Preliminary Study of Reconstruction of a Dynamic System Using an One－Dimensional Time Series

This paper concerns the reconstruction of a dynamic system based on phase space continuation of monthly meantemperature iD time series and the assumption that the equation for the time-varying evolution of phase space statevariables contains linear and nonlinear quadratic terms. followed by the fitting of the dataset subjected to continuation so as to get, by the least square method. the coefficients of the terms, of which those of greater variance contribution are retained for use. Results show that the obtained low-order system may be used to describe nonlinear properties of the short range climate variation shown by monthly mean temperature series.

Fuzzy Time Series Forecasting Based On K-Means Clustering

Many forecasting models based on the concepts of Fuzzy time series have been proposed in the past decades. These models have been widely applied to various problem domains, especially in dealing with forecasting problems in which historical data are linguistic values. In this paper, we present a new fuzzy time series forecasting model, which uses the historical data as the universe of discourse and uses the K-means clustering algorithm to cluster the universe of discourse, then adjust the clusters into intervals. The proposed method is applied for forecasting University enrollment of Alabama. It is shown that the proposed model achieves a significant improvement in forecasting accuracy as compared to other fuzzy time series forecasting models.

Grey Relation between Nonlinear Characteristic and Dynamic Uncertainty of Rolling Bearing Friction Torque

The rolling bearing friction torque which is characterized by its uncertainty and nonlinearity affects heavily the dynamic performance of a system such as missiles, spacecrafts and radars, etc. It is difficult to use the classical statistical theory to evaluate the dynamic evaluation of the rolling bearing friction torque for the lack of prior information about both probability distribution and trends. For this reason, based on the information poor system theory and combined with the correlation dimension in chaos theory, the concepts about the mean of the dynamic fluctuant range （MDFR） and the grey relation are proposed to resolve the problem about evaluating the nonlinear characteristic and the dynamic uncertainty of the rolling bearing friction torque. Friction torque experiments are done for three types of the rolling bearings marked with HKTA, HKTB and HKTC separately; meantime, the correlation dimension and MDFR are calculated to describe the nonlinear characteristic and the dynamic uncertainty of the friction torque, respectively. And the experiments reveal that there is a certain grey relation between the nonlinear characteristic and the dynamic uncertainty of the rolling bearing friction torque, viz. MDFR will become the nonlinear increasing trend with the correlation dimension increasing. Under the condition of fewer characteristic data and the lack of prior information about both probability distribution and trends, the unitive evaluation for the nonlinear characteristic and the dynamic uncertainty of the rolling bearing friction torque is realized with the grey confidence level of 87.7%-96.3%.

Integrated parallel forecasting model based on modified fuzzy time series and SVM

A dynamic parallel forecasting model is proposed, which is based on the problem of current forecasting models and their combined model. According to the process of the model, the fuzzy C-means clustering algorithm is improved in outliers operation and distance in the clusters and among the clusters. Firstly, the input data sets are optimized and their coherence is ensured, the region scale algorithm is modified and non-isometric multi scale region fuzzy time series model is built. At the same time, the particle swarm optimization algorithm about the particle speed, location and inertia weight value is improved, this method is used to optimize the parameters of support vector machine, construct the combined forecast model, build the dynamic parallel forecast model, and calculate the dynamic weight values and regard the product of the weight value and forecast value to be the final forecast values. At last, the example shows the improved forecast model is effective and accurate.

Time Series Forecasting with Multiple Deep Learners: Selection from a Bayesian Network

Considering the recent developments in deep learning, it has become increasingly important to verify what methods are valid for the prediction of multivariate time-series data. In this study, we propose a novel method of time-series prediction employing multiple deep learners combined with a Bayesian network where training data is divided into clusters using K-means clustering. We decided how many clusters are the best for K-means with the Bayesian information criteria. Depending on each cluster, the multiple deep learners are trained. We used three types of deep learners: deep neural network (DNN), recurrent neural network (RNN), and long short-term memory (LSTM). A naive Bayes classifier is used to determine which deep learner is in charge of predicting a particular time-series. Our proposed method will be applied to a set of financial time-series data, the Nikkei Average Stock price, to assess the accuracy of the predictions made. Compared with the conventional method of employing a single deep learner to acquire all the data, it is demonstrated by our proposed method that F-value and accuracy are improved.

