一种面向弹道再入目标跟踪的HPD-SRCQSPF算法

针对弹道再入目标轨迹跟踪问题，提出基于混合建议分布的平方根容积求积采样粒子滤波（HPD-SRCQSPF）算法，该算法以混合建议分布为框架，由两个基本建议分布组成。其中一个基本建议分布为先验分布，另一个基本建议分布为平方根容积求积卡尔曼滤波估计后的值。该混合建议分布与真实的后验分布很接近，因此有着高效性、高精度等特点。仿真结果表明，对于弹道再入目标轨迹跟踪模型，相比于标准粒子滤波（SPF）算法和平方根容积求积粒子滤波（SRCQPF）算法，HPD-SRCQSPF算法可以在较低运算负载的情况下获得更好的跟踪性能。特别是在弹道目标变轨机动的情况时，所提出算法的性能增益更为显著。     

粒子滤波评述

以最优Bayesian滤波的求解为起点,综述了粒子滤波的发展历程、基本思想、算法的各个基本环节、基本的滤波算法及其收敛性以及算法的多种重要衍变形式,包括辅助变量粒子滤波、自适应粒子滤波、实时粒子滤波、分布式粒子滤波、Rao\|Blackwellised粒子滤波、免重采样粒子滤波和裂变自举粒子滤波,并通过一个复杂的递推非线性滤波估计例子,用Monte Carlo仿真实验的方法对几种典型的粒子滤波算法进行了比较研究,最后总结了粒子滤波的应用并展望了进一步研究的方向。     

复杂电磁环境下组网雷达对隐身目标的跟踪研究

在复杂电磁环境背景下,研究了组网雷达基于粒子滤波算法对隐身目标的跟踪滤波问题,建立了雷达观测方程和隐身目标的多基地雷达截面积模型,分析了基于U PF的组网雷达目标跟踪滤波算法。利用计算机仿真验证了复杂电磁环境下组网雷达利用粒子滤波技术跟踪隐身目标的有效性,并对几种跟踪算法的效率进行了比较。     

基于方位信息的组网雷达跟踪滤波研究

干扰对组网雷达目标跟踪系统模型的影响,决定了系统需要用非线性滤波技术进行纯方位跟踪.粒子滤波避免了传统非线性滤波方法的缺陷,但是存在粒子退化,于是用EKF和UKF在每一时刻更新粒子,用更新的粒子及其协方差构造重要性函数,然后重采样.仿真实验表明了这两种改进粒子滤波方法的有效性.     

改进高斯混合粒子滤波的纯方位目标跟踪算法

针对纯方位跟踪系统非线性较强、传统跟踪滤波方法收敛速度慢且容易发散的问题,提出了一种基于改进高斯混合粒子滤波的纯方位跟踪算法。该算法基于Sigma点卡尔曼滤波(SPKF)和粒子滤波的特点,用有限的高斯混合模型来近似后验状态密度、系统噪声和观测噪声的分布。利用贪心EM算法实现模型的降阶,一定程度上克服了EM算法假定混合成分数为已知、迭代的结果需要依赖初始值、可能收敛到局部最大点和可能收敛到参数空间的边界的缺点,从而改善粒子枯竭的问题。仿真实验结果表明在纯方位跟踪领域,与传统粒子滤波(PF)和基于EM的高斯混合粒子滤波相比,该算法在保持高精度估计能力的同时,具有较强的鲁棒性,是解决非线性系统状态估计问题的一种有效方法。     

矢量观测确定卫星姿态的预测粒子滤波算法

在矢量观测的基础上,针对单独的星敏感器定姿,提出了一种将粒子滤波(PF)和预测滤波相结合的姿态确定算法,通过设计粒子初始化,结合重要性采样、重采样和规则化等手段,成功地将姿态四元数作为状态粒子进行更新和传递,避免了状态方程的线性化和协方差矩阵的计算;利用预测滤波算法估计模型误差和姿态角速度,在保证滤波精度的同时,有效降低了粒子滤波器的维数.实验在某对地观测通用小卫星平台上进行,选取卫星自由飞行状态和飞轮控制对地稳定模式,分别对滤波器进行了仿真,实验结果验证了该算法对本质非线性、非高斯的卫星姿态估计问题具有快速的收敛性能和良好的稳定精度.该方法还为粒子滤波器的设计和无角速度敏感器测量的飞行器姿态确定提供了借鉴.     

