求解具有多层试验数据成败型单元混联系统可靠性近似限的信息论方法

研究了求解具有多层成败型试验数据时,系统可靠性置信下限近似解的信息论方法,给出了各层试验数据综合的方法和步骤,推导了各层数据信息论方法折合及综合的计算公式。根据单元试验信息提供的信息量与系统折合试验应提供的信息量相等的原则,将多层试验数据逐级折合为系统的等效成败型数据,导出了等效成败型数据信息论方法折合的基本公式;结合可靠性工程的基本原理,根据单个成败型单元可靠性评定的经典精确置信限方法,给出了系统可靠性置信下限的近似解。并以实例形式说明了该方法的具体应用。     

纤维缠绕结构的一种可靠性优化设计方法研究

将纽曼级数展开的蒙特卡罗随机有限元应用于具有随机因素的纤维缠绕结构的应力求解;然后利用蒙特卡罗直接比较法和蔡希布（Ｔｓａｉ－Ｈｉｌｌ）失效判据给出该结构的可靠性算式,最后建立了此类结构的可靠性优化设计迭代法,并给出了一算例。     

基于支持向量机的航空发动机滑油监控分析

提出了一种基于支持向量机的航空发动机滑油金属含量预测方法。详细分析了支持向量机用于时间序列预测的理论基础,并给出了运用支持向量回归进行多步预测的一般公式,提出了用最终预报误差(FPE)准则优化选取嵌入维数。与传统的AR预测模型相比,支持向量机由于采用了新型的结构风险最小化准则表现出优秀的推广能力。经过数值仿真得出自回归(AR)模型仅适合于短期预测;支持向量机预测推广能力强、具有较强的鲁棒性和容错性,对较长区间预测仍具有较好的效果。最后,将其应用于某型发动机滑油的铁金属含量预测,取得了较好的效果。      

CHAMP卫星加速度计标校参数序列的谱分析

给出了利用能量守恒方法求定CHAMP卫星加速度计标校参数的基本原理和算法．对2001年8月-2003年12月共约两年半的CHAMP卫星加速度计数据进行了标校,导出了尺度、偏差和偏差漂移参数的时间序列,分别采用最小二乘谱分析和抗差谱分析对这些时间序列进行了分析．结果表明,尺度、偏差和偏差漂移参数都有明显的周期变化,但各主周期项略有差异．尺度的主周期是441d、220d、126d和13．4d的周期变化,其中220d和441d的周期变化的贡献相对较大．偏差中有明显的220d、126d和13．4d周期变化,各周期变化的贡献几乎相当;偏差漂移的主周期是220d、13．4d和9d的周期变化,其中13．4d的周期变化最为显著,几乎比其他周期项变化高出一个数量级．另外尺度和偏差参数还有明显的长期变化,而偏差漂移参数的长期变化不明显．     

Shaping low-thrust trajectories with thrust-handling feature

Shape-based methods are becoming popular in low-thrust trajectory optimization due to their fast computation speeds. In existing shape-based methods constraints are treated at the acceleration level but not at the thrust level. These two constraint types are not equivalent since spacecraft mass decreases over time as fuel is expended. This paper develops a shape-based method based on a Fourier series approximation that is capable of representing trajectories defined in spherical coordinates and that enforces thrust constraints. An objective function can be incorporated to minimize overall mission cost, i.e., achieve minimum $\mathrm{\Delta }V$. A representative mission from Earth to Mars is studied. The proposed Fourier series technique is demonstrated capable of generating feasible and near-optimal trajectories. These attributes can facilitate future low-thrust mission designs where different trajectory alternatives must be rapidly constructed and evaluated.     

管道喷涂机器人:结构与位姿调整

针对工业上大型管道内壁喷涂的需求,设计了用于喷涂管道的机器人,该机器人采用串并联结构形式,由支撑平台和串联机械手组成.支撑平台采用几何的方法建立数学模型,串联机械手采用D-H参数法建模;并在数学模型基础上给出了逆解方程.位姿调整时,对机器人进行初定位,利用激光跟踪仪测量出管道和机器人的位姿,求解出调整量,对机器人进行位姿调整,使喷枪旋转轴和管道轴线重合.管道喷涂机器人在工厂进行试验测试,测试结果表明该装备可以很好地实现管道的喷涂,提高喷涂质量和效率.      

The Big Pine Creek watershed and climate change: A trend analysis of Landsat surface reflectance and PRISM datasets over the last 3 decades

Recent variations in normal meteorological conditions indicate the earth’s climate is changing in ways that may impact delicate ecological balances in sensitive regions. Identifying how those changes are affecting the biosphere is essential if we are going to be able to adapt to those changes and to potentially mitigate their harmful consequences. This paper presents a time series study of an alpine ecosystem in the Big Pine Creek watershed in California’s Eastern Sierra Nevada Mountain’s. Raw Landsat data covering the years 1984 through 2011 is converted to observed surface reflectance and analyzed for trends that would indicate a change in the ecosystem. We found that over the time period of the study, observed surface reflectance shows a general decline across the spectrum while our analysis of environmental data demonstrates statistically significant increases in temperatures. While declining reflectance in the visible and short wave bands are indicators of increased surface cover, the fact that the IR band also shows declines is consistent with a decline in tree density. This study provides a useful insight into the ecological response of the Big Pine Creek watershed to recent climate change. These findings suggest that alpine ecosystems are particularly sensitive to increasing temperatures. If these results are replicated in other alpine watersheds it will demonstrate that the biosphere is already showing the effects of a warmer environment.     

利用神经元拓展正则极端学习机预测时间序列

为实现对于时间序列预测数据的准确预测,提出一种神经元拓展正则极端学习机(NERELM,Neuron-Expanding Regularized Extreme Learning Machine),并研究了其在时间序列预测中的应用.NERELM根据结构风险最小化原理权衡经验风险与结构风险,以逐次拓展隐层神经元的方式自动确定最佳的网络结构,以避免传统神经网络训练过程中需人为确定网络结构的弊端.应用于时间序列的仿真结果表明:NERELM可有效实现对于RELM最佳网络结构的自动确定,具有预测精度高与计算速度快的优点.      

串并混联空间机械臂全自由度轨迹规划算法

目前,国内外对串并混联空间机械臂的研究较少;而且大多在舍弃冗余自由度的情况下进行轨迹规划.针对7自由度串并混联空间机械臂,提出了一种基于解析解、并且充分放开全部7个自由度的机械臂轨迹规划算法.该规划算法既扩大了串并混联空间机械臂的规划可达空间,又保证了规划的实时性,同时还保留了结构承载能力强的优点,可以充分发挥其作为空间抓捕用臂的优势.通过仿真和地面空间操作抓捕试验,验证了算法的有效性.