基于异构网络表示学习的评分预测模型

深入分析电商行业的用户个性化数据并提供推荐服务近年来已成为业界的热点。推荐服务的基础是对用户的潜在兴趣进行挖掘，并对商品的感兴趣程度进行预测。因此，以此为背景，研究用户对商品的评分预测。对电商业的关系型数据在推荐系统中的应用进行了研究，提出了通过使用网络表示学习进行评分预测的方法。首先，将关系型数据构建成异构网络，用户和商品为网络中的节点。然后，设计了兼顾网络结构信息和节点之间相似性的个性化异构网络采样方法，并对节点进行表示学习。最后，将学习到的用户、商品表示向量输入到神经网络中进行训练，利用优化后的神经网络模型进行评分预测。实验结果表明:所提方法在YELP 13、Movielens 100k、Movielens 1m数据集上都有较高的准确率，对比常用方法，准确率提升6.5%以上。      

阳极氧化铝模板法制备CoPt纳米管

采用电化学直流沉积,首次在氧化铝模板中成功制备出直径为50 nm的CoPt纳米管高度有序阵列。利用透射电子显微镜(TEM)、扫描电子显微镜(SEM)、X射线衍射仪(XRD)以及振动样品磁强计(VSM)对样品的结构和磁学特性进行测试。分析表明,CoPt纳米管是以面心立方(fcc)结构存在。当外加磁场与纳米线平行时,CoPt合金纳米管的矫顽力达450 Oe,这种大面积制备的CoPt纳米管希望在超高密度信息磁存储上得到应用。     

超声振动体的耦合振动研究

对一些典型的大尺寸超声振动体如圆柱体、空心圆柱体、长条形六面体和大矩形六面体。本文用通用有限元软件对耦合系数的分布进行了分析，并在耦合振动频率方程的基础上研究了耦合系数与耦合振动的强度之间的关系。     

Modelling of the optical observations of a geosynchronous fragmentation event

A study of the evolution and optical detectability of a fragmentation debris cloud in geosynchronous orbit has been carried out. The 1998 NASA breakup model has been used to generate orbit data for 95 fragments larger than 10 cm size from a 1000 kg satellite. The orbital evolution of these fragments is studied using a precision numerical propagator, employing a high-fidelity force model. Although the fragments rapidly disperse throughout the geostationary arc, they remain localised in right ascension of ascending node and inclination, and are driven along a narrow inertial corridor by luni-solar perturbations. The ESA PROOF software is used to study the detectability of the fragments using a 1- and 0.5-m telescope design. The 1-m telescope can detect 82% of the fragments (down to 13 cm in size) whilst the 0.5-m telescope can detect 39% of the fragments (down to 30 cm size). Due to the large along-track spread of the fragments, a time limit of 1-month post-breakup can be established for a space surveillance system to catalogue the breakup fragments. After this time the angular separation is such that the fragments disperse into the background population, and are no longer distinguishable as originating from a common breakup event.     

深空探测通信技术发展趋势及思考

阐述了深空探测通信的重要地位和意义，就深空探测通信中的主要问题和特点开展论述，分析了深空探测通信的国内外研究现状。提出了利用地球静止轨道卫星编队的方法构建我国自主的天基连续测控和通信系统的构想，并以月球为中继、利用地月系统内的拉格朗日点开展深空探测。建议当前开展深空探测通信中星际互联网络体系结构、高效信道编码技术和空间协议研究的构想。     

Coupled thermal,structural and vibrational analysis of a hypersonic engine for flight test

This paper describes a relatively simple and quick method for implementing aerodynamic heating models into a finite element code for non-linear transient thermal-structural and thermal-structural-vibrational analyses of a Mach 10 generic HyShot scramjet engine. The thermal-structural-vibrational response of the engine was studied for the descent trajectory from 60 to 26 km. Aerodynamic heating fluxes, as a function of spatial position and time for varying trajectory points, were implemented in the transient heat analysis. Additionally, the combined effect of varying dynamic pressure and thermal loads with altitude was considered. This aero-thermal-structural analysis capability was used to assess the temperature distribution, engine geometry distortion and yielding of the structural material due to aerodynamic heating during the descent trajectory, and for optimising the wall thickness, nose radius of leading edge, etc. of the engine intake. A structural vibration analysis was also performed following the aero-thermal-structural analysis to determine the changes in natural frequencies of the structural vibration modes that occur at the various temperatures associated with the descent trajectory. This analysis provides a unique and relatively simple design strategy for predicting and mitigating the thermal-structural-vibrational response of hypersonic engines.     