基于K-means-LSTM组合算法的地质灾害监测设备故障排查模型设计

为提高地质灾害预警的准确性和效率,提出了一种基于K-means聚类和长短期记忆网络(LSTM)的地质灾害监测设备故障排查模型。通过对地质监测数据的聚类分析,该模型能有效区分正常和异常运行状态的设备,为后续的深度学习分析提供了精准的数据基础。LSTM时序分析部分利用聚类结果,深入挖掘时间序列数据中的潜在模式和趋势,实现对设备故障类型及其发展趋势的准确预测。实验验证表明,该组合模型在地质灾害监测领域具有良好的应用潜力,能够为灾害预防和减灾提供有力的技术支持。未来研究将集中于进一步提升模型的准确性和泛化能力,探索更多算法组合和数据处理方法,以适应更加复杂的监测环境,推进监测系统的自动化和智能化。

基于K-means和MTLS-SVM算法的生理参数监测系统

在非医模式的生理参数监测系统中,对监测参数进行学习,可以提高诊断和预测精度。针对多任务时间序列中存在的信息挖掘不充分、预测精度低等问题,将机器学习中的监督和半监督学习方式结合起来对远程健康监护对象进行生理状况预测。该方法用K-means算法将相同类别的数据集群,并使用多任务最小二乘支持向量机(MTLS-SVM)来训练历史数据来进行趋势预测。为了评估该方法的有效性,将MTLS-SVM方法与K-means、MTLS-SVM方法比较,实验结果表明该方法具有较高的预测精度。

一种融合SOM与K-means算法的动态信用评价方法及应用

针对传统信用评价方法多是静态评价的不足,本文提出了一种融合SOM与K-means算法的动态信用评价方法。文章首先对动态信用评价问题进行了介绍,并利用E-TOPSIS方法对单时点下的静态信息进行集结,以确定被评价对象的信用评价值;然后在融合SOM算法和K-means算法各自优势的基础上,提出了SOM-K算法的原理和步骤;最后以SOM-K算法对被评价对象进行聚类,并确定相应信用等级。文章最后进行了实例验证。验证结果表明,该方法能够较好地克服静态信息下由于信息突变造成评价结果失真的问题。

基于ICA K-Means的产品口碑演化聚类与营销分析

对于产品而言,其在线口碑的活跃度是非常具有代表性的一个指标。在线口碑活跃度的高低,直接揭示产品的生命周期演化模式,对于产品生命周期有全面的了解有助于决策者制定营销计划以及战略。但由于产品在线评论的高维性和复杂性,使得其聚类的难度加大。所以,在普通的K均值算法的基础上引入独立成份分析,对异类产品之间或同类产品在线口碑的活跃度之间进行聚类分析,可以大大降低复杂性和提升聚类准确性;同时深入分析提取出的产品生命周期曲线,有效提升在线口碑信息在电子商务营销管理与决策支持中的作用,深化在线口碑活跃度的管理学视角研究。

基于K-means MCMC算法的中长期风电时间序列建模方法研究

构建风电功率时间序列模型对电力系统中长期规划、年/月调度和安全稳定运行具有重要意义。针对传统马尔科夫链-蒙特卡洛法(Markovchain-MonteCarlo,MCMC)法存在的缺陷,提出一种基于粒子群优化的K-means MCMC风电时间序列建模新方法。首先,对历史风电功率数据进行聚类,并对聚类后的不同类别风电功率序列选取最优状态数,分别建立状态转移矩阵;其次,用拟合度较好的混合高斯分布拟合多时间尺度的风电最大波动率的概率分布特性;最后,采用基于类间转移概率矩阵的MCMC方法依次生成模拟风电出力时间序列;同时,在生成模拟序列过程中叠加高频波动分量,使模拟序列延续历史风电序列的波动特性。通过对比本所提方法和传统MCMC法分别生成的模拟风电出力序列以及历史风电功率序列,验证了所提方法的有效性和准确性。

一种时间序列数据的动态k-means聚类算法

传统k-means聚类算法是对某个时间片上的静态数据集合进行独立的聚类分析,但对于时间序列数据仅仅是多次静态聚类分析的重复应用。当数据量过大时,算法的时间开销将大大增加。为此,本文提出了一种时间序列数据的动态k-means聚类算法(Dynamic k-means Clustering Algorithm for Time Series Data,DKCA/TSD)。该算法通过时间序列的前一时刻最优质心的结果,利用数据之间的关联性进行下一时刻的聚类,从而减少算法的迭代次数,提高时间效率。实验结果表明:对于时间序列数据,DKCA/TSD算法相对于k-means算法时间效率上有很大提高。

