基于轨迹聚类的天光光谱特征分析

天光背景扣除是LAMOST 1D光谱数据处理中重要的环节,其扣除好坏直接影响光谱产品质量,因此构造理想的超级天光光谱模型具有重要的意义。通常超级天光是由与目标天体同时观测的天光光纤光谱构造而成,同一区域的天光背景可能随着不同的观测时刻有着规律性的变化特征（如月相变化）,如果能充分分析并利用这些特征,可有效校正超级天光模型,从而提高减天光效果。轨迹聚类方法是一种分析目标随时、空变化特征的有效工具,针对LAMOST天光光谱中可能存在的变化规律,给出一种基于轨迹聚类的天光光谱特征分析方法。主要分以下三部分：首先是天光光谱的时序化描述。LAMOST pipeline采用且提供了每个观测天体的即时超级天光光谱,为了获取特定天区背景天光的光变特征,需选择天光光纤光谱以及扣除目标天体光谱的背景光谱,以5°视场（LAMOST望远镜视场）为单位,按观测日期MJD均匀分组,从而对特定区域的天光光谱进行了时序化表征;其次给出基于密度的天光光谱数据聚类算法STK-means。为解决随机参数导致收敛及聚类效果不理想的问题,在分析天光光谱时序数据特征的基础上,给出基于密度的相似性度量公式,并作为传统k-means聚类的初始参数选择依据,从而给出基于密度的天光光谱数据聚类算法STK-means;最后进行实验分析。实验验证了该方法的正确性和有效性以及不同初始参数K值的选择对聚类结果的影响。在此基础上,利用STK-means聚类方法,对LAMOST第一期巡天中一个完备小天区的天光光谱时序数据进行了轨迹特征分析,结果表明,除个别光谱质量较差或常说异常外,该特定区域的天光背景以农历每月十五、十六为中心向两边呈对称分布,反映了该区域观测过程中受月相的影响变化情况,该特征经量化后可为校正超级天光模型提供一种有效途径。同时,由于时序化描述过程中均匀采样的要求,该方法可适用于反银心、盘、晕等高天体数密度区域,而对于高银纬低数密度区域则需要更长时间的巡天观测。此外,该方法还可有效发现特定区域的离群(异常)天光光谱,为天文学家进一步分析提供珍稀样本。     

Segmentation and classification of time series using ordinal pattern distributions

The distribution of ordinal patterns in time series has been found to reflect important qualitative features of the underlying system dynamics. Abrupt changes in the dynamics typically result in clearly visible differences between the distributions before and after the break. Recurring dynamical regimes can be discovered by classifying the distributions in different parts of the time series. This paper discusses two algorithms which exploit the relation between ordinal pattern distributions and system dynamics for the segmentation and classification of time series. The first algorithm employs a kernel-based statistic, the Maximum Mean Discrepancy of ordinal pattern distributions, to detect and locate change points in the time series. The second algorithm uses clustering of the ordinal pattern distributions to classify time series segments with similar dynamics. The methodology is applied to various real-life time series from physiology and economics.     

Clustering of financial time series

This paper addresses the topic of classifying financial time series in a fuzzy framework proposing two fuzzy clustering models both based on GARCH models. In general clustering of financial time series, due to their peculiar features, needs the definition of suitable distance measures. At this aim, the first fuzzy clustering model exploits the autoregressive representation of GARCH models and employs, in the framework of a partitioning around medoids algorithm, the classical autoregressive metric. The second fuzzy clustering model, also based on partitioning around medoids algorithm, uses the Caiado distance, a Mahalanobis-like distance, based on estimated GARCH parameters and covariances that takes into account the information about the volatility structure of time series. In order to illustrate the merits of the proposed fuzzy approaches an application to the problem of classifying 29 time series of Euro exchange rates against international currencies is presented and discussed, also comparing the fuzzy models with their crisp version.     

Grafting of higher-order correlations of real financial markets into herding models

In this work, we graft the volatility clustering observed in empirical financial time series into the Equiluz and Zimmermann (EZ) model, which was introduced to reproduce the herding behaviors of a financial time series. The original EZ model failed to reproduce the empirically observed power-law exponents of real financial data. The EZ model ordinarily produces a more fat-tailed distribution compared to real data, and a long-range correlation of absolute returns that underlie the volatility clustering. As it is not appropriate to capture the empirically observed correlations in a modified EZ model, we apply a sorting method to incorporate the nonlinear correlation structure of a real financial time series into the generated returns. By doing so, we observe that the slow convergence of distribution of returns is well established for returns generated from the EZ model and its modified version. It is also found that the modified EZ model leads to a less fat-tailed distribution.     

