时间序列趋势相似性度量方法研究

为了进一步改善和提高基于模式的时间序列趋势相似性度量效果,在时间序列分段线性表示的基础上,依据分段子序列的均值及其线性拟合函数的导数符号,实现时间序列的分段模式化,以模式之间的异同性定义模式匹配距离,借鉴动态时间弯曲（Dynamic Time Warping,DTW）的动态规划原理,提出一种动态模式匹配方法（Dynamic Pattern Matching,DPM）。实验结果表明,该方法能够在不同压缩率条件下,准确度量等长时间序列的趋势相似性,而且时间消耗较低。时间序列不等长作为存在数据缺失的一种表现形式,该方法的度量效果与数据缺失比例之间的关系值得进一步的深入研究。     

基于分段极值DTW距离的时间序列相似性度量

在时间序列相似性的研究中,通常采用的欧氏距离及其变形无法对在时间轴上发生伸缩或弯曲的序列进行相似性度量,本文提出了一种基于分段极值DTW距离的时间序列相似性度量方法可以解决这一问题。在动态时间弯曲(DTW)距离的基础上,本文定义了序列的分段极值DTW距离,并阐述了其完整的算法实现。与传统的DTW距离相比,分段极值DTW距离在保证度量准确性的同时大大提高了相似性计算的效率。文中最后运用MATLAB作对比实验,并给出实验结果数据,验证了该度量方法的有效性与准确性。     

基于多维分段和动态权重DTW的多元时间序列相似性度量方法

针对常用方法无法准确度量多元时间序列相似程度的问题,提出一种基于多维分段和动态权重动态时间弯曲距离的多元时间序列相似性度量方法.首先对多元时间序列进行多维分段拟合,选取拟合段的斜率、均值和时间跨度作为每一段的特征,在对多元时间序列降维的同时也保留了变量之间的相关性;然后提出一种动态权重动态时间弯曲距离度量方法计算多元时间序列特征矩阵之间的距离,避免了直接使用动态时间弯曲距离造成的畸形匹配问题.最终实验结果也验证了该方法在多种类型的数据集上都能取得较高的度量精度,表明了该方法的有效性.     

时间序列的层次分段及相似性度量

时间序列的相似性度量是时间序列数据挖掘的研究基础,为数据挖掘任务的效率和准确度提供可靠的保障。提出一种时间序列的层次分段及相似性度量方法,方法首先识别时间序列中的极值点,依据极值点的特征对时间序列进行分层次分段,并以此为基础,通过定义新的距离公式来度量时间序列间的相似性。使用新提出的相似性度量方法对时间序列进行聚类计算,实验结果表明,该方法能够有效地度量时间序列间的相似性,聚类效果明显,具有较好的实用性和良好的应用前景。     

多元时间序列相似性度量方法

现有的多元时间序列相似性度量方法 难以平衡度量准确性和计算效率之间的矛盾.针对该问题,首先,对多元时间序列进行多维分段拟合;然后,选取各分段上序列点的均值作为特征;最后,以特征序列作为输入,利用动态时间弯曲算法实现相似性度量.实验结果表明,所提出方法参数配置简单,能够在保证度量准确性的前提下有效降低计算复杂度.     

基于增量动态时间弯曲的时间序列相似性度量方法

针对动态时间弯曲方法计算时间过长的问题,提出增量动态时间弯曲来度量较长时间序列之间的相似性。首先利用动态时间弯曲方法对历史时间序列数据进行相似性度量,得到相应的历史最优弯曲路径和路径中各元素的累积距离代价。其次,通过逆向弯曲度量方法完成当前序列数据 的相似性度量,结合历史数据信息找到与历史弯曲路径相交且度量时间序列距离为当前最小值的新路径,进而实现增量动态时间弯曲的相似性度量。该方法不仅具有良好的度量质量,还具有较高的时间效率。数值实验表明,对于大部分时间序列数据集,新方法的分类准确率和计算性能要优于经典动态时间弯曲。     

基于分段特征及自适应加权的DTW相似性度量

利用动态时间弯曲（DTW）技术在原始多元时间序列进行相似性度量时时间复杂度较高,且DTW在追求最小弯曲距离的过程中可能会出现过渡拉伸和压缩的问题。提出一种基于分段特征及自适应加权的DTW多元时间序列相似性度量方法。对原始时间序列在各个变量维度上进行统一分段,选取分段后拟合线段的斜率、分段区间的最大值和最小值以及时间跨度作为每一段的特征,实现对原始序列的大幅降维,提高计算效率。在DTW计算最佳弯曲路径的过程中为每个点设置自适应代价权重,限制弯曲路径中点列的重复使用次数,改善时间序列因过度拉伸或压缩所导致的度量精度低的问题,以得到最优路径路线。实验结果表明,该方法能很好地度量多元时间序列之间的相似性,在多个数据集上都能取得较好的度量结果。     

