LSTM神经网络在太阳F10.7射电流量中期预报中的应用

F_10.7指数作为大气密度经验模型的重要输入参量,其预报精度直接影响航天器轨道预报精度.研究发现,太阳活动表现出长时间尺度上平均11年和中短时间尺度平均27天的周期性变化特征.依据这一观测事实,基于长短期记忆单元（Long Short-term Memory,LSTM）递归神经网络方法进行F_10.7指数未来27天的中期预报.利用一个连续长时段F_10.7数据作为训练数据,构建LSTM神经网络训练和预测模型,分别预测太阳活动高低年未来27天的F_10.7指数.结果表明,太阳活动高年的第27天F_10.7指数预报平均相对误差最优可达10%以内,低年最优可达2%以内.      

太阳10.7 cm射电流量中期预报模型研究(Ⅱ)

太阳活动指数中期预报一直是空间环境业务预报的难点之一.本文在自回归方法模型的基础上,利用太阳活动区面积、位置等参数与10.7cm辐射流量之间的定量关系,根据活动区面积衰减规律,建立了一个基于活动区参数及演化规律的改进型太阳活动指数中期预报模型.通过对预报测试实例分析发现,在日面出现较大活动区导致F_(10.7)迅速增长并超过历史数据峰值的情况,在日面活动区消亡导致指数突然出现平静期的情况,新模型的预报准确性相比自回归模型有很大提高,预报的平均相对误差下降约5%～9%.由此可见,新模型在某些特定条件下提高了原有模型的精度.该研究为提高业务型太阳10.7cm射电流量中期预报模型的预报精度奠定了基础.     

利用F_(10.7)短期预报电离层F_0F_2

通过对电离层历史数据和太阳射电流量F10.7的回归分析,提出了一种单站电离层f0F2的短期预报方法,以F10.7的流动平均值fc为输入,以未米3天的f0F2为输出,分别利用中国地区8个台站的数据进行检验,分析不同太阳活动水平、季节以及地方时预报误差的分布特征.结果表明,该方法能有效地预测未来1～3天的f0F2.该方法还可应用于其他电离层参量的短期预报.     

太阳10.7 cm射电流量中期预报模型研究(I)

采用时间序列模型中的自回归方法开展了F10.7中期预报研究. 预报试验和误差分析结果表明, 在太阳活动水平较低、F10.7 的27天周期性明显时自回归预报方法的预报精度高, 具有较为理想的预报效果, 但在日面有大活动区产生和消亡时预报效果不理想. 这说明时间序列模型中的自回归方法能够较好地反映太阳F10.7的27天周期性特征, 对F10.7中期预报模型的建立有一定适用性.通过对2005年9月21日至2007年6月7日期间预报结果的比较可以看出, 自回归分析方法预报的精度与美国空军预报的相当.     

奇异谱分析在太阳10.7cm射电流量中期预测中的应用

首次尝试利用信号处理技术奇异谱分析方法预测太阳活动低年未来27天太阳10.7cm射电流量.选取的预报试验时间段是2004年4月30日至5月30日,此试验期内太阳活动水平相对较低.在样本时间序列的构建上,吸取了相似周数据分析思路,采取的是23周实时观测数据与其相似周第20周下降年部分数据相结合的方式,既增加了样本长度又避开了太阳活动的活跃期.这31天的预报试验结果表明,大部分情况下,预报值基本上体现出F_10.7的变化趋势,平均相对误差为10.5%;比同时期美国空军预测值的平均相对误差小,前者为11.3%,后者为14.6%;除两天外,SSA每一次27天的预报结果的平均相对误差比美国空军(AAF)的要小;对不同的时间提前量而言,AAF提前1天到提前12天的预报准确性较奇异谱分析方法要高,即AAF较短期的预报效果更好.     

基于BP神经网络的飞行动态实时预测方法

针对传统Back-Propagation网络(简称BP网络)收敛速度慢、网络灵敏度过高和隐含层数难以确定等缺陷,提出一个改进型BP网络,提高了网络预测的实时性和精确性;然后将之应用到飞行动态预测问题上,充分发挥网络模型的学习、记忆和动态自适应性的优势,力图解决飞行器轨迹的描述和预测问题.      

利用EMD方法提取太阳活动周期成分

EMD(经验模态分解)方法在处理非线性及非平稳时间序列时表现出了很大的优势和应用潜力.利用EMD方法研究太阳活动周期,对110年(1894-2003)和55年(1949-2003)的太阳黑子数月均值进行分解,分别得到一系列模式和一个趋势项,其中都可能包含有1.3至1.4年周期分量,25至30个月QBO(准双年振荡)分量,11年太阳周分量和22年Hale周分量.其中11年周期分量幅度最大,变化特征与太阳黑子数原始数据具有很高的相似性.不同于传统方法,EMD方法给出了太阳活动在不同时间尺度上各自分离的变化特征.     

