光纤陀螺中分形噪声的参数估计和去除

光纤陀螺零漂信号中主要存在分形噪声和高斯白噪声,采用传统的时间序列分析法很难去除这类混合噪声.基于小波分析,提出一种分步估计加性白噪声强度和分形随机过程参数的新方法.先通过拟合自相关函数,估计出白噪声的强度;再在小波变换域估计分形噪声参数,并选取适当软阈值对测量值滤波.实验结果表明该估计和降噪方法有效地去除了光纤陀螺中的混合噪声,且不需要噪声的先验知识,具有很好的适应性.      

一种基于小波变换的多分形布朗运动合成算法

为了更有效地描述点状奇异性指数沿样本路径变化的信号,利用多分形的概念,在基于离散小波变换技术的基础上,提出了一种合成多分形布朗运动的新算法.该算法通过控制高斯白噪声的小波系数权来获得期望的信号局部正则性,而合成过程的收敛性由收敛因子保证.通过与基于Durbin-Levinson 和轮换矩阵嵌入技术算法的比较以及数字仿真试验,表明提出的算法不仅计算复杂度低,而且适用于生成非高斯的、自协方差函数事先未知的多分形过程.      

基于多重分形参数的高光谱数据特征提取

针对单一分形维数不能表征高光谱数据光谱局部吸收特征的问题,提出了基于光谱概率测度的多重分形参数特征提取方法.基于光谱信息度量进行光谱概率测度的计算,基于配分函数法估计得到尺度函数;通过对尺度函数求导计算出Holder指数,并对尺度函数勒让德Legendre变换计算出多重分形谱;从多重分形谱和Holder指数之间的函数关系提取表征多重分形谱形态的4个多重分形谱参数作为光谱特征参数;并应用于基于最小距离准则的航空推扫式高光谱成像仪(PHI,Prush-broom Hyperspectral Imager)图像监督分类.结果证明:利用基于光谱概率测度的多重分形参数特征提取方法提取的光谱特征参数进行分类得到的总体分类正确率达94.789%,分类精度明显高于利用信息量维数和多重分形谱特征提取方法进行分类的结果,证明了基于光谱概率测度的多重分形参数特征提取方法提取的多重分形参数的有效性和可靠性.     

小波变换在大地回波噪声处理中的应用

基于小波变换的多尺度分辨特性,作者对连续信号、脉冲信号和宽平稳白噪声在小波域中不同分解尺度上的模极大值变化规律进行分析,根据它们的变化规律用Mallat快速算法设计出一套用于抑制低信噪比雷达回波中的干扰噪声、最大限度提纯真实回波信号的算法.仿真实验证明,该算法对抑制噪声作用明显,能较好地提纯雷达回波信号,这为低信噪比下雷达回波信号检测提供了一条有效途径.     

光纤陀螺分形噪声的预白化滤波方法

光纤陀螺分形噪声往往具有长相关特性,导致陀螺输出信号慢漂移.为了滤除这类分形噪声,提出了一种分数阶预白化差分方法,通过该方法可以将分形噪声转化为高斯白噪声;利用小波变换,在小波变换域中采用Bayes软阈值去噪方法去除该白化噪声,从而形成了一种光纤陀螺分形噪声的预白化滤波方法.对VG949p型光纤陀螺实测数据处理结果表明,该方法能够有效地去除其中的分形噪声,抑制了陀螺低频漂移,同时也滤掉了其高频噪声项.      

基于多小波基多信源融合异常值剔除方法研究

提出了一种基于多小波基多信源融合和偏度分析相结合的异常值剔除方法。基于多个小波基的数据融合算法,先对各信源的数据进行多个不同小波基的多尺度分解,对相同小波基分解的信号在多尺度上加权融合,之后进行不同小波基的逆变换得到重构信号,将重构信号融合出的结果作为目标状态估值,对各信源的异常值进行检测,将偏度分析与检测门限相结合,实现了对异常值的剔除。实验结果表明,该方法准确、高效地实现了对各信源异常值的检测与剔除,提高了数据处理的精度,且易于工程实现。     

基于调频高斯小波变换的导航传感器故障诊断

为提高多传感器组合导航系统对各导航传感器的在线故障检测能力,在分析调频高斯小波特性的基础上,提出了一种基于调频高斯小波变换的导航传感器故障诊断算法.该算法仅利用传感器的观测量来直接诊断传感器工作是否正常.它可以对方差突变等软故障进行有效的在线检测.仿真结果证明了所提方法的有效性.     

基于EEMD分解的直驱式机电作动器故障诊断

基于集合经验模式分解 (EEMD,Ensemble Empirical Mode Decomposition)算法,给出一种机载直驱式双余度机电作动器(DDDR-EMA,Direct-Driven Dual-Redundancy Electro-Mechanical Actuator)复合故障诊断方法.EEMD对信号加入有限幅度的高斯白噪声,利用高斯白噪声频率均匀分布的统计特性使信号在不同尺度上保持连续性,解决了经验模式分解的模式混叠缺陷并保留了自适应性.将EEMD方法应用于机载DDDR-EMA故障诊断实验振动信号分析,先对实测信号进行分解,得到一组无模式混叠的固有模式函数;再采用不同的方法分析各频段,提取各频段包含的故障特征.实验结果表明:与经验模式分解相比EEMD能提高故障信号的分析精度,准确诊断机载DDDR-EMA的复合故障.     

