷达虚拟阵列导向矢量的特点,通过降维矩阵的设计及回波数据的降维变换,将高维回波数据转换至低维信号空间;然后基于传播算子获得对应信号子空间的估计,利用收、发阵列阵元间长基线对应的旋转不变性和极化矢量中电场矢量和磁场矢量的叉积进行2维高精度DOA估计和解模糊处理,同时利用与阵列结构无关的极化域旋转不变性进行极化辅角和极化相位差的联合估计。该矢量传感器MIMO雷达阵列可同时获取MIMO雷达的波形分集和矢量传感器的极化分集,无需额外增加阵元和硬件开销,能够有效扩展阵列孔径,提高参数估计性能;同时通过降维变换及传播算子,在获取信噪比增益的同时,能够实现2维高精度DOA和2维极化矢量的联合估计及参数的自动配对,有效降低数据处理维数和参数估计的运算复杂度;最后,仿真结果验证了理论分析的正确性和算法的有效性。     

电磁矢量传感器原位误差校正方法

 现有基于电磁矢量传感器阵列的信号DOA和极化参数联合估计算法,大都假设电磁矢量传感器的三个电偶极子和三个磁偶极子严格指向参考坐标系的三个坐标轴,即不存在原位误差.然而在实际应用场合,电磁矢量传感器是存在原位误差的,因此其实际极化-角度域导向矢量与理想情况下的极化-角度域导向矢量有一定的偏差,导致现有方法的估计性能显著下降,因此必须对原位误差进行校正.通过对存在偏差的极化-角度域导向矢量进行一阶Taylor近似展开,并利用一个辅助校正源,提出了电磁矢量传感器原位误差校正方法,给出了原位误差估计的CRB界.仿真结果验证了该方法的有效性.     

基于三维波结构向量的飞行器姿态快速估计

飞行器姿态测量的主要参照为重力和地磁场,其在高加速度环境中的缺点难以克服.将电磁波三维结构向量作为飞行器姿态测量的参照,根据机载电磁矢量接收阵列的姿态与接收电磁信号的变化关系建立导向矢量.通过MUSIC算法三维搜索飞行器姿态参数使得信号空间谱最大化,实现飞行器平台姿态估计.针对MUSIC算法计算量过大导致姿态估计较慢的...     

基于单电磁矢量传感器的多参数联合估计

利用电磁矢量传感器来实现信号多参数的联合估计是近年来信号处理领域的一种新的方法和途径。基于单个电磁矢量传感器通过构造两组具有固定时延的接收数据,利用TLSESPRIT算法和矢量叉积相结合实现了多参数的联合估计,参数自动完成配对,可以最多同时估计5个不相关的信号,且空间到达角、极化参数的估计和信号频率的估计是两个相互独立的过程互不影响,与利用极化敏感阵列进行参数估计相比计算量小,并通过计算机仿真验证了此方法的有效性。     

基于多特征融合与支持向量机的手势识别

针对手势识别中人的手部特征描述易受到环境因素影响,手势识别率低等问题,并考虑到单个特征的局限性,提出了一种基于Hu矩和HOG特征融合的支持向量机手势识别新方法。该方法首先对处理后的手势图像提取局部的HOG特征,然后针对手势的轮廓提取全局Hu矩特征,再将两种特征融合成混合特征,并通过主成分分析法对混合特征进行降维形成最终分类特征,并将新特征输入到支持向量机中进行识别。实验表明,该方法具有较好的鲁棒性和较高的识别率。     

Self-initiating MUSIC-based direction finding and polarization estimation in spatio-polarizational beamspace

A novel self-initiating multiple signal classification (MUSIC)-based direction-finding (DF) and polarization-estimation algorithm in spatio-polarizational beamspace is presented for an arbitrarily spaced array of identically oriented electromagnetic vector sensors. An electromagnetic vector sensor, already commercially available, is composed of six colocated, but diversely polarized, antennas distinctly measuring all six electromagnetic-field components of a multisource incident wave field. This proposed algorithm: (1) exploits the incident sources' polarization diversity; (2) decouples the estimation of the sources' arrival angles from the estimation of the sources' polarization parameters; (3) uses ESPRIT on pairs of vector sensors to self-generate coarse estimates of the arrival angles to start off its MUSIC-based iterative search without any a priori information on the incident sources' parameters; (4) estimate the sources' polarization states; and (5) automatically pairs the x-axis direction-cosine estimates with the y-axis direction-cosine estimates and with the polarization estimates. Monte Carlo simulation results verify the efficacy of the proposed method.     

Joint DOD–DOA Estimation Based on Signal Subspace Approach in MIMO Radars

This paper deals with joint direction of departure (DOD) and direction of arrival (DOA) estimation using a subspace-based approach in bistatic multiple-input multiple-output radars. First, a conventional rotational invariance technique is used to estimate DOD. Then the steering vector of DOA corresponding to the estimated DOD is obtained using the proposed signal subspace estimation method. The steering vector is then converted to the angle of arrival. This way, the DOD and DOA pairing are automatically obtained. Numerical simulations show that the estimated angles using the proposed approach are more accurate than those using the conventional methods.     

