基于深度置信网络的近距空战态势评估

针对传统态势评估方法权值确定困难、大规模数据处理和特征提取能力不足的问题,结合当前空战数据特征,将深度置信网络(DBN)应用于近距空战态势评估。通过密度峰值算法对空战特征数据进行聚类分析,并结合态势函数和专家判读进行修正,建立标准空战态势样本库;以重构误差和测试错误率为基础,建立网络拓扑结构和最优参数确定方法,提高模型的训练效率,并通过样本数据,对模型进行训练和验证。实验表明,模型态势分类正确率达到92.7%,模型运行时间满足应用需求,实例评估结果与客观态势一致性强。     

分布式系统的容错测度研究(英文)

容错性是全面评价分布系统的重要方面。本文讨论了三种评价测度：节点／边连通度,生成树的数目及综合连通度。本文还给出了一个分析算例,算例表明,文中提出的综合连通度测度是合理的。     

基于神经网络的旁路活门自适应控制

ＦＬ－７风洞旁路活门非线性控制系统稳Ｍ数神经网络自适应控制中的网络拓扑结构，权值调整，自构性学习，及其实现方法表明，整个系统控制性能优良，不仅控制精度高，而且节省燃油，可靠性性能良好。     

INVERSE KINEMATICS FOR A 6 DOF MANIPULATOR BASED ON NEURAL NETWORK

运用神经网络求解机器人运动学位姿逆解，突破了文献局限于研究位置逆解的状况，首次实现自组织神经网络求解机器人姿态逆解。通过深入分析基于Ｋｏｈｏｎｅｎ网络原理和Ｖｉｄｒｏｗ－Ｈｏｆ误差修正的Ｍ．Ｒ．Ｓ．自组织神经网络及机器人运动学特性，创新了自组织神经网络训练算法并建立了一类工业机器人位姿逆解的神经网络方法。对ＰＵＭＡ５６０机器人的计算机仿真结果表明，本算法在自组织能力和定位控制精度方面大幅度提高。     

三维碎片拼合的轮廓提取和特征点检测

针对三维空间碎片的拼接,提出一种轮廓提取和特征点检测的方法。首先,采用一种简单、有效的方法来构建三维碎片数据的拓扑关系,并在此基础上提取碎片的轮廓曲线。对存在缺陷、拓扑关系难以建立的三维测量数据,中给出了相应的修改、调整方法。同时,提出了一种对轮廓进行重采样的方法来计算空间轮廓上每一点的曲率和挠率。将曲率和挠率相结合得出轮廓曲线上各点的全曲率,然后由全曲率值来检测碎片轮廓的特征点。轮廓曲线的描述就由其上每一点的曲率和挠率以及特征点来表征。实验表明了该算法的有效性和稳定性。     

优化驱动的起落架结构设计方法

起落架结构是飞机上最复杂最重要的结构之一,传统的设计方法依靠人工经验反复迭代,没有充分利用结构优化技术,具有设计周期长且不能最大限度得到最优设计的缺点。根据结构优化技术的发展,提出了优化驱动的起落架结构设计方法,实现了优化驱动的设计方法在起落架领域的完整工程应用。结构优化技术作为整个设计流程的驱动者,在其中发挥贯穿全程的主体作用,根据不同设计阶段的需求,先后运用拓扑优化、尺寸优化和形状优化技术。以某型飞机前起落架外筒的设计为例可以发现,相比传统设计方法,新方法在相同的设计约束条件下,能更快得到设计方案,结构质量减少了24.1%,实现了起落架结构的快速设计和轻量化设计。     

Material optimization: bridging the gap between conceptual and preliminary design

This paper presents a collection of tools for conceptual structure design. The underlying model is the ‘free material optimization’ problem. This problem gives the best physically attainable material and is considered as a generalization of the sizing/shape optimization problem. The method is supported by powerful optimization and numerical techniques, which allow us to work with bodies of complex initial design and with very fine finite-element meshes. The computed results are realized by composite materials. We consider a particular class of fibre-reinforced composite materials, manufactured by the so-called tape-laying technology. In the post-processing phase, stress/strain-trajectories are plotted which indicate how to place these tapes; they also show the proposed thickness of the tapes. Several examples demonstrate the efficiency of this approach to conceptual design of engineering structures.     

Locating Current Sheets in the Solar Corona

Current sheets are essential for energy dissipation in the solar corona, in particular by enabling magnetic reconnection. Unfortunately, sufficiently thin current sheets cannot be resolved observationally and the theory of their formation is an unresolved issue as well. We consider two predictors of coronal current concentrations, both based on geometrical or even topological properties of a force-free coronal magnetic field. First, there are separatrices related to magnetic nulls. Through separatrices the magnetic connectivity changes discontinuously. Coronal magnetic nulls are, however, very rare. Second, inspired by the concept of generalized magnetic reconnection without nulls, quasi-separatrix layers (QSL) were suggested. Through QSL the magnetic connectivity changes continuously, though strongly. The strength of the connectivity change can be quantified by measuring the squashing of the flux tubes which connect the magnetically conjugated photospheres. We verify the QSL and separatrix concepts by comparing the sites of magnetic nulls and enhanced squashing with the location of current concentrations in the corona. Due to the known difficulties of their direct observation, we simulated coronal current sheets by numerically calculating the response of the corona to energy input from the photosphere, heating a simultaneously observed Extreme Ultraviolet Bright Point. We did not find coronal current sheets at separatrices but at several QSL locations. The reason is that, although the geometrical properties of force-free extrapolated magnetic fields can indeed hint at possible current concentrations, a necessary condition for current sheet formation is the local energy input into the corona.     

The knowledge of knots: an interdisciplinary literature review

ABSTRACT

Knots can be found and used in a variety of situations in the 3D world, such as in vines, in the DNA, polymer chains, electrical wires, in mountaineering, seamanship and when ropes or other flexible objects are involved for exerting forces and holding objects in place. Research on knots as topological entities has contributed with a number of findings, not only of interest to pure mathematics, but also to statistical mechanics, quantum physics, genetics, and chemistry. Yet, the cognitive (or algorithmic) aspects involved in the act of tying a knot are a largely uncharted territory. This paper presents a review of the literature related to the investigation of knots from the topological, physical, cognitive and computational (including robotics) standpoints, aiming at bridging the gap between pure mathematical work on knot theory and macroscopic physical knots, with an eye to applications and modeling.     

连续dcpo上lawson拓扑的紧性

由于dcpo上赋予lawson拓扑后相对于s cot t拓扑会用更好的性质(由于lawson拓扑细于s cot t拓扑)。因此人们对lawson拓扑及lawson开集都很关注。本文主要讨论了连续dcpo上lawson拓扑紧性的问题。指出即使连续dcpo D是凝聚的，也不能保证D是lawson紧的，同时给出了一些lawson拓扑紧的例子。