Whistler waves in plasmas with time-varying magnetic field: Laboratory investigation

Modulation of whistler waves in a plasma with time-dependant magnetic field perturbations was investigated experimentally. The experiments were performed on large “Krot” device, which was specially designed to study space plasma physics phenomena. It is shown that magnetic field variations on the wave propagation path can lead to splitting of initially continuous whistler wave into the sequence of bursts, whose repetition rate corresponds to magnetic field perturbation period. The frequency inside each burst is changing from its front edge to the back edge. Relative shift of the wave frequency can be as large as the relative magnetic disturbance. Distortion of whistler wave frequency spectrum after its passing through magnetically disturbed areas can be used as a diagnostics for low-frequency magnetic field variations. The applicability of our laboratory results to space plasma is discussed.     

Impact of the time span selected to calibrate the ballistic parameter on spacecraft re-entry predictions

The trajectory modeling of satellites that are re-entering the Earth’s atmosphere, as a result of natural orbital decay, has always been a challenging task. Residual lifetime estimations and re-entry predictions are affected by substantial uncertainties, associated with atmospheric density models, with the forecasts of the relevant solar and geomagnetic activity indices and with tracking data, which for uncontrolled re-entries are usually sparse and not particularly accurate. Furthermore, modeling the aerodynamic forces that act on low altitude satellites is a formidable task, especially for objects of a complex shape and unknown attitude evolution.     

空间遥感器大口径轻量化扫描镜材料发展与制作方法探讨

扫描镜是光机扫描型空间遥感器的重要组成元件。文章针对目前空间遥感器上应用的大口径轻量化扫描镜,从材料和制作方法方面进行了一定的技术研究。     

结构方程模型的预测建模方法

提出一种结构方程模型的动态预测建模方法,从而可以在无须未来样本数据的情况下,预测系统要素之间未来的因果关系。采用矩阵谱分解,将协方差矩阵唯一分解为特征值矩阵和特征向量矩阵乘积的形式.分别应用经典的线性回归方法和高维群点主轴旋转预测方法对特征值矩阵和特征向量矩阵建立预测模型,提出一种协方差矩阵的后推预测算法.采用极大似然法,迭代估计未来结构方程模型的各种参数.仿真实验例示了该方法的主要计算步骤.计算结果显示,利用本模型得到的拟合值精度较高,预测模型真实可信,表明这种方法可以用于分析和预测结构方程模型.      

飞机大迎角滚转运动频率域建模与稳定性分析

本文运用频率域建模法,分别建立了飞机大迎角滚转运动的Ｆｏｕｒｉｅｒ变换模型及非线性代数模型,并从数学模型中提取动导数。基于非线性非定常数学模型与局部线化的动导数数学模型,计算了飞机滚转运动的稳定性,并运用李雅普诺夫稳定性分析理论,进行飞机滚转运动稳定性分析。结果表明,大振幅非定常滚转运动实验可以用于飞机绕体轴滚转运动的稳定性分析。从大振幅实验数据提取的动导数,基本反映了飞机的非定常气动力特性。通过     

HISTORICAL PROCEDURES AND G-DSG METHOD BASED MANUFACTURING PLANNING

Oriented to CAD/CAM seamless integration, this paper presents an idea for synthetically considering the qualitative history model, which represents the whole course of modeling, and the quantitative geometry model, which contains the extended Brep model, CSG and feature pedigree. History model building captures in background the dynamic interactive definition of engineering requirement and then explicitly conveys the original intention to successive application layers, which isc onductive to the decision support of manufacturing planning in not only automatic geometry re-constructing but also machining set-up. G-DSG theory as an earlier achievement is applied to generate the topology-independent generalized mid-model as an input to manufacturing planning, which istherefore simplified and its accuracy is simultaneously improved. Manufacturing planning lays emphasis on optimizing the mapping in both geometry and function from part itself to detail machining scheme. Comparatively, process planning pays more attention to the mapping to those items like tool, fixture, etc. Theoretically, such an idea is also beneficial to realizing the parametric NC machining trajectory generation and maintaining its dynamic consistency with the update of the design model.     

BP神经网络用于传感器故障诊断的仿真研究

在标准逆传播神经网络的基础上 ,提出了一种全新的改进逆传播神经网络 ,用来消除标准逆传播神经网络收敛速度慢和易陷入局部最小点的缺陷 ;并且设计了一个主神经网络和三个分布神经网络的结构 ,通过神经网络的在线学习 ,得到需要的参数估计。用它与传感器测得的实际值比较 ,可判断出故障 ,并且给出了仿真实例。     

射频接收机系统级建模中的噪声谱分析

提出了一种在有限的系统参数下进行射频接收机系统建模时,各子电路噪声参数的分析和计算方法.从噪声功率谱的等效计算出发,给出了加性高斯白噪声通过有噪线性电路的分析方法.以一通信电台为例进行了分析计算,在给定有限的系统参数(信噪比,灵敏度,增益等)下分析出合理的各子电路的噪声参数,并进行建模仿真验证,理论结果与仿真结果较好吻合,表明了分析方法的有效性.该噪声功率谱的等效分析计算方法对射频系统建模有很好的指导作用.      

嵌入式隐Markov模型的分段训练方法

针对嵌入式隐Markov模型再学习问题,提出了分段训练方法用于人脸识别:把当前的训练样本看作整体训练样本的一部分,训练结束后存储训练后的模型参数和中间变量;增加新样本后,以当前模型参数作为初始模型参数,用新增样本训练模型,得到新的中间变量,最后将已存储的中间变量和用新样本计算得到的中间变量合成,得到最终的模型.人脸识别实验结果表明了该方法的可行性.      

一种面向星载计算机的功能级功耗估计方法

在星载计算机的软硬件协同设计过程中需要解决处理器功耗估计问题.现有的工具和方法主要面向ARM或DSP处理器,缺乏可用于LEON3处理器的软件功耗估计方法.针对这一不足,基于功能级功耗分析方法进行功耗估计,重点解决LEON3处理器的功耗建模问题.提出了从C代码中提取模型参数的方法,并且对序贯最小优化支持向量回归算法进行了改进,提高模型的精度.在数值试验部分,通过常用的C语言算法试验对估计方法进行了验证.结果表明功耗估计误差小于3%,能够满足星载计算机软硬件协同设计的需要.