基于小波域的分形图像压缩

在分析产生块效应原因基础上,定义了一种度量块效应的客观方法.为了消除传统分形图像压缩中产生的块效应,提出先将原图像变换到小波域,再进行分形压缩.这种方法既继承了传统分形图像压缩的编码原理,又利用了小波变换的优良特性,从而取得了更好的压缩效果.此外,根据小波系数的能量分布特征,提出了一种加快相似匹配的方法.     

二维分形海面电磁后向散射系数的分析与计算

根据改进的二维分形海面模型推导了一种在不同极化方式下(HH水平极化,VV垂直极化)新的电磁后向散射系数计算方法,对多个海浪波长区域内所有分辨单元后向散射系数进行平均统计.在时域内研究了海水介电常数、电磁波入射角、入射频率以及风速风向对后向散射系数的影响,运用盒维数方法计算了该散射系数的分形维数.最后通过数值分析方法证明了该散射系数实部与虚部曲线是分形的,并且与二维分形海面的分形维数相等,从而表明该海面的雷达回波信号也具有分形特性,为运用分形理论研究海面微弱目标检测提供了一个重要方法.     

一种基于支持向量机的小波图像压缩方法

提出了一种结合支持向量机(SVM,Support Vector Machines)回归与小波变换的新的静态图像压缩方法.SVM回归方法可以学习原始数据之间的相关性,并采用小部分训练样本,即支持向量来稀疏表示原始数据集,利用这一特性来逼近和约减小波系数,可以达到数据压缩的效果.首先采用小波变换把原始图像分解成不同尺度的多个子带,由于最低频子带系数非常重要,采用DPCM直接编码,然后对其它频带系数采用SVM回归进行压缩.由于不同尺度和方向的小波系数特征不同,为尽可能去除小波系数间的各种相关性,给出了适合SVM回归的小波系数的有效组织方式.最后研究了支持向量及其相应权重的混合编码方法.实验结果表明:与同类压缩方法相比,本算法获得的恢复图像的主客观质量有明显提高.      

Chaos and multifractals in the solar wind

By using the false-nearest-neighbours method, we have argued that the deterministic component of solar wind plasma dynamics should be low-dimensional. In fact, the results we have obtained using the method of topological embedding indicate that the behaviour of the solar wind can be approximately described by a low-dimensional chaotic attractor in the inertial manifold, which is a subspace of system phase space. We have also shown that the multifractal spectrum of the solar wind attractor is consistent with that for the multifractal measure of the self-similar generalized weighted Cantor set with two different scaling parameters and one probability measure parameter responsible for nonuniform compression in phase space and multifractality. The values of the parameters fitted also demonstrate that the complex solar wind system could only be weakly non-conservative (small dissipation) and quantify nonlinear dynamics; some parts of the attractor in phase space are visited much more frequently than other parts. In addition, to quantify the multifractality of space plasma intermittent turbulence, we consider that generalized Cantor set also in the context of scaling properties of solar wind turbulence. We investigate the resulting multifractal spectrum of a one-dimensional phenomenological model of turbulence cascade depending on its parameters, especially for asymmetric scaling. In particular, we have shown that intermittent pulses are stronger for the cascade model with two different scaling parameters. Even thought solar wind turbulence appears to be rather space filling, a better agreement with the data is obtained, especially for the negative index of generalized dimensions. Therefore we argue that there is a need to use a two-scale asymmetric cascade model. We hope that this generalized multifractal model will be a useful tool for analysis of intermittent turbulence in space plasmas. We thus believe that fractal analysis of chaotic systems could lead us to a deeper understanding of their nature, and maybe even to predict their seemingly unpredictable behaviour.     

基于视觉模型和图像特征的遥感图像压缩

遥感图像具有纹理复杂、边缘丰富的特点,通常难以实现高压缩比.统计了图像经过小波变换后在不同方向、不同级别子频带的能量分布.然后基于人类视觉系统(HVS)模型中的对比敏感度函数(CSF)和遥感图像的能量分布特征,提出一种不同频带小波系数的变步长量化方法,并把该方法应用于嵌入式小波图像编码器.实验结果表明,与不采用心理视觉量化的压缩方法相比,在相同压缩比下,峰值信噪比略为降低,但恢复图像的主观质量有一定程度的改善.     