基于分段降维和路径修正DTW的时序特征分类器设计

针对传统的动态时间弯曲（DTW）度量方法易出现过度弯曲现象且计算复杂度高、算法效率低等问题,提出一种基于路径修正的动态时间弯曲（UDTW）度量方法。首先通过分段降维方法——分段局部最大值平滑法（PLM）有效提取序列特征信息,减少UDTW的计算代价;其次,考虑了时间序列形态特征的相似性要求,给过度弯曲路径设置动态惩罚系数,以此修正路径的弯曲程度;最后,在改进度量距离基础上,采用1-近邻分类算法对时序数据进行分类,以提高时间序列相似性度量的准确率和效率。实验结果表明,在15个UCR数据集上,UDTW度量方法与传统DTW度量方法相比具有更高的分类准确率,UDTW在其中3个数据集上能实现100%分类正确;与导数DTW（DDTW）度量方法相比,UDTW分类准确率最多提高了71.8%,而PLM-UDTW在不影响分类准确率的前提下执行时间减小了99%。     

基于滑动平均与分段线性回归的时间序列相似性

针对时间序列相似性度量中欧氏距离对异常数据敏感以及DTW距离算法效率低的问题,提出基于滑动平均与分段线性回归的时间序列相似性方法。首先,使用初始可变滑动平均算法以及分段线性回归对原始时间序列进行数据变换,并将分段线性回归的参数(截距与距离)集作为时间序列的特征,以实现时间序列的特征提取和数据降维;然后,利用动态时间弯曲距离进行距离计算。该方法在时间序列相似性上与DTW算法的性能相近,但是在算法效率上几乎提高了96%。实验结果验证了该方法的有效性与准确性。     

一种基于DTW的新型故事时间序列相似性度量方法

现有时间序列相似性度量方法在进行股市序列相似性分析时,通常忽略成交量等其他重要因素对股价的影响,从而导致序列聚类、分类不精确。针对这一问题,本文提出了新的股市时间序列相似性度量方法。该方法在动态时间弯曲算法的基础上,通过引进时间衰竭因子,并结合成交量因素,给出了股市序列的最终度量公式。为了证明提出方法的可行性和有效性,本文实验部分通过选取家电等三个行业中的股票数据进行测试。实验结果表明,基于动态时间弯曲(Dynamic time warping,DTW)的新型股市时间序列相似性度量方法能够在保持股票序列形态特征的基础上,较好地解决股市技术分析中量价关系问题,从而更有效地应用于股市技术分析里关于模式发现等领域。     

流式时间序列的实时相似度研究

动态时间弯曲算法虽然适合度量时间序列的相似度,但是在大数据背景下,对于序列个数多、潜在长度可能是无穷、实时性要求高的流式时间序列,面临着算法简单、计算不简单的可计算问题。以Spark计算平台为基础,针对流式时间序列的特点,提出了一种流式动态时间弯曲算法,能实时计算动态时间序列近似值,误差可控、稳定,且具备大数据计算能力。最后通过实验验证了算法的可行性和稳定性。     

Multiple time scale decomposition of discrete time Markov chains

The multiple time scale decomposition of discrete time, finite state Markov chains is addressed. In [1, 2], the behavior of a continuous time Markov chain is approximated using a fast time scale, ε-independent, continuous time process, and a reduced order perturbed process. The procedure can then be iterated to obtain a complete multiple time scale decomposition. In the discrete time case presented in this paper, the basic approximation has a ‘hybrid’ form. In this form, the fast time scale behavior is approximated using an ε-independent, discrete time Markov chain, and the slow behavior is captured by a perturbed, continuous time process. Further time scale decomposition then involves the continuous time procedure in [1, 2]. This extension to discrete time chains bridges previous multiple time scale decomposition results, which have dealt exclusively with either continuous time or discrete time processes, and provides a uniform framework for the analysis of both types of systems.     

Electromagnetic scattered field time series from finite difference time domain trained time delay neural network

This paper uses time delay neural network (TDNN) for predicting electromagnetic (EM) fields scattered from dielectric objects (cylinder, cylinder‐hemisphere, and cylinder‐cone) using: (a) FDTD generated initial field data for similar conducting objects and (b) Statistical information for the nature of fields. Statistical data indicated that the scattered field nature is close to deterministic. The TDNN structure determination uses statistical data for fixing the number of delays and tabular technique to obtain the number of hidden neurons. The TDNN training uses the Levenberg‐Marquardt (LM) algorithm. The model outputs follow standard FDTD results closely.     

Finite‐time distributed control with time transformation

In this article, we propose distributed control algorithms for first‐ and second‐order multiagent systems for addressing finite‐time control problem with a priori given, user‐defined finite‐time convergence guarantees. The proposed control frameworks are predicated on a recently developed time transformation approach. Specifically, our contribution is twofold: First, a generalized time transformation function is proposed that converts the user‐defined finite‐time interval to a stretched infinite‐time interval, where one can design a distributed control algorithm on this stretched interval and then transform it back to the original finite‐time interval for achieving a given multiagent system objective. Second, for a specific time transformation function, we analytically establish the robustness properties of the resulting finite‐time distributed control algorithms against vanishing and nonvanishing system uncertainties. By contrast to existing finite‐time approaches, it is shown that the proposed algorithms can preserve a priori given, user‐defined finite‐time convergence regardless of the initial conditions of the multiagent system, the graph topology, and without requiring a knowledge of the upper bounds of the considered class of system uncertainties. Illustrative numerical examples are included to further demonstrate the efficacy of the presented results.     