A Dynamic Forecasting System with Applications in Production Logistics

Production logistics involve the co-ordination of ac tivities such as production and materials control (PMC), inventory management, p roduct life cycle management, etc. Those activities demand for an accurate forec asting model. However, the conventional methods of making sell and buy decision based on human forecast or conventional moving average and exponential smoothing methods is no longer be sufficient to meet the future need. Furthermore, the un derlying statistics of the market information change from time to time due to a number of reasons such as change of global economic environment, government poli cies and business risks. This demands for highly adaptive forecasting model which is robust enough to response and adapt well to the fast changes in the dat a characteristics, in other words, the trajectory of the "dynamic characteristic s" of the data. In this paper, an adaptive time-series modelling method was proposed for short -term dynamic forecasting. The method employs an autoregressive (AR) time-seri es model to carry out the forecasting process. A modified least mean square (MLM S) adaptive filter algorithm was established for adjusting the AR model coeffici ents so as to minimise the sum of squared of forecasting errors. A prototype dyn amic forecasting system was built based on the adaptive time-series modelling m ethod. Basically, the dynamic forecasting system can be divided into two phases, i.e. the Learning Phase and the Application Phase. The learning procedures star t with the determination of upper limit of the adaptation gain based on the conv ergence in the mean square criterion. Hence, the optimum ELMS filter parameters are determined using an iteration algorithm which changes each filter parameter i.e. the order, the adaptation gain andthe values initial coefficient vector on e by one inside a predetermined iteration range. The set of parameters which giv es the minimum value for sum of squared errors within the iteration range is sel ected as the optimum set of filter parameters. In the Application Phase, the sys tem is operated under a real-time environment. The sampled data is processed by the optimised ELMS filter and the forecasted data are calculated based on the a daptive time-series model. The error of forecasting is continuously monitored w ithin the predefined tolerance. When the system detects excessive forecasting er ror, a feedback alarm signal was issued for system re-calibration. Experimental results indicated that the convergence rate and sum of squared erro rs during initial adaptation could be significantly improved using the MLMS algorithm. The performance of the system was verified through a series of experi ments conducted on the forecast of materials demand and costing in productio n logistics. Satisfactory results were achieved with the forecast errors confini ng within in most instances. Further applications of the system can be found i n sales demand forecast, inventory management as well as collaborative planning, forecast and replenishment (CPFR) in logistics engineering.

Empirical analysis on risk of security investment

The paper analyzes the theory and application of Markowitz Mean-Variance Model and CAPM model. Firstly, it explains the development process and standpoints of two models and deduces the whole process in detail. Then 30 stocks are choosen from Shangzheng 50 stocks and are testified whether the prices of Shanghai stocks conform to the two models. With the technique of time series and panel data analysis, the research on the stock risk and effective portfolio by ORIGIN and MATLAB software is conducted. The result shows that Shanghai stock market conforms to Markowitz Mean-Variance Model to a certain extent and can give investors reliable suggestion to gain higher return, but there is no positive relation between system risk and profit ratio and CAPM doesn＇t function well in China＇s security market.

Detection of Chaoticity in Daily Mean Temperature Time Series

The chaoticity of daily mean temperature time series is investigated with complex systems theory. The data set that has been used in this analysis consists of daily mean temperature recorded at four stations in China. The power spectrums axe used to obtain some preliminaxy information regaxding the temperature time series. R/S analysis provides evidence for fractaiity in temperature time series. Cao method and correlation dimension, as well as the largest Lyapunov exponent, give consistent results, which does not exclude the possibility of deterministic chaos for the four daily mean temperature series. The research provides a principled basis for further study of temperature data with nonlinear dynamical methods.

Clustering Countries on COVID-19 Data among Different Waves Using K-Means Clustering

The COVID-19 pandemic has caused an unprecedented spike in confirmed cases in 230 countries globally. In this work, a set of data from the COVID-19 coronavirus outbreak has been subjected to two well-known unsupervised learning techniques: K-means clustering and correlation. The COVID-19 virus has infected several nations, and K-means automatically looks for undiscovered clusters of those infections. To examine the spread of COVID-19 before a vaccine becomes widely available, this work has used unsupervised approaches to identify the crucial county-level confirmed cases, death cases, recover cases, total_cases_per_million, and total_deaths_per_million aspects of county-level variables. We combined countries into significant clusters using this feature subspace to assist more in-depth disease analysis efforts. As a result, we used a clustering technique to examine various trends in COVID-19 incidence and mortality across nations. This technique took the key components of a trajectory and incorporates them into a K-means clustering process. We separated the trend lines into measures that characterize various features of a trend. The measurements were first reduced in dimension, then clustered using a K-means algorithm. This method was used to individually calculate the incidence and death rates and then compare them.

