类构件潜在可重用性预测技术

首先分析了Judith Barnard类构件潜在可重用性预测模型,针对其不足提出了改进意见.在此基础上,将"类构件"的概念进行扩展,提出了"可重用类群"的概念及相应的判定算法,该技术使得对类构件潜在可重用性的预测更加合理、准确.      

帧差法和Mean shift算法融合的高速无人机目标跟踪

为实时跟踪高速飞行无人机,图像跟踪算法必须满足快速性和准确性要求。文章给出一个融合算法,将帧差法和 Mean shift算法的优势结合起来。2个算法平行运行,差帧法实现快速跟踪,Mean shift算法则用于对帧差法结果进行准确度修正。还利用 Kalman滤波技术对计算周期内的无人机运动位移进行补偿,进一步提高实时跟踪的准确性,并给出 Matlab仿真例子验证本文方法的有效性。     

快速可重组测控试验平台布线子系统的设计

在一种基于虚拟仪器的多功能快速可重组测控实验环境中,为了实现快速可重组思想,需要实现模块之间布线系统的设计。本文讨论了在Visual C＋＋开发平台下布线系统中线类的设计思想和实现过程,并对其中设计的主要函数和算法进行详细描述。     

激光冲击强化提高外物打伤钛合金模拟叶片高周疲劳性能

为指导钛合金叶片抗外物打伤激光冲击强化工艺设计,根据真实叶片叶型特征设计了刃口型模拟叶片,采用两种激光冲击强化工艺对模拟叶片进行预处理,并采用空气炮系统进行外物打伤模拟试验,最后通过疲劳试验和应力场预测进行疲劳性能影响规律及机理分析.试验结果表明:模拟叶片外物打伤后疲劳强度由518.45MPa降为290.72MPa,而...     

基于小样本数据增强的航天器表面损伤智能检测方法

在轨运行的航天器表面形成损伤有可能导致严重的后果,需要对航天器进行在轨实时损伤检测。针对航天器损伤检测图像样本难以获取的问题,本文采用智能化检测方法,提出了一种用于航天器表面损伤样本扩充的生成对抗网络,该网络能够学习单张输入图像的特征纹理表示,从而生成大量与输入图像特征相似的细粒度尺度样本,实现了少量图像数据样本的扩充。利用YOLO目标检测算法在扩充的图像样本中进行表面缺陷与损伤的检测识别,获取了较高的检测精度,为未来航天器健康状态监测与评估、通用化服务机器人应用及太空原位建设等提供了技术支撑。     

Reconciling space object observed and solar pressure albedo-areas via astrometric and photometric data fusion

There are many Resident Space Objects (RSOs) in the Geostationary Earth Orbit (GEO) regime, both operational and debris. The primary non-gravitational force acting on these RSOs is Solar Radiation Pressure (SRP), which is sensitive to the RSO’s area-to-mass ratio. Sparse observation data and mismodeling of non-gravitational forces has constrained the state of practice in tracking and characterizing RSOs. Accurate identification, characterization, tracking, and motion prediction of RSOs is a high priority research issue as it shall aid in assessing collision probabilities in the GEO regime, and orbital safety writ large. Previous work in characterizing RSOs has taken a preliminary step in exploiting fused astrometric and photometric data to estimate the RSO mass, shape, attitude, and size. This works, in theory, since angles data are sensitive to SRP albedo-area-to-mass ratio, and photometric data are sensitive to shape, attitude, and observed albedo-area. By fusing these two data types, mass and albedo-area both become observable parameters and can be estimated as independent quantities. However, previous work in mass and albedo-area estimation has not quantified and assessed the fundamental physical link between SRP albedo-area and observed albedo-area. The observed albedo-area is always a function of the SRP albedo-area along the line of sight of the observer. This is the physical relationship that this current research exploits. It is shown through simulation that due to this physical link, and through the fusion of astrometric and photometric data, it is possible to observe the mass of a space object when the area is not known. Results for data from 100 trajectories generated from randomly sampled initial conditions are shown. It is seen that even when the area of the object is not known, the uncertainty in mass can be lowered from an initial value of 800?kg to the range 500–700?kg for 72% of the samples, 200–500?kg for 13% of the samples, and 0–200?kg for 15% of the samples. It is further shown that although the uncertainties are large, the actual errors in mass are much lower, with the error RMS being less than 100?kg for 30% of the samples, between 100 and 200?kg for another 30%, and between 200 and 300?kg for 24% of the samples.     

Combining MHD and kinetic modelling of solar flares

Solar flares are explosive events in the solar corona, representing fast conversion of magnetic energy into thermal and kinetic energy, and hence radiation, due to magnetic reconnection. Modelling is essential for understanding and predicting these events. However, self-consistent modelling is extremely difficult due to the vast spatial and temporal scale separation between processes involving thermal plasma (normally considered using magnetohydrodynamic (MHD) approach) and non-thermal plasma (requiring a kinetic approach). In this mini-review we consider different approaches aimed at bridging the gap between fluid and kinetic modelling of solar flares. Two types of approaches are discussed: combined MHD/test-particle (MHDTP) models, which can be used for modelling the flaring corona with relatively small numbers of energetic particles, and hybrid fluid-kinetic methods, which can be used for modelling stronger events with higher numbers of energetic particles. Two specific examples are discussed in more detail: MHDTP models of magnetic reconnection and particle acceleration in kink-unstable twisted coronal loops, and a novel reduced-kinetic model of particle transport in converging magnetic fields.     

MCMC-Particle-based group tracking of space objects within Bayesian framework

With the intense increase in space objects, especially space debris, it is necessary to efficiently track and catalog the extensive dense clusters of space objects. As the main instrument for low earth orbit (LEO) space surveillance, ground-based radar system is usually limited by its resolution while tracking small space debris with high density. Thus, the obtained measurement information could have been seriously missed, which makes the traditional tracking method inefficient. To address this issue, we conceived the concept of group tracking. For group tracking, the overall tendency of the group objects is expected to be revealed, and the trajectories of individual objects are simultaneously reconstructed explicitly. According to model the interaction between the group center and individual trajectories using the Markov random field (MRF) within Bayesian framework, the objects’ number and individual trajectory can be estimated more accurately in the condition of high miss alarm probability. The Markov chain Monte Carlo (MCMC)-Particle algorithm was utilized for solving the Bayesian integral problem. Furthermore, we introduced the mechanism for describing the behaviors of groups merging and splitting, which can expand the single group tracking algorithm to track variable multiple groups. Finally, simulation of the group tracking of space objects was carried out to validate the efficiency of the proposed method.     

电子出版物的数据递归定义设计

】介绍了电子出版物递归数据结构设计。在分析了电子出版物中数据安排的实际情况后，深入探讨了数据递归定义的内在机理，揭示了各种安排的实质。     

自然写生与工业设计

本文论述了自然界中生命存在的基本规律，进而指出这些基本规律对提高设计者的创造力，从而设计出更为实用的、完美的工业产品是非常重要的。