基于选择性支持向量机集成的混沌时间序列预测

提出了一种基于聚类的选择性支持向量机集成预测模型.为提高支持向量机集成的泛化能力，采用自组织映射和K均值聚类算法结合的聚类组合算法，从每簇中选择出精度最高的子支持向量机进行集成，可以保证子支持向量机有较高精度并提高了子支持向量机之间的差异度.该方法能以较小的代价显著提高支持向量机集成的泛化能力.采用该方法对Mackey-Glass混沌时间序列和Lorenz系统生成的混沌时间序列进行预测实验，结果表明可以对混沌时间序列进行准确预测，验证了该方法的有效性. 关键词： 支持向量机 集成 混沌时间序列 聚类     

基于大数据分析的移动对象轨迹预测方法

对移动对象的轨迹预测将在移动目标跟踪识别中具有较好的应用价值。移动对象轨迹预测的基础是移动目标运动参量的采集和估计,移动目标的运动参量信息特征规模较大,传统的单分量时间序列分析方法难以实现准确的参量估计和轨迹预测。提出一种基于大数据多传感信息融合跟踪的移动对象轨迹预测算法。首先进行移动目标对象进行轨迹跟踪的控制对象描述和约束参量分析,对轨迹预测的大规模运动参量信息进行信息融合和自正整定性控制,通过大数据分析方法实现对移动对象运动参量的准确估计和检测,由此指导移动对象轨迹的准确预测,提高预测精度。仿真结果表明,采用该算法进行移动对象的运动参量估计和轨迹预测的精度较高,自适应性能较强,稳健性较好,相关的指标性能优于传统方法。     

Fault Recognition of Rolling Bearings Based on Parameter Optimized Multi-Scale Permutation Entropy and Gath-Geva

To extract fault features of rolling bearing vibration signals precisely, a fault diagnosis method based on parameter optimized multi-scale permutation entropy (MPE) and Gath-Geva (GG) clustering is proposed. The method can select the important parameters of MPE method adaptively, overcome the disadvantages of fixed MPE parameters and greatly improve the accuracy of fault identification. Firstly, aiming at the problem of parameter determination and considering the interaction among parameters comprehensively of MPE, taking skewness of MPE as fitness function, the time series length and embedding dimension were optimized respectively by particle swarm optimization (PSO) algorithm. Then the fault features of rolling bearing were extracted by parameter optimized MPE and the standard clustering centers is obtained with GG clustering. Finally, the samples are clustered with the Euclid nearness degree to obtain recognition rate. The validity of the parameter optimization is proved by calculating the partition coefficient and average fuzzy entropy. Compared with unoptimized MPE, the propose method has a higher fault recognition rate.     

基于有限穿越水平可视图的短时睡眠心率变异性研究

心率数据是最易于获取的人体生理数据之一,基于心率变异性的睡眠分析是近年来各种用于日常健康管理的可穿戴设备功能的一个重要发展方向,需要不断探索可以应用于标准睡眠分期时间窗(约30 s)的各类短时特征参数.利用近期报道的有限穿越水平可视图,并进一步提出一种加权有限穿越水平可视图,将不同睡眠状态下的短时心率变异序列映射为网络,进而提取平均集聚系数、特征路径长度、集聚系数熵、路径分布熵、加权集聚系数熵和加权路径分布熵等网络特征参数进行统计分析.结果表明,各网络参数值在醒觉、浅睡期、深睡期和快速眼动期的幅度水平具有显著差异,体现了所述方法在基于短时心率变异数据的睡眠分期中的有效性.同时,进一步研究了健康年轻人和中老年人在不同睡眠状态下的网络参数值,发现两者虽然存在整体的水平差异,但是在不同睡眠状态间的变化仍具有相同的趋势,反映出相对于正常的年龄老化,睡眠调制对心脏动力学系统具有更显著的影响,也说明所述方法可作为基于心率变异性的睡眠研究的一种新的辅助工具.     