基于粒子滤波的被动多基站跟踪算法

针对被动多基站跟踪的实际应用，提出了基于粒子滤波的被动多基站跟踪方法。该方法首先利用两个基站的观测数据融合目标位置，并用第三个基站的观测数据对目标位置进行检验，消除了错误融合位置；然后，推导了被动多基站粒子滤波模型，并根据被动多基站跟踪的特点，提出了顺序重抽样方法，有效地解决了被动多基站跟踪中的高度非线性、非高斯的影响；最后，给出了算法的性能仿真比较。仿真结果表明提出的方法性能明显优于其他跟踪方法（如PMHT、IMM-PDA）。     

空间非合作机动目标跟踪相对导航方法研究

就基于雷达测量的空间非合作机动目标跟踪相对导航问题进行了研究,提出了一套 既能够处理测量粗差,又能使滤波器在存在未知输入情况下快速收敛的相对导航算法。本文 将强跟踪滤波器的思想引入到二阶分离插值滤波算法（DDF2）,设计DDF2\|STF滤波器;另 外,本文将DDF2与Huber滤波算法相结合,设计DDF2\|Huber滤波器;最后,本文设计了相应 的滤波器切换准则,根据不同情况选用DDF2\|Huber和DDF2\|STF滤波器,从而得到一套 完整的相对导航算法。数学仿真证实了所提出相对导航算法的有效性。本文的研究成果为工 程设计提供了理论参考。
     

基于测量数据积累的弱目标检测前跟踪算法

检测前跟踪(TBD)算法是一种有效的弱目标联合检测与跟踪方法。当目标强度参数未知时,基于点目标假设的主流TBD算法通常将其建模为增广目标状态变量,在贝叶斯框架下联合估计目标运动状态与强度信息。本文研究发现强度参数估计精度受限于点扩散函数(PSF)的设置,提出了一种基于测量数据积累的弱目标TBD算法。该方法利用目标的航迹,积累多帧间目标占据分辨单元的测量数据,在实现弱目标的联合检测与跟踪的同时,对目标强度参数完成准确估计,为后续目标分类或识别提供有用的目标强度信息。仿真结果表明:在低信噪比条件下,所提算法的强度参数估计精度和稳健性优于主流TBD算法,并最终收敛于参数完全已知时的最优联合检测与跟踪结果。     

基于粒子滤波的红外弱小多运动目标检测

针对低信噪比条件下红外图像序列中的多个弱小运动目标检测问题,提出了一种基于粒子滤波的多目标检测前跟踪方法.首先分析了红外图像的状态及观测模型,然后将弱小运动目标的检测转化为贝叶斯滤波框架下的检测前跟踪,并利用多个粒子滤波器予以实现.为了评估该算法的有效性,采用人工合成的红外图像序列对其进行测试,结果说明该算法可同时处理低信噪比条件下的多个弱小运动目标,检测和跟踪性能良好.     

