高精度秒脉冲信号的产生

通过介绍高精度秒脉冲信号的实现原理以及实现中的关键技术,分析了如何减小由码跟踪环路测量误差导致的秒脉冲信号随机抖动误差,提出了减小由直接数字频率合成器导致的秒脉冲信号系统钟模糊误差的实现方法。最后给出了在＂北斗＂一号卫星标校机工程应用中的实际性能,统计精度（1倍标准差）为1.7ns,保证了标校设备能高精度地服务于BD-1导航定位系统。     

Inversion of satellite gradiometry data using statistically modified integral formulas for local gravity field recovery

The satellite gravity gradiometric data can be used directly to recover the gravity anomaly at sea level using inversion of integral formulas. This approach suffers by the spatial truncation errors of the integrals, but these errors can be reduced by modifying the formulas. It allows us to consider smaller coverage of the satellite data over the region of recovery. In this study, we consider the second-order radial derivative (SORD) of disturbing potential (T_rr) and determine the gravity anomaly with a resolution of 1° × 1° at sea level by inverting the statistically modified version of SORD of extended Stokes’ formula. Also we investigate the effect of the spatial truncation error on the quality of inversion considering noise of T_rr. The numerical investigations show satisfactory results when the area of T_rr coverage is the same with that of the gravity anomaly and the integral formula is modified by the biased least-squares modification. The error of recovery will be about 6 mGal after removing the regularization bias in the presence of 1 mE noise in T_rr measured on the orbit.     

相对差分单向测距（△DOR）技术在探月工程中的应用研究

相对差分单向测距（△DOR）是一种高精度测角技术,在深空探测任务的高精度测定轨方案中得到广泛应用．介绍了△DOR测量的基本原理及CCSDS（国际空间数据系统委员会）关于△测量的相关标准．针对中国探月工程月面软着陆任务提出的高精度测定轨要求,分析了△DOR测量的信号形式、工作模式以及测量精度,推导了解模糊过程．研究表明,△DOR测角精度达到25nrad,对应到月球上的距离约为10m．     

The photometric study of light scattering from the surface of alumina powder and interpretations by Hapke formula

A laboratory experiment helps to understand the light scattering property of regolith like samples with known compositions and other physical parameters. The laboratory data so obtained can be compared with the existing in situ data on celestial objects like asteroids. Further, it may be analyzed with the help of various theoretical models to understand the light scattering processes from regolith more clearly. In this work we have performed laboratory based photometry of the light scattered from the surfaces of powdered alumina (Al₂O₃) at various tilt angles of the sample and at large phase angles, with the particles having diameter 0.3 μm. The wavelength of observation was 632.8 nm. These data have been fitted by a surface scattering model originally suggested by Hapke. Instead of using empirical Henyey–Greenstein phase function to fix the values of albedo and phase function to be used within Hapke formula, we have used Mie theory for the same. This approach helped us to determine the single particle properties such as particle diameter and complex refractive index from surface scattering phase curve alone. Mie theory depends only on the size parameter X(=2π(radius/wavelength)) and complex refractive index (n, k) of the material. Since the absorption coefficient (k) for alumina is known to be very low but not exactly zero, the best fit to the experimental data was obtained by least square technique with k as a free parameter, as the other parameters are known. Finally, we compare our results with other published results and discuss the scope of application of the method we adopted.     

基于人工神经网络的柔性机翼挠度预测

为实现对承载后柔性机翼挠度的准确预测,在全面分析柔性机翼挠度的影响因素基础上,应用正交试验法确定的影响柔性机翼挠度的主要因子作为输入变量,挠度作为输出变量,以大量试验数据为训练样本,通过多次试取隐含层和各隐含单元,并选取trainlm作为最优训练函数,最终建立了预测柔性机翼挠度的BP(Back Propagation)人工神经网络模型.在此基础上,随机选取试验结果中的12组试验样本,连续进行10次挠度预测,预测结果和试验实测值最大相对误差和标准方差分别为4.481%,1.033 7.解析结果表明:柔性机翼挠度预测结果与实验值吻合的较好,建立的人工神经网络预测模型具有较高的预测精度.     

基于神经网络的二维随机翼型优化设计方法(英文)

为避免翼型单点优化设计存在的非设计状态气动性能损失,改进对实际飞行环境中不确定性因素的适应能力,提出了基于神经网络的二维随机翼型优化设计方法。采用4个BP神经网络作为代理模型,用于预测翼型的气动系数与几何参数,以提供高效和可靠的分析。同时联合服从正态分布的概率密度函数和遗传算法构成了优化设计方法。采用该方法,对GA(W)-2翼型,在关于马赫数和迎角的二维飞行区域内进行了随机优化设计。通过与原始翼型和单点优化设计翼型的结果对比,表明该二维随机优化方法能够在指定飞行区域内改进翼型的整体性能,提高了翼型对多个飞行参数随机变化的适应能力。     

自适应神经模糊推理模型在复杂社会技术系统中的应用(英文)

考虑复杂社会技术系统的特点,运用自适应神经模糊推理模型对提高公益科研机构的组织效率进行了实例研究。建立了针对公益科研机构员工激励决策的自适应模糊推理完整模型,利用调研数据及其处理结果验证了模型的有效性。该模型融合了神经网络的学习机制和模糊系统的语言推理能力,弥补了神经网络和模糊逻辑系统各自的不足。研究结果表明,自适应神经模糊推理模型适用于复杂社会技术系统,在解决管理领域的预测、评估和决策问题中有广阔的应用前景。     

SVM和神经网络在电能质量扰动分类应用中的对比

提出了一种电能质量动态扰动特征向量的提取方法,分析比较了多种分类器对电能质量动态扰动的分类能力。首先采用小波包分解算法对电能质量信号某一频段内的信息进行精细分解从而提取出特征向量,然后针对该特征向量构造了相应的BP神经网络、学习向量量化(Learning vector quantization,LVQ)神经网络、自组织特征映射(Self-organizing map,SOM)神经网络及支持向量机(Support vector machine,SVM)分类器,并模拟实际电网中的复杂扰动信号提取其特征向量集,对多种分类器的分类能力进行对比。仿真结果表明,在较复杂的电能质量扰动情况中,支持向量机分类器仍能实现对信号的精确分类,对电能质量监测具有很好的应用价值。     

基于BIT的工程机械智能在线检测方法研究与实现

以某工程机械为研究对象,研究基于CAN总线的嵌入式BIT智能检测方法,研制具有高可信度的工程机械智能故障诊断仪,以提高工程机械作业的可靠性和安全性。     

L形声学谐振器的吸声特性试验

从试验方法、试验装置、试验数据处理等方面对如何有效测试L型声学谐振器法向和切向的吸声特性(包括吸声系数和声阻抗)展开了详细论述,并得出有一定工程价值的试验结论.