一种公共网络攻击数据挖掘智能算法研究

公共网络的开放性和自组织特性导致网络容易受到病毒干扰和入侵攻击,对攻击数据的准确高效挖掘能确保网络安全。传统方法采用时频指向性波束特征聚类方法实现攻击数据挖掘,在信噪比较低时攻击数据准确挖掘概率较低。提出一种基于自适应滤波检测和时频特征提取的公共网络攻击数据挖掘智能算法。首先进行公共网络攻击数据的信号拟合和时间序列分析,对含噪的攻击数据拟合信号进行自适应滤波检测,提高信号纯度,对滤波输出数据进行时频特征提取,实现攻击数据的准确挖掘。仿真结果表明,采用该算法进行网络攻击数据挖掘,对攻击数据特征的准确检测性能较高,对干扰的抑制性能较强,能有效实现网络安全防御。     

基于多路可视图的健康与心梗患者心电图信号复杂网络识别

可视图(visibility graph, VG)算法已被证明是将时间序列转换为复杂网络的简单且高效的方法,其构成的复杂网络在拓扑结构中继承了原始时间序列的动力学特性.目前,单维时间序列的可视图分析已趋于成熟,但应用于复杂系统时,单变量往往无法描述系统的全局特征.本文提出一种新的多元时间序列分析方法,将心梗和健康人的12导联心电图(electrocardiograph, ECG)信号转换为多路可视图,以每个导联为一个节点,两个导联构成可视图的层间互信息为连边权重,将其映射到复杂网络.由于不同人群的全连通网络表现为完全相同的拓扑结构,无法唯一表征不同个体的动力学特征,根据层间互信息大小重构网络,提取权重度和加权聚类系数,实现对不同人群12导联ECG信号的识别.为判断序列长度对识别效果的影响,引入多尺度权重度分布熵.由于健康受试者拥有更高的平均权重度和平均加权聚类系数,其映射网络表现为更加规则的结构、更高的复杂性和连接性,可以与心梗患者进行区分,两个参数的识别准确率均达到93.3%.     

基于灰度聚类算法的红外图像增强研究

根据红外灰度图像的特点，提出了一种基于K-均值聚类的图像增强的新算法。该算法首先根据具体图像确定K值，其次对红外图像的辐射温度数据进行统计学习，把不同温度值按升序排列，然后按等差原则选取温度值作为初始聚类中心，再依据初始聚类中心采用K-均值聚类算法对温度进行聚类，最后由聚类结果对图像进行自适应增强。通过对红外灰度图像进行实验，得到了满意的结果: 对比直方图均衡，具有更丰富的图像细节信息和层次感，视觉效果更好。     

一种基于模糊C均值聚类小数据量计算最大Lyapunov指数的新方法

在小数据量计算最大Lyapunov指数的过程中,为了减少人为因素识别线性区域带来的误差,提出一种基于模糊C均值聚类的新方法.该方法根据平均发散程度指数曲线的变化特征,利用分类算法进行识别.首先,利用小数据量算法对混沌时间序列进行计算得到平均发散程度指数集合;其次,利用模糊C均值聚类算法对平均发散程度指数集合进行分类,得到不饱和数据;然后,对不饱和的二阶差分数据进行分类,得到零附近波动数据并剔除粗大误差,再对保留的有效数据利用统计方法识别出线性区域;最后,对线性区域进行最小二乘法拟合得到最大Lyapunov指数.为了验证该算法的有效性,对著名Logistic和Hénon混沌系统进行了仿真,所得结果接近理论值.实验表明,所提出的新方法与主观识别方法比较,计算结果更加准确.     

离散时间序列的网络模体分析

时间序列可以被转换成网络的形式,复杂网络理论也因此可以用于刻画时间序列的时域和相空间特性.本文针对可视图算法和相空间重构算法这两种时间序列的转换算法,研究了它们的伴生网络在倍周期分岔和混沌等各种类型时间序列的模体分布特征,分析了这两种算法各自的优点.     

考虑柔性关节的机械臂自适应运动控制策略研究

具有柔性关节的轻型机械臂因其自重轻、响应迅速、操作灵活等优点,取得了广泛应用。针对具有柔性关节的机械臂系统的关节空间轨迹跟踪控制系统动力学参数不精确的问题,提出一种结合滑膜变结构设计的自适应控制器算法。通过自适应控制的思想对系统动力学参数进行在线辨识,并采用Lyapunov方法证明了闭环系统的稳定性。仿真结果表明,该控制策略保证了机械臂系统对期望轨迹的快速跟踪,具有良好的跟踪精度,系统具有稳定性。     