一种不等差错保护的多用户随机接入方法

针对在随机接入中由于碰撞导致吞吐率低、信道拥塞时不能满足高优先级数据有效传输的问题,提出一种具有不等差错保护性能的多用户随机接入方法。该方法将数据包复制多份,根据选定的度分布在不同时隙中传输,将随机接入过程映射到二分图上;接收端利用连续干扰消除(SIC)技术,将碰撞的数据包恢复出来。在高链路负载下,从每帧中选取若干时隙,只允许高优先级用户接入,以提高对高优先级数据包的译码概率。使用"与或树"理论对提出的方法进行了理论分析,并进行了实验仿真验证。仿真结果表明:提出的多用户随机接入方法与传统ALOHA协议相比,在低链路负载下有效避免由于碰撞引起的吞吐率下降;与IRSA协议相比,高链路负载下有效提高高优先级用户的吞吐率。     

Statistical analysis of the total electron content observed at 23°S in the Brazilian sector

The vertical total electron content (VTEC) time series obtained at São José dos Campos (23.2°S, 45.9°W), Brazil, were statistically analyzed to study the low latitude ionosphere in the Brazilian sector during the year of 2006 (a period of low solar activity). Statistical analysis showed that Probability Density Functions (PDFs) and kurtosis have an intermittent behavior on small-scales (periods from minutes to one day) and presence of two functions on large-scales (periods from 3 to 30 days). The skewness result suggests the presence of some kind of waves due to the action of tropospheric sources (lower atmospheric origin). Results obtained by wavelet transform show strong oscillations with scale-sizes between 3 and 30 days, possibly associated with the planetary oscillations. According to these statistical and wavelet analyses we conjecture that there exist two important factors regarding the ionospheric effects: one factor is due to turbulent states found in small scales and the other factor consists in a more or less deterministic state provided by planetary waves (3–16 days or full solar rotation (27–28 days)). Further, these strong oscillations were also noted in multifractal analysis. We found a decrease of multifractality degree of the same scale-sizes.     

利用小波方差进行原子钟频率稳定度的估计

离散小波变换可以在不同尺度上分解时间序列,而不同尺度的波动性可用小波方差来表征。从小波方差的定义入手,系统地归纳了基于极大重叠离散小波变换(MODWT)的小波方差估计方法,及其等效自由度(EDF)的实用计算方法。最后利用一个实测算例进行计算分析,并与相应的重叠阿伦方差、重叠哈达玛方差进行比较,通过实验分析可以看出小波方差可有效消除原子钟信号非线性和非平稳性的影响,通过选择适当的小波基函数,如D4、D6小波,其方差可以像哈达玛方差一样,减少调频闪变噪声和调频随机游走噪声的泄露,适用于原子钟频率稳定度的表征。     

基于相关分析和组合神经网络的退化预测

产品剩余寿命预测是加速退化试验和故障预测与健康管理两大热点领域中的关键技术之一.为了解决复杂退化的预测问题,提出了一种新型预测方法,对退化轨迹能够实现较长距离的预测.此方法首先对复杂退化数据进行小波变换,通过Durbin-Watson方法和偏相关图分析各级分解序列的自相关性,最后根据序列的特点,组合BP(Back Propagation)和小波神经网络对退化轨迹进行预测.为了验证所提组合神经网络方法的有效性,采用小波神经网络的预测结果进行对比分析.实际退化数据的预测结果表明,所提方法比单独采用小波神经网络,具有更小的均方差(MSE,Mean Square Error),对剩余寿命(RUL,Remaining Useful Life)也具有更高的预测精度.     

Planetary radio astronomy from voyager

The Planetary Radio Astronomy instruments on Voyager 1 and 2 provided new, highly detailed measurements of several different kinds of strong, nonthermal radiation generated in the inner magnetospheres and upper ionospheres of Jupiter and Saturn. At Jupiter, an intense decameter-wavelength component (between a few tenths of a MHz and 39.5 MHz) is characterized by complex, highly organized structure in the frequency-time domain and by a strong dependence on the longitude of the observer and, in some cases, of Io. At frequencies below about 1 MHz there exists a (principally) kilometer-wavelength component of emission that is bursty, relatively broadbanded (typically covering 10 to 1000 kHz), and strongly modulated by planetary rotation. The properties of this component are consistent with a source confined to high latitudes on the dayside hemisphere of Jupiter. A second kilometric component is narrow-banded, relatively weak and exhibits a spectral peak near 100 kHz. The narrowband component also occurs periodically but at a repetition rate that is a few percent slower than that corresponding to the planetary rotation rate. This component is thought to originate at a frequency near the electron plasma frequency in the outer part of the Io plasma torus (8 to 10 R_J) and to reflect the small departures from perfect corotation experienced by plasma there.The Voyager instruments also detected intense, low frequency, radio emissions from the Saturn system. The Saturnian kilometric radiation is observed in a relatively narrow frequency band between 3 kHz and 1.2 MHz, is elliptically or circularly polarized, and is strongly modulated in intensity at Saturn's 10.66-hr rotation period. This emission is believed to be emitted in the right-hand extraordinary mode from regions near or in Saturn's dayside, polar, magnetospheric cusps. Variations in intensity at Saturn's rotation period may correspond to the rotation of a localized magnetic anomaly into the vicinity of the ionospheric footprint of the polar cusp. Variations in activity on time scales of a few days and longer seem to indicate that both the solar wind and the satellite Dione can also influence the generation of the radio emission.