2D DOA and Frequency Estimation Method with One Vector-Sensor and Two Pressure Sensors Based on ESPRIT and Signal Power

We use one vector and two pressure sensors to form a sparse large aperture L-shape array for high performance two-dimensional (2D) direction of arrival (DOA) and frequency estimation. Because the number of sensors is small and there is only one vector sensor in the presented array, thus, the installation of sensors in the array is simpler and installation error is smaller, than the conventional array. Meanwhile, a high performance 2D DOA and frequency estimation method is presented. Firstly, utilizing single vector sensor and based on the ESPRIT, a group coarse 2D DOA and frequency parameters are obtained. Secondly, to restrain space noise or interference, a matrix filter is utilized to process the covariance matrix which comes from sensor array, so as to form a new covariance matrix which possesses high signal to noise ratio. Thirdly, utilizing the new covariance matrix and based on the ESPRIT again, accurate but ambiguity angles estimates are obtained. Fourthly, one signal power estimator and one optimization method are presented to solve the angle ambiguity and frequency ambiguity problems, respectively. The proposed method gains a high performance 2D DOA and frequency estimation results. Numerical simulations are performed to verify the feasibility of the proposed method.

     

Khatri-Rao积变换下的离格信号DOA估计

当存在离格信号时,基于稀疏表示的波达角(DOA)估计算法性能损失严重.为解决这个问题,在对接收数据协方差矩阵进行Khatri-Rao积变换的基础上,推导了离格信号网格偏离量与紧邻信号原子系数之间的关系,提出了一种单一离格信号DOA估计方法.为提高对邻近离格信号DOA的估计性能,利用矩阵的广义逆性质提出了基于多原子系数的联合估计方法.仿真实验表明,单一离格信号DOA估计方法在低信噪比下有较好的性能,联合估计方法在高信噪比条件下对邻近离格信号DOA有较高的估计精度,同时所提算法估计性能几乎不受网格划分间距的影响,可以通过增大网格间距降低算法运算量.相关研究对阵列天线DOA估计具有一定的参考价值.     

基于波束域的子空间正交性测试宽带DOA估计方法研究

投影子空间正交性测试(Test of Orthogonality of Projected Subspace:TOPS)算法通过测试宽带信号各频率点上噪声子空间和信号子空间之间的正交性对目标方位进行到达角估计(DOA:direction-of-arrival)。此算法对参考频点上的信号子空间的估计依赖性较大,因此存在较多伪峰,低信噪比条件下性能差等缺点。针对该问题,提出一种基于波束域的宽带DOA估计方法。该方法通过将阵列接收信号转换到波束域,在波束域中利用信号带宽内各频率分量的波束域方向向量与噪声子空间之间的正交关系构造判决向量,根据判决向量搜索空间谱的极大值对应的角度进行DOA估计。该方法不需要进行角度预估,避免了TOPS算法中常出现的伪峰,降低了信噪比分辨门限,减少了计算量,具有较好的估计效果。将该方法分别运用到均匀圆阵和线阵上,通过仿真对比和海试实验数据的处理,证明了本文所提方法的有效性。      

A vector space approach to direction finding in a coherent multipath environment

The superresolution methods of spatial spectrum estimation heretofore reported are known to break down when arriving signals from different directions are coherent. It is shown here that this deficiency can be overcome by replacing the direction-of-arrival (DOA) search vector ordinarily used with a linear combination of such vectors. Particular attention is here given to extending the multiple signal classification (MUSIC) method to cover such cases. A procedure for determining the proper linear combination is presented so that determination of the separate coherent signal arrival angles involves a multidimensional search in angle only. Such a search is demonstrated for coherent two-ray and three-ray multipath environments.     

RAM-based angle estimation with linear spatially separated polarisation sensitive array

Two-dimensional direction of arrival (2D-DOA) and polarisation estimation of multiple incident signals for linear spatially separated polarisation sensitive array (SS-PSA) is investigated with reweighted atomic norm minimisation (RAM) algorithm. Single vector sensor in this paper is composed of spatially separated three dipoles and three loops. Firstly, the received data of the proposed array with single snapshot is converted into the new received data with six virtual snapshots. Each virtual snapshot data is the data sensing by dipoles or loops located in the same direction. Secondly, based on each virtual snapshot data, the complex amplitudes and frequencies of the virtual signals are restored using RAM algorithm. Lastly, 2D-DOA and polarisation angles are derived according to the restored complex amplitudes and frequencies. Simulation results demonstrate that the RAM algorithm used in 2D-DOA and polarisation estimation with SS-PSA exhibits superior performance, which is better compared with polarisation multiple signal classi?cation (PMUSIC) algorithm and rotational invariance techniques (ESPRIT) algorithm.     

利用DBSCAN和概率密度估计的欠定盲源分离混合矩阵估计

针对欠定盲源分离中混合矩阵估计精度不佳的问题,本文提出了一种结合带噪声的基于密度的空间聚类（combining density-based spatial clustering of application with noise, DBSCAN）和概率密度估计的混合矩阵估计算法。首先,通过向量转换方式获得单声源时频点检测准则,并基于此准则从混合信号中检测出单声源点。其次,利用基于密度的空间聚类算法对单声源点进行聚类,由此估计出声源个数以及各类别所属的单声源点。再次,利用概率密度估计获得各类别的聚类中心,并构成混合矩阵。所提混合矩阵估计方法不需要提前设定声源个数,并且避免了由于数据分布不均所造成的聚类效果差的问题。最后,采用压缩感知技术实现源信号恢复,从而从混合信号中分离出各个声源信号。实验结果表明,本文所提的混合矩阵估计方法在声源个数未知的情况下,能够准确估计出混合矩阵;并且分离出的信号具有较高的质量。