基于局部线性嵌入的特征融合方法在岩石破裂 状态分类的应用*

为了获取岩石破裂过程有效的声发射信号特征,更好的对岩石破裂状态进行分类,提出一种基于流形学习算法的LLE特征融合方法进行数据降维。以红砂岩为研究对象设计室内单轴压缩实验采集信号,然后对原始声发射信号预处理并对信号进行特征提取,以时域、频域下的特征向量重新组合成一组新的多维特征向量,采用线性主元(PCA)和流形学习LLE算法分别进行降维。比较两种算法降维后融合特征的聚类效果二维和三维分布图,使用LLE算法降维后,四种状态分布相对更近,呈一条水平线趋势,且各状态交叉混叠数目较少,第一状态没有一个样本错判,且四个状态相比于PCA降维后的聚类效果更集中。再比较两种算法降维后融合特征的敏感度之和,LLE算法融合特征敏感度之和远大于PCA算法,说明经过LLE算法降维后得到的融合特征更多地表征了原始信号包含的局部信息同时证明了LLE算法相比PCA算法具有更好的聚类效果。最后经LLE特征融合下的砂岩破裂状态分类实验验证,融合特征后的识别率相对单一的时域特征识别提高了6%。表明该方法能显著提高岩石破裂状态分类的识别率,降维性能相对突出。     

Algorithm for solving the inverse problem of ballistics in the case of an asymmetric object

An algorithm to identify the aerodynamic characteristics of an asymmetric object from its trajectory data obtained in a ballistic experiment is developed based on the technique for estimating the nonlinear system’s parameters. Using the method of successive approximations, the coefficients of the aerodynamic function polynomial representation are found that best describe measuring data. The essence of the algorithm is the solution of the direct problem of the symmetric object’s dynamics using the complete set of Euler dynamic equations. The variation of the desired parameters is statistically estimated during calculations. The algorithm allows for jointly processing data of a series of experiments with similar models. Thereby, the volume of processed data is augmented and the final result becomes more accurate.     

基于K-均值聚类与夹角余弦法的多光谱分类算法

近年来对高光谱与多光谱进行分类去混的研究方法很多,K-均值聚类算法与光谱相似度计算算法都属于成熟的分类算法.作者在对其研究基础上,将K-均值算法进行改进,并融入光谱相似度匹配算法,形成一种新的光谱分类算法,找出两条距离最远的光谱作为参考光谱,用欧氏距离法或夹角余弦法对数据立方体进行分类,并且从数据立方体中删除属于这两条谱线的其余谱线,同时找出与两条参考光谱距离最远或者夹角最大者作为第三条参考光谱,对剩余数据立方体进行新的分类,并在此算法上用多光谱数据立方体进行了试验验证.通过ENVI用K-均值(K-means)进行分类,与改进的K-means算法和夹角余弦法Matlab仿真结果进行比较,后两种对于两种气泡的分类效果都很好,对背景的分类改进的K-means算法效果较好,尤其是欧氏距离法能将背景完整地分离出来.     

Application of permutation entropy method in the analysis of chaotic,noisy, and chaotic noisy series

We have considered a permutation entropy method for analyzing chaotic, noisy, and chaotic noisy series. We have introduced the concept of permutation entropy from a survey of some features of information entropy (Shannon entropy), described the algorithm for its calculation, and indicated the advantages of this approach in the analysis of time series; the application of this method in the analysis of various model systems and experimental data has also been demonstrated.     

The Cambridge jet algorithm: features and applications

Jet clustering algorithms are widely used to analyse hadronic events in high energy collisions. Recently a new clustering method, known as ‘Cambridge’, has been introduced. In this article we present an algorithm to determine the transition values of for this clustering scheme, which allows to resolve any event to a definite number of jets in the final state. We discuss some particularities of the Cambridge clustering method and compare its performance to the Durham clustering scheme for Monte Carlo generated annihilation events. Received: 11 March 1998 / Published online: 2 July 1998     

Rules extraction in short memory time series using genetic algorithms

Data mining is performed using genetic algorithm on artificially generated time series data with short memory. The extraction of rules from a training set and the subsequent testing of these rules provide a basis for the predictions on the test set. The artificial time series are generated using the inverse whitening transformation, and the correlation function has an exponential form with given time constant indicative of short memory. A vector quantization technique is employed to classify the daily rate of return of this artificial time series into four categories. A simple genetic algorithm based on a fixed format of rules is introduced to do the forecasting. Comparing to the benchmark tests with random walk and random guess, genetic algorithms yield substantially better prediction rates, between 50% to 60%. This is an improvement compared with the 47% for random walk prediction and 25% for random guessing method. Received 29 August 2000