Particle filter using a new resampling approach applied to LEO satellite autonomous orbit determination with a magnetometer

Particle filter (PF) is widely used in nonlinear and non-Gaussian systems. Resampling is one of the significant steps in PF. However, PF using conventional resampling approaches may lead to divergent solutions because of the degeneracy phenomenon or sample impoverishment associated with a multidimensional system. In this article, an efficient alternative to conventional resampling approaches, called adaptive partial systematic resampling (APSR) with Markov chain Monte Carlo move and intelligent roughening is proposed for satellite orbit determination using a magnetometer. The results of the new resampling approach are compared with conventional resampling approaches and with unscented Kalman filter (UKF) for various initial errors in position and velocity, measurement sampling periods, and measurement noises to evaluate and verify the performance of the new resampling approach. The results of the new resampling approach in all cases are significantly better than the results of conventional resampling approaches. The velocity accuracy of the orbit determination of APSR is slightly poorer than UKF for relatively small initial errors, and small Gaussian measurement noise. However, the proposed approach yields more robust and stable convergence than UKF under large initial errors, long measurement sampling period, large Gaussian measurement noise, or non-Gaussian noise.     

多普勒盲区条件下的改进粒子滤波跟踪算法

机载预警雷达采用脉冲多普勒体制,具有良好的低空探测性能,但其存在不可忽略的多普勒盲区问题。在目标跟踪的过程中,该盲区容易造成目标航迹中断和重起批。针对多普勒盲区条件下的目标连续跟踪问题,提出了一种基于多假设运动模型的改进粒子滤波跟踪方法。该方法根据多普勒盲区对目标状态的约束形成多个可能的运动模型,粒子在每个模型下再进行带有状态约束的更新,当新出现的量测值落入任何一个运动模型形成的粒子云内,则航迹关联成功。仿真结果表明,该算法在不同盲区范围条件下,针对不同机动能力的目标均具有较高的航迹关联率,能够实现目标的连续跟踪。     

地月平动点尘埃动力学特性原位探测方法研究

为考察在地月平动点EML4、EML5可能存在的尘埃云的动力学特性，需研究能够覆盖较宽速度范围的尘埃颗粒动力学原位探测方法及技术。针对国内即将实施的Kordylewski尘埃云（KDC）探测任务，已有的栅网式尘埃速度矢量探测方法难以满足对高速颗粒的探测需求。作者提出通过在栅网式探测装置最后增加靶标以同时获得高速和低速颗粒的动力学特性参数，并对气凝胶靶标、电阻丝靶标、金属板靶标、电容式靶标、薄膜靶标及光帘式靶标等的适用性和优缺点进行分析比较后，建议采用栅网?金属电离复合式探测方法。     

基于上升轨迹可达范围的目标拦截发射窗口计算

针对低轨目标拦截任务,提出一种利用上升轨迹可达范围分析的发射窗口计算方法。首先,建立以上升时长和航程为性能指标的上升轨迹和优化模型,确定拦截器上升轨迹可达范围。然后,根据目标星下点与上升轨迹可达范围外包络的穿越关系以及拦截器的上升时长范围,对发射窗口进行初筛,得到准发射窗口。最后,针对每一段筛选出的准发射窗口,通过精确判定目标星下点与每一上升时长可达范围子环的位置关系,得到每段准窗口中能够实现目标拦截的发射窗口,将其取并集得到针对目标拦截的精确发射窗口。仿真表明本文提出的计算方法能够快速准确得到目标拦截的发射窗口。     

Influence of estimation errors on wayfinding-decisions in unknown street networks – analyzing the least-angle strategy

The least-angle strategy is a common wayfinding method that can be applied in unknown environments if the target direction is known. The strategy is based on the navigator's heuristic to select the street segment at an intersection which is most in line with the target direction. To use this strategy, the navigator needs to know the angles between the target direction and the street segments leading out from the intersection. If the direct view to the target is blocked and the target vector cannot be perceived, the target direction that is needed for the decision process is based on the agent's believed position and orientation (estimated through path integration). The agent's believed position and target direction are distorted by human errors in estimation of distances and directions, mainly affecting the path integration process. In this paper we examine how human estimation errors of distance and rotation influence the decision behavior in the wayfinding process in an unknown street environment. To demonstrate the geometrical consequences for a specific test case, we use a simulated software agent which navigates in a simulated street environment.     