Using TensorFlow-based Neural Network to estimate GNSS single frequency ionospheric delay (IONONet)

In the last 20?years, and in particular in the last decade, the availability of propagation data for GNSS has increased substantially. In this sense, the ionosphere has been sounded with a large number of receivers that provide an enormous amount of ionospheric data. Moreover, the maturity of the models has also been increased in the same period of time. As an example, IGS has ionospheric maps from GNSS data back to 1998, which would allow for the correlation of these data with other quantities relevant for the user and space weather (such as Solar Flux and Kp). These large datasets would account for almost half a billion points to be analyzed. With the advent and explosion of Big Data algorithms to analyze large databases and find correlations with different kinds of data, and the availability of open source code libraries (for example, the TensorFlow libraries from Google that are used in this paper), the possibility of merging these two worlds has been widely opened. In this paper, a proof of concept for a single frequency correction algorithm based in GNSS GIM vTEC and Fully Connected Neural Networks is provided. Different Neural Network architectures have been tested, including shallow (one hidden layer) and deep (up to five hidden layers) Neural Network models. The error in training data of such models ranges from 50% to 1% depending on the architecture used. Moreover, it is shown that by adjusting a Neural Network with data from 2005 to 2009 but tested with data from 2016 to 2017, Neural Network models could be suitable for the forecast of vTEC for single frequency users. The results indicate that this kind of model can be used in combination with the Galileo Signal-in-Space (SiS) NeQuick G parameters. This combination provides a broadcast model with equivalent performances to NeQuick G and better than GPS ICA for the years 2016 and 2017, showing a 3D position Root Mean Squared (RMS) error of approximately 2?m.     

神经网络与D-S理论结合的HRR识别研究

基于D-S(Dempster-Shafer)证据理论,比较和研究了相关数据和不相关数据的融合方法,分析了多传感器数据融合的算法:集中式融合算法和分布式融合算法.经过实验证明,执行分布式有反馈融合算法时的效果最好.然后利用该算法,提出了和线性内插神经网络相结合的识别方法.利用4种飞机的步进频率雷达的高分辨率一维距离像,将神经网络的识别结果作为证据分别送入传感器进行融合,进行识别研究.实验证明,与单纯利用神经网络的方法比较,目标的正确识别率得到了改善.     

2007-2009谷年期间高热层大气密度的变化

在第23至第24太阳活动周的峰年之间,太阳活动谷年具有持续时间长,极低F_10.7太阳辐射通量（低至65）和超长期的零太阳黑子数记录等特点,因此是观测和研究在这种特殊背景下热层大气变化的极好机会.尤其是能充分理解和掌握在宁静环境下热层大气密度对弱太阳活动和小地磁扰动的响应特性.本文利用高度650 km以上星载大气密度探测器2007—2009年的连续探测数据进行分析,结果表明,在太阳辐射通量F_10.7极低值期间,较高热层大气密度对F_10.7的起伏具有更显著的响应变化.当F_10.7由70降至65时,日均大气密度会有4～5倍的显著降变,远大于通常大气模式中的降变值.同样在F_10.7极低值期间,较高热层大气密度对小地磁扰动也具有显著的响应增变,当日K_p指数之和由23增至30时,较高热层大气密度则会有80%～160%的强增变.     

A laboratory study on the thermally induced transformation of hydrogen cyanide (HCN) to the cyanogen anion (CN) in Solar System analog ices

Mixtures of molecular nitrogen and methane have been identified in numerous outer Solar Systemices including the icy surfaces of Pluto and Triton. We have simulated the interaction of ionizing radiation in the Solar System by carrying out a radiolysis experiment on a methane – molecular nitrogen ice mixture with energetic electrons. We have identified the hydrogen cyanide molecule as the most prominent carbon–nitrogen-bearing reaction product formed. Upon warming the irradiated sample, we followed for the first time the kinetics and temporal evolution of the underlying acid–base chemistry which resulted in the formation of the cyanide ion from hydrogen cyanide. On the surfaces of Triton and Pluto and on comets in Oort’s cloud this sort of complex chemistry is likely to occur. In particular, hydrogen cyanide can be produced in low temperature environments (Oort cloud comets) and may be converted into cyanide ions once the comets reach the warmer regions of the Solar System.     

瞬态平面热源法热物理性能测量准确度和适用范围的标定——常温下标准Pyroceram 9606材料热物理性能测量

瞬态平面热源法作为一种非稳态热物理性能测试技术,其测量范围、测量准确度和试验参数的确定是正确评价和应用这种测试技术的前提条件。详细介绍了瞬态平面热源法的测量原理以及常温下的测量装置,描述了采用Pyroceram 9606热物理性能标准材料研究瞬态平面热源法在实际测试中各种试验参数对测量准确度的影响,并由此考察这种测试方法和测试装置的测量准确度。     

Using vegetation indices and texture measures to estimate vegetation fractional coverage (VFC) of planted and natural forests in Nanjing city,China

Vegetation fractional coverage (VFC) is an important vegetation parameter affecting exchanges of carbon, water, energy between the atmosphere and surface. In this study, the applicability of tonal and texture measures calculated using an IKONOS_2 image in retrieving VFC of forests was investigated in the urban area of Nanjing city, China. Four spectral vegetation indices (VI) and six texture measures (TEX) were related to VFCs acquired from in situ measurements. Models for estimating VFC based on VIs or/and TEXs were established and validated for planted low broad-leaf forest plots (PLB), planted mature forest plots (PMF), natural broad-leaf forest plots (NBF), and all forest plots (ALLv), respectively. The results show that high spatial resolution remote sensing data is applicable to estimate VFC in urban areas, and TEXs may act as effective supplements of vegetation indices (VIs) for the retrieval of VFC. VIs are suitable for VFC estimation of mature forests (such as NBF and PMF) with high vegetation density, and TEXs can yield a more accurate estimate for planted forests (such as PLB and PMF) with regular spatial distribution if they are calculated with proper parameters, such as window size. The combination of VIs and TEXs improve the estimation of VFC if forest types are not previously differentiated. The results can be used as a reference for determining effective spectral or texture parameters in VFC estimation under similar environmental conditions according to vegetation maturity and regularity.