多种观测数据驱动的三维行星际太阳风MHD模拟

三维磁流体力学（MHD）数值模拟是行星际太阳风研究的重要手段.本文发展了一种由多种观测数据驱动的三维行星际太阳风MHD数值模型.模型的计算区域为0.1AU到1AU附近，使用Lax-Friedrich差分格式在六片网格系统中进行数值求解.边界条件中磁场使用GONG台站观测的光球磁图外推获得，密度通过LASCO观测的白光偏振亮度反演得到，速度根据以上两种观测数据并利用一种基于人工神经网络技术（ANN）的方法得到，温度通过自洽方法根据磁场和密度导出.利用该模型模拟了第2062卡灵顿周（CR2062）时期的行星际太阳风，模拟结果显示出丰富的观测特征，并与OMNI以及Ulysses的实际观测值符合得较好.该模型可用于提供接近真实的行星际太阳风，有助于提高空间天气预报的精度.      

Aerodynamic performance enhancement for flapping airfoils by co-flow jet

Introducing active flow control into the design of flapping wing is an effective way to enhance its aerodynamic performance. In this paper, a novel active flow control technology called Co-Flow Jet (CFJ) is applied to flapping airfoils. The effect of CFJ on aerodynamic performance of flapping airfoils at low Reynolds number is numerically investigated using Unsteady Reynolds Averaged Navier-Stokes (URANS) simulation with Spalart-Allmaras (SA) turbulence model. Numerical methods are validated by a NACA6415-based CFJ airfoil case and a S809 pitching airfoil case. Then NACA6415 baseline airfoil and NACA6415-based CFJ airfoil with jet-off and jet-on are simulated in flapping motion, with Reynolds number 70,000 and reduced frequency 0.2. As a result, CFJ airfoils with jet-on generally have better lift and thrust characteristics than baseline airfoils and jet-off airfoil when C_μ is greater than 0.04, which results from the CFJ effect of reducing flow separation by injecting high-energy fluid into boundary layer. Besides, typical kinematic and geometric parameters, including the reduced frequency and the positions of the suction and injection slot, are systematically studied to figure out their influence on aerodynamic performance of the CFJ airfoil. And a variable C_μ jet control strategy is proposed to further improve effective propulsive efficiency. Compared with using constant C_μ, an increase of effective propulsive efficiency by 22.6% has been achieved by using prescribed variable C_μ for NACA6415-based CFJ airfoil at frequency 0.2. This study may provide some guidance to performance enhancement for Flapping wing Micro Air Vehicles (FMAV).     

Hybrid integration method for highly maneuvering radar target detection based on a Markov motion model

To detect highly maneuvering radar targets in low signal-to-noise ratio conditions, a hybrid long-time integration method is proposed, which combines Radon-Fourier Transform (RFT), Dynamic Programming (DP), and Binary Integration (BI), named RFT-DP-BI. A Markov model with unified range-velocity quantification is formulated to describe the maneuvering target’s motion. Based on this model, long-time hybrid integration is performed. Firstly, the whole integration time is divided into multiple time segments and coherent integration is performed in each segment via RFT. Secondly, non-coherent integration is performed in all segments via DP. Thirdly, 2/4 binary integration is performed to further improve the detection performance. Finally, the detection results are exported together with target range and velocity trajectories. The proposed method can perform the long-time integration of highly maneuvering targets with arbitrary forms of motion. Additionally, it has a low computational cost that is linear to the integration time. Both simulated and real radar data demonstrate that it offers good detection and estimation performances.     

基于时序相关性的信息效用评价模型研究

事务处理中的时间相关性数据蕴涵着对预测与决策具有情报效用的信息,可以通过构建一定的特征模式对其效用大小进行评价。文章对时变序列数据的相关性特征进行了分析,研究了基于时序与波动变化的特征抽取模式;在时变序列的差分及相对变化基础上,进一步综合利用偏态、离散、相关等特征构建了影响变量与响应变量之间的相关度计算模型。该模型可降低效用评价计算复杂度,增强评价的可靠性并可用于模型预测。