基于ARIMA模型和K−Means的组合异常检测方法

针对时间序列中异常点的检测计算问题,提出了一种基于ARIMA模型和基于K−Means模型的组合异常检测计算方法。首先测试训练采用差分自回归移动平均模型(ARIMA),之后采用滑动窗口配合ARIMA模型对测试集进行预测得到异常预测值,然后计算误差项以及误差项的置信区间,误差项在置信区间判定范围以外的,将其对应的原始值判定为异常值。在检测出异常值之后,采用K−Means算法对原始数据进行聚类,然后通过计算出状态转移概率,对检测出的异常的取值结果进行质量评估,最后确定出异常值。实验探讨了算法中的滑动窗口对异常检测的影响,并以NAB部分数据集对算法进行了验证。实验结果表明,与同类经典算法相比,该算法不仅能够有效检测出时间序列中的异常点,而且在提高精准率和正确率方面取得了很好的效果。

基于符号表示的可度量shapelets提取的时序分类研究

在时序分类问题中,基于符号表示的shapelets提取方法具有良好的分类精度和分类效率,但对符号进行质量度量的过程,如计算TFIDF分数,耗时较长且计算量大,导致分类效率较低。此外,提取的shapelets候选数量仍然较多,判别力有待提高。针对这些问题,本文提出了一种基于符号表示的可度量shapelets提取方法,该方法包含时间序列数据预处理、确定shapelets候选集和学习shapelets 3个阶段,可以快速得到高质量shapelets。在数据预处理阶段,将时间序列转化为符号聚合近似(SAX)表示以降低原始时间序列的维度。在确定shapelets候选集阶段,利用Bloom过滤器过滤重复的SAX词,并将过滤后的SAX词存储在哈希表中进行质量度量。随后,对SAX词的相似性进行判别,基于相似性和覆盖度等概念确定最终的shapelets候选集。在学习shapelets阶段,采用logistic回归模型学得真正的shapelets用于时序分类。在32个数据集上进行了大量实验,实验结果表明,所提方法的平均分类精度和平均分类效率均排名第二。与现有的基于shapelets的时序分类方法相比,该方法可以在保证精度的同时提高分类效率,并且具有良好的可解释性。

基于时域卷积网络的精轧出口厚度预测

以精轧过程为研究对象,引入时域卷积网络算法,构建了基于时域卷积网络的精轧出口厚度预测模型。利用时域卷积网络模型提取精轧过程时序数据的特征信息,通过优化模型结构和参数,提升精轧出口厚度预测性能。实际钢种数据集仿真实验结果表明,相较于传统方法,本文所提出的时域卷积网络算法在均方根误差、平均绝对百分比误差及决定系数等评价指标方面存在较大优势,可为现场工程师提供重要的决策信息。

一种二模态天气分型方法及其在光伏功率概率预测的应用

天气分型是光伏功率预测中不可或缺的预处理步骤,为精细刻画光伏出力的不确定性,提出一种新的基于光伏功率聚类的二模态天气分类方法。该方法结合气象信息和功率信息进行天气分型,为天气分型在光伏功率预测的应用提供了一条有效的新路径。此外,该方法使用数据融合技术,依据融合数值天气预报(numeric weather prediction,NWP)气象和实际气象二者间的相关信息进行天气分型,以减少模型对NWP准确度的依赖并提高模型的鲁棒性。以吉林某光伏电站数据为例,验证了该天气分型方法的合理性,同时,将天气分型方法与功率概率预测相结合,其测算结果表明,使用所提方法进行天气分型概率预测的区间覆盖率更接近预设的置信水平,且平均带宽更窄。

地铁隧道通风系统设备故障统计分析及RAM指标分配

通过调研华南某地区3条地铁线路近五年隧道通风系统故障数据,分析设备运营维护薄弱环节,得到隧道通风系统各关键设备的故障类型、故障原因以及各设备类型的故障数目和比例,提供故障检修及维护建议;综合考虑设备故障率和设备数量,结合各设备故障数和平均修复时间MTTR,计算隧道风系统各关键设备平均无故障工作时间MTBR,该RAM指标可供同行参考。