Estimated time of arrival and debiasing the time saving bias

The time saving bias predicts that the time saved when increasing speed from a high speed is overestimated, and underestimated when increasing speed from a slow speed. In a questionnaire, time saving judgements were investigated when information of estimated time to arrival was provided. In an active driving task, an alternative meter indicating the inverted speed was used to debias judgements. The simulated task was to first drive a distance at a given speed, and then drive the same distance again at the speed the driver judged was required to gain exactly 3 min in travel time compared with the first drive. A control group performed the same task with a speedometer and saved less than the targeted 3 min when increasing speed from a high speed, and more than 3 min when increasing from a low speed. Participants in the alternative meter condition were closer to the target. The two studies corroborate a time saving bias and show that biased intuitive judgements can be debiased by displaying the inverted speed.     

Enhancing the settling time estimation of a class of fixed‐time stable systems

In this paper, we provide a new nonconservative upper bound for the settling time of a class of fixed‐time stable systems. To expose the value and the applicability of this result, we present four main contributions. First, we revisit the well‐known class of fixed‐time stable systems, to show the conservatism of the classical upper estimate of its settling time. Second, we provide the smallest constant that the uniformly upper bounds the settling time of any trajectory of the system under consideration. Third, introducing a slight modification of the previous class of fixed‐time systems, we propose a new predefined‐time convergent algorithm where the least upper bound of the settling time is set a priori as a parameter of the system. At last, we design a class of predefined‐time controllers for first‐ and second‐order systems based on the exposed stability analysis. Simulation results highlight the performance of the proposed scheme regarding settling time estimation compared to existing methods.     

On the design of nonautonomous fixed‐time controllers with a predefined upper bound of the settling time

Recently, there has been a great deal of attention in a class of finite‐time stable dynamical systems, called fixed‐time stable, that exhibit uniform convergence with respect to its initial condition, that is, there exists an upper bound for the settling‐time (UBST) function, independent of the initial condition of the system. Of particular interest is the development of stabilizing controllers where the desired UBST can be selected a priori by the user since it allows the design of controllers to satisfy real‐time constraints. Unfortunately, existing methodologies for the design of controllers for fixed‐time stability exhibit the following drawbacks: on the one hand, in methods based on autonomous systems, either the UBST is unknown or its estimate is very conservative, leading to over‐engineered solutions; on the other hand, in methods based on time‐varying gains, the gain tends to infinity, which makes these methods unrealizable in practice. To bridge these gaps, we introduce a design methodology to stabilize a perturbed chain of integrators in a fixed‐time, with the desired UBST that can be set arbitrarily tight. Our approach consists of redesigning autonomous stabilizing controllers by adding time‐varying gains. However, unlike existing methods, we provide sufficient conditions such that the time‐varying gain remains bounded, making our approach realizable in practice.     

基于可拓变换与时间模块的产品族加工时间快速确定方法

为提高定制企业工时定额制定的效率以满足顾客对产品交货期的需求,提出一种新的工时定额计算方法。该方法以产品族的结构组成为基础,将产品族加工时间分为静态工时、柔性工时、特有工时三大工时模块,提出了时间模块的概念,并运用可拓变换的思想对柔性工时模块进行了可拓学描述,运用工艺聚类分析实现了各工时模块时间的计算。通过对制造对象物元的三级匹配搜索和事元的匹配搜索提取相似的制造对象,以实现加工时间的快速重用。最后通过实例验证了方法的可行性和有效性。     

Forecasting daily time series using periodic unobserved components time series models

A periodic time series analysis is explored in the context of unobserved components time series models that include stochastic time functions for trend, seasonal and irregular effects. Periodic time series models allow dynamic characteristics (autocovariances) to depend on the period of the year, month, week or day. In the standard multivariate approach one can interpret a periodic time series analysis as a simultaneous treatment of typically yearly time series where each series is related to a particular season. Here, the periodic analysis applies to a vector of monthly time series related to each day of the month. Particular focus is on the forecasting performance and therefore on the underlying periodic forecast function, defined by the in-sample observation weights for producing (multi-step) forecasts. These weight patterns facilitate the interpretation of periodic model extensions. A statistical state space approach is used to estimate the model and allows for irregularly spaced observations in daily time series. Recent algorithms are adopted for the computation of observation weights for forecasting based on state space models with regressor variables. The methodology is illustrated for daily Dutch tax revenues that appear to have periodic dynamic properties. The dimension of our periodic unobserved components model is relatively large as we allow each element (day) of the vector of monthly time series to have a changing seasonal pattern. Nevertheless, even with only five years of data we find that the increased periodic flexibility can help in out-of-sample forecasting for two extra years of data.