基于MapReduce的CME参数识别模型并行计算技术

日冕物质抛射（Coronal Mass Ejection，CME）参数识别模型是太阳风预报过程的重要组成部分.在空间环境预报业务中，为提高太阳风预报的准确率，需要提高CME参数识别的精度.模型以计算任务串行的方式运行，运算效率低导致模型运算时间长，不能满足这种需求.CME参数识别模型的物理运算过程相互不独立，其在单节点上的运行方式不能满足并行化要求.基于MapReduce的并行计算框架，改进了CME参数识别模型的计算流程，提出CDMR（CME detection under MapReduce）方法，实现了CME参数识别模型的并行计算，并对比分析CME参数识别模型在串行计算和MapReduce并行计算下的运行时间，提高了模型的识别精度和计算效率.      

Real-time clock comparison and monitoring with multi-GNSS precise point positioning: GPS,GLONASS and Galileo

The precise point positioning (PPP) technique is widely used in time and frequency applications. Because of the real-time service (RTS) project of the International GNSS Service, we can use the PPP technique for real-time clock comparison and monitoring. As a participant in the RTS, the Centre National d’Etudes Spatiales (CNES) implements the PPPWIZARD (Precise Point Positioning with Integer and Zero-difference Ambiguity Resolution Demonstrator) project to validate carrier phase ambiguity resolution. Unlike the Integer-PPP (IPPP) of the CNES, fixing ambiguities in the post-processing mode, the PPPWIZARD operates in the real-time mode, which is also called real-time IPPP (RT-IPPP). This paper focuses on applying the RT-IPPP for real-time clock comparison and monitoring. We review the principle of real-time clock comparison and monitoring, and introduce the methodology of the RT-IPPP technique. The observations of GPS, GLONASS and Galileo were processed for the experiments. Five processing modes were provided in the experiment to analyze the benefits of ambiguity resolution and multi-GNSS. In the clock comparison experiment, the average reduction ratios of standard deviations with respect to the G PPP mode range from 9.7% to 35.0%. In the clock monitoring experiment, G PPP mode can detect clock jumps whose magnitudes are larger than 0.9 ns. The RT-IPPP technique with GRE PPP AR (G) mode allows for the detection of any clock jumps larger than 0.6 ns. For frequency monitoring, G PPP mode allows detection of frequency changes larger than 1.1 × 10⁻¹⁴. When the RT-IPPP technique is applied, monitoring with GRE PPP AR (G) mode can detect frequency changes larger than 6.1 × 10⁻¹⁵.     

Characterizing space debris longitude-dependent distribution based on RAAN perturbation rate

The space debris environment is one of the major threats against payloads. Space debris orbital distribution is of great importance for space debris environment modeling. Due to perturbation factors, the Right Ascension of Ascending Node (RAAN) of space objects changes consistently, causing regular rotation of the orbit plane around Earth’s axis. Based on the investigation of the RAAN perturbation rate of concerned objects, this paper proposes a RAAN discretization method in order to present the space debris longitude-dependent distribution. Combined with two line element (TLE) data provided by the US Space Surveillance Network, the estimated value from RAAN discretization method is compared with the real case. The results suggest that using only the initial orbital data at the beginning of the time interval of interest, the RAAN discretization method is able to provide reliable longitude distribution of concerned targets in the next following period. Furthermore, spacecraft cumulative flux against space debris is calculated in this paper. The results suggest that the relevance between spacecraft RAAN setup and flux output is much smaller for LEO targets than MEO targets, which corresponds with the theory analysis. Since the nonspherical perturbation is the major factor for RAAN variation, the RAAN perturbation rate has little connection with the size of orbital objects. In other words, the RAAN discretization method introduced in this paper also applies to space debris of different size range, proposing a possible suggestion for the improvement of space debris environment engineering models.     

Do RadioAstron detections correlate with flaring states? An initial study of seven southern AGN

We examine the state of seven southern radio sources at the time of their RadioAstron AGN Survey observations. Both ATCA flux density monitoring data and Fermi light-curves are considered in determining the relative activity of the source. A simple hypothesis, that sufficiently compact source structure exists for detections on RadioAstron baselines when the source is in a flaring state, is qualitatively tested. We find four instances of RadioAstron detections during flaring radio states and four instances of RadioAstron non-detections during fading or quiescent radio states, in support of the hypothesis. However, we also find three instances of RadioAstron detections during quiescent or fading radio states, and two non-detections during a flaring state, indicating that the situation is (not unexpectedly) more complex. Radio and gamma-ray monitoring such as that described here, together with the full RadioAstron AGN Survey results, will allow a more thorough investigation of the dependencies of detections on baselines of >10 Earth diameters.     