Assessing the Risk of Orbital Debris Impact

A Space Debris Impact Risk Analysis Tool (SDIRAT) was developed and implemented to assess the orbital debris impact risk on a specified target in Earth orbit, in terms of flux, relative velocity, impact velocity, direction of the incoming particles, debris mass and diameter. Based on a deterministic approach, SDIRAT uses a realistic orbital debris population where each representative particle is identified by its rectangular coordinates (position and velocity) at a reference epoch. Using this information, some geometrical algorithms were developed and implemented to evaluate the contribution of each particle to the incoming flux. The position of the particle with respect to a specified target drives the selection criteria to reject, or select, it as a possible projectile. On the other hand, the relative velocity vector can be used to estimate the impact direction of the incoming flux. SDIRAT was conceived as a general tool for a variety of scenarios, such as low circular and elliptical orbits, up to the geosynchronous ring. This paper presents some examples of possible applications, including the computation of the incoming debris flux on SAX (low Earth orbit), SIRIO (geosynchronous orbit) and the IRIS upper stage (elliptical orbit). Other applications assess the impact risk for the Soviet Radar Ocean Reconnaissance Satellites Cosmos 1900 and Cosmos 1932.     

Effects of long-duration space flight on target acquisition

The purpose of the study was to explore the effects of long-duration space flight on the acquisition of specific visual targets in the horizontal plane. Seven cosmonauts (4 high performance pilots and 3 non-pilots) who had flown between 186–198 days on Mir served as subjects. Baseline testing was performed 4 times prior to launch and 4 times following landing at different intervals totrack recovery. During testing the subjects were required to acquire targets that were randomly presented with both a head and eye movement using a time optimal strategy. Prior to flight two unique head movement strategies, related primarily to piloting experience, were used for target acquisition. Non-pilots employed a Type-I strategy consisting of high velocity head movements with large peak amplitudes, while high performance pilots used primarily low velocity, small amplitude head movements (Type-II) to acquire the targets (p<0.02). For both strategies peak head velocities increased as the angular distance to the target increased (p<0.01) resulting in greater discrimination between strategies for the 60° targets. While preflight eye velocity between strategies did not reach statistical significance, postflight testing revealed a decrease in eye velocity for Type-I compared with their preflight performance (p<0.02) for the 60° targets. Postflight, the Type-I group showed a decrease in head velocity (p<0.20) while the Type-II group compensated by increasing head velocity (p<0.02). Variability for both of the head and eye parameters tended to increase postflight for both types of strategies.     

改进的迭代粒子滤波法在GPS数据处理中的应用

研究粒子滤波方法(PF)处理GPS测量数据.采用迭代Kalman滤波方法产生粒子滤波器中的粒子,得到更精确的目标状态估计.同时利用目标运动方程向后外推目标状态,综合利用外推的目标状态以及历史现测数据确定各粒子的权值,降低现测随机误差的影响,从而提高粒子滤波器的性能.采用改进的粒子滤波方法处理GPS测量数据,仿真结果表明...     

使用解析目标分流策略的海洋卫星多学科优化

在海洋卫星总体设计阶段,为了提高卫星整体设计性能和效率,采用多学科设计优化（MDO）技术对其总体参数进行设计优化。考虑轨道、有效载荷、结构和电源四个学科的设计变量和约束条件,以卫星总质量最小为优化目标建立海洋卫星MDO模型;针对各学科优化问题属性设计高效的优化方案并采用解析目标分流（ATC）这一多级MDO策略来协调整个优化过程。优化后海洋卫星总质量相对海洋一号（HY-1）卫星下降17.6%,整个优化过程花费的时间也比采用单级All at once(AAO)策略减少85.7%。一方面表明所建立的海洋卫星MDO模型的合理有效性,另一方面表明ATC策略结合学科定制优化方案的求解技术的有效性。