逆设计方法研究

本文以梁元结构为例,推导满足结构静、动力特性的结构逆设计计算方法。     

喷管喉衬结构瞬态温度测量及导热反问题方法应用

为了获得实时可靠的固体火箭发动机喉衬结构瞬态温度,采用内埋快响应烧蚀热电偶对发动机喉衬结构瞬态温度进行测量,并应用导热反问题方法(Beck序列函数法)计算其喉部内壁面温度和热流密度。采用的试验方案有效避免了传感器探头与喉衬内壁面烧蚀速率不同而干扰流场或中途烧毁等现象发生。试验结果显示越靠近内壁面的测量点温度峰值越高,测点温度瞬态变化与发动机工作状态同步性越好。计算结果显示,计算的平均误差为5.6%,最大误差为7.9%,其内壁温度最高出现在发动机工作结束时,热流密度在0.1s内迅速上升,而后又迅速下降,最大值约为30MW/m2。在发动机停止工作2s后,喉衬开始向外界传热。     

基于随机过程的可靠性建模

基于Wiener过程、Gamma过程和逆高斯(Inverse Gaussian,IG)过程等随机过程的可靠性建模方法在装备可靠性评估中逐步得到应用.针对3类考虑随机影响的随机过程模型,分别阐述了各类模型在一元性能退化建模、二元性能退化建模、加速退化建模和竞争失效建模等情形下的应用情况,并展望了下一步研究方向.研究工作对...     

开链多刚体系统逆动力学的一种并行计算模型

本文通过采用方向余弦描述物体的转动，物体的质心坐标描述物体的平动，用Ｎｅｗｔｏｎ－Ｅｕｌｅｒ方程建立了开链多刚体系统逆动力学的并行计算模型。在建模过程中由于选用了不独立广义坐标并利用坐标变换方法使这种模型成为非递推形式，与其它模型相比这种模型具有更好的内在并行性。最后以ＰＵＭＡ５６０机器人的前三个臂为例，进行了并行计算效率分析，并与其它方法进行了比较，结果是令人满意的。     

行人导航系统中的MEMS误差在线修正技术

在基于微惯性器件的行人导航系统中,陀螺仪和加速度计的偏移是降低系统定位精度的重要因素。传统的标定方法大多在实验室中进行,后续导航解算都是基于标定后的固定模型,然而MEMS器件长时间工作后,标定模型参数发生变化会导致系统导航性能下降。通过分析行人导航系统及MEMS器件的特点,提出了一种基于误差模型的MEMS器件参数在线修正方法。根据行人行走的特点,检测并区分行走过程中的可修正区间与不修正区间。在可修正区间基于逆向解算算法实现了对陀螺仪和加速度计零偏的在线修正,并提出了主航向反馈修正算法,提高了行人导航系统长时间导航性能。实验结果表明,40m行走实验中,系统定位精度提升了9.07%;300m行走实验中,系统定位精度提升了13.14%。     

反抽样数字滤波器设计

根据 DAC的幅频特性 ,当输入信号频率接近奈奎斯特采样率时 ,输出信号幅度衰减约 3 .9d B。文中设计了补偿此衰减的数字滤波器 ,并提出一种节省资源的高效实现方法。最后 ,给出了实验结果 ,以证明此滤波器补偿功能的可行性     

基于模糊逻辑的月球车逆运动学求解方法

研究了月球车逆运动学问题，即在月面数字地形图、月球车水平面坐标和航向角给定时，实时估计6轮摇臂转向架式月球车质心高度、位形（configuration）和姿态中的俯仰角和横滚角的问题。分析了6轮摇臂转向架式（rocker-bogie）月球车运动学模型，提出使用模糊逻辑结合虚拟传感器快速实时求解月球车逆运动学问题的新方法。此方法在月球车三维仿真平台中得到验证。实验结果表明此方法求解速度小于1／40s，同时保证月球车车轮到地面的距离小于0．02m。     

考虑臂杆柔性的空间机器人逆动力学仿真研究

本文在考虑臂杆柔性的情况下，利用拉格朗日公式推导出空间机器人操作臂的动力学方程。基于所得到的非线性模型，在最优控制理论的基础上，用Ｔｒｅａｎｏｒ法结合打靶法研究了其动力学逆问题的解法及振动的有效控制问题，并以平面二臂杆空间机器人操作臂为例进行了逆动力学仿真的研究，仿真结果很好。     

Ionosphere tomography based on spaceborne SAR

Two models of ionosphere tomography based on spaceborne SAR (Synthetic Aperture Radar) are proposed. For HF-SAR the signal with sweeping frequency lower than the characteristic frequency of ionosphere will be scatted during the ionosphere propagation and completely reflected at a corresponding height. The ionospheric electron density isolines looked as series of random surfaces can be reconstructed from the HF-SAR echoes by using the inverse scattering technique for layered rough surfaces and the method of moment (MoM). The numerical simulation show that due to the MoM can provide a full wave solution, the ionosphere tomography with high resolution can be obtained as long as enough sampling data of HF-SAR echoes are used. For VHF/UHF/P/L-band SAR the TEC (Total Electron Content) can be obtained from the SAR echoes scattered by some strong point targets (such as the calibrators, etc.) appeared in the SAR imaged ground region, and the ionosphere tomography can be performed by CT technique.     

Modeling of LEO orbital debris populations for ORDEM2008

The NASA Orbital Debris Engineering Model, ORDEM2000, is in the process of being updated to a new version: ORDEM2008. The data-driven ORDEM covers a spectrum of object size from 10 μm to greater than 1 m, and ranging from LEO (low Earth orbit) to GEO (geosynchronous orbit) altitude regimes. ORDEM2008 centimeter-sized populations are statistically derived from Haystack and HAX (the Haystack Auxiliary) radar data, while micron-sized populations are estimated from shuttle impact records. Each of the model populations consists of a large number of orbits with specified orbital elements, the number of objects on each orbit (with corresponding uncertainty), and the size, type, and material assignment for each object. This paper describes the general methodology and procedure commonly used in the statistical inference of the ORDEM2008 LEO debris populations. Major steps in the population derivations include data analysis, reference-population construction, definition of model parameters in terms of reference populations, linking model parameters with data, seeking best estimates for the model parameters, uncertainty analysis, and assessment of the outcomes. To demonstrate the population-derivation process and to validate the Bayesian statistical model applied in the population derivations throughout, this paper uses illustrative examples for the special cases of large-size (>1 m, >32 cm, and >10 cm) populations that are tracked by SSN (the Space Surveillance Network) and also monitored by Haystack and HAX radars operating in a staring mode.