基于类脑模型与深度神经网络的目标检测与跟踪技术研究

目标检测与跟踪技术广泛应用于交通、医疗、安保和航天等领域.目前，目标检测与跟踪技术面临目标微弱、背景复杂、目标被遮挡等挑战.同时，随着脑科学研究的不断深入，人们对人脑视觉系统的理解逐渐透彻，利用类脑计算解决复杂背景下高精度目标检测与跟踪问题成为相关领域的重要研究方向.本文结合神经工程导向的类脑模型和计算机工程导向的深度神经网络(Deep Neural Networks, DNNs)，提出多种基于类脑模型与深度神经网络的目标检测与跟踪算法，包括：基于演算侧抑制的目标检测算法，基于结构 对比度(Structure Contrast, SC)视觉注意模型的弱小目标检测算法和基于记忆机制与分层卷积特征的目标跟踪算法.实验结果表明，将类脑模型和深度神经网络应用于目标检测和跟踪领域，有利于实现复杂条件下的高精度目标检测和鲁棒性目标跟踪.     

大跨时空任务背景下的太阳能无人机任务规划技术研究进展

通过任务规划技术合理的优化太阳能无人机的飞行轨迹和动力学参数，能够有效提高太阳能无人机的能量利用率，使其胜任许多大范围跨时间跨空间飞行任务。从能量建模、续航评估和能量管理策略3个方面对大跨时空任务背景下太阳能无人机任务规划技术的研究进展进行了综述。在能量建模方面，介绍了当前主流的太阳辐射模型和能量生产基本框架；在续航评估方面，分析了目前的指标设计和应用方法；在能量管理策略方面，从能量综合应用、风力滑翔机制、轨迹优化方法和面向特定任务的应用4个角度，梳理了当前的研究现状。最后，对该领域未来可能的研究方向进行了展望。     

Effect of curing condition on bonding quality of scarf-repaired composite laminates

Both single-face vacuum bag curing (SVC) and double-face vacuum bag curing (DVC) can be used in scarf repair of composite structures. But different curing conditions caused by the sealing state may affect the bonding quality of scarf-repaired structures. In this paper, the effect of curing condition on bonding quality of scarf-repaired laminates was experimentally investigated in terms of surface profiles, moisture absorption curves and section profiles. In order to further explore the moisture absorption mechanism, finite element model of the repaired laminates using DVC was established with moisture diffusion of both the adhesive and composite laminates considered. This model was verified by experimental results. Based on the model of DVC case, the model of SVC case was built by changing moisture absorption parameters of the adhesive. Results show that SVC reduces the bonding quality, mainly reflecting in more adhesive inner voids and patch-to-parent dislocation. And SVC increases moisture absorption rate and moisture equilibrium content of the adhesive, and its effect on the former is far greater than that on the latter.     

智能方法在惯性系统误差参数辨识中的应用

总结了目前我国提高惯性系统导航精度的技术途径,阐述了国内外惯性系统误差参数辨识方法的研究现状,介绍了当前滤波算法、智能优化算法和人工神经网络方法在惯性系统导航领域的应用情况以及存在的不足。最后,分析了空间飞行器惯性系统误差参数辨识技术的未来研究方向,即智能优化算法和人工神经网络方法等智能方法将在惯性系统误差参数辨识中发挥越来越重要的作用,通过将误差系数标定问题转换为参数辨识问题,采用智能方法在庞大的解空间内实现对惯性系统误差参数的快速辨识。     

再入飞行器自适应最优姿态控制

针对再入飞行器姿态控制问题，应用自适应动态规划（ADP）理论设计了姿态控制器。将再入飞行器的姿态控制建模为非线性系统的最优控制问题，提出单网络积分型强化学习（SNIRL）算法进行求解，该算法简化了积分型强化学习（IRL）算法在迭代计算中的执行-评价双网络结构，只需要采用评价网络估计值函数就可以求得最优控制律，其收敛性得到了理论证明。基于SNIRL算法设计了自适应最优控制器，并证明了闭环系统的稳定性。通过数值仿真校验了SNIRL算法比IRL算法计算效率更高，收敛速度更快，并校验了自适应最优姿态控制器的有效性 。