TSCND:Temporal Subsequence-Based Convolutional Network with Difference for Time Series Forecasting

Time series forecasting plays an important role in various fields, such as energy, finance, transport, and weather. Temporal convolutional networks (TCNs) based on dilated causal convolution have been widely used in time series forecasting. However, two problems weaken the performance of TCNs. One is that in dilated casual convolution, causal convolution leads to the receptive fields of outputs being concentrated in the earlier part of the input sequence, whereas the recent input information will be severely lost. The other is that the distribution shift problem in time series has not been adequately solved. To address the first problem, we propose a subsequence-based dilated convolution method (SDC). By using multiple convolutional filters to convolve elements of neighboring subsequences, the method extracts temporal features from a growing receptive field via a growing subsequence rather than a single element. Ultimately, the receptive field of each output element can cover the whole input sequence. To address the second problem, we propose a difference and compensation method (DCM). The method reduces the discrepancies between and within the input sequences by difference operations and then compensates the outputs for the information lost due to difference operations. Based on SDC and DCM, we further construct a temporal subsequence-based convolutional network with difference (TSCND) for time series forecasting. The experimental results show that TSCND can reduce prediction mean squared error by 7.3% and save runtime, compared with state-of-the-art models and vanilla TCN.

Spatio-temporal changes of underground coal fires during 2008-2016 in Khanh Hoa coal field(North-east of Viet Nam) using Landsat time-series data

Underground coal fires are one of the most common and serious geohazards in most coal producing countries in the world. Monitoring their spatio-temporal changes plays an important role in controlling and preventing the effects of coal fires, and their environmental impact. In this study, the spatio-temporal changes of underground coal fires in Khanh Hoa coal field(North-East of Viet Nam) were analyzed using Landsat time-series data during the 2008-2016 period. Based on land surface temperatures retrieved from Landsat thermal data, underground coal fires related to thermal anomalies were identified using the MEDIAN+1.5×IQR(IQR: Interquartile range) threshold technique. The locations of underground coal fires were validated using a coal fire map produced by the field survey data and cross-validated using the daytime ASTER thermal infrared imagery. Based on the fires extracted from seven Landsat thermal imageries, the spatiotemporal changes of underground coal fire areas were analyzed. The results showed that the thermalanomalous zones have been correlated with known coal fires. Cross-validation of coal fires using ASTER TIR data showed a high consistency of 79.3%. The largest coal fire area of 184.6 hectares was detected in 2010, followed by 2014(181.1 hectares) and 2016(178.5 hectares). The smaller coal fire areas were extracted with areas of 133.6 and 152.5 hectares in 2011 and 2009 respectively. Underground coal fires were mainly detected in the northern and southern part, and tend to spread to north-west of the coal field.

Adaptive Modeling and Forecasting of Time Series by Combining the Methods of Temporal Differences with Neural Networks

This paper discusses the modeling method of time series with neural network. In order to improve the adaptability of direct multi-step prediction models, this paper proposes a method of combining the temporal differences methods with back-propagation algorithm for updating the parameters continuously on the basis of recent data. This method can make the neural network model fit the recent characteristic of the time series as close as possible, therefore improves the prediction accuracy. We built models and made predictions for the sunspot series. The prediction results of adaptive modeling method are better than that of non-adaptive modeling methods.

A Time Series Intrusion Detection Method Based on SSAE,TCN and Bi-LSTM

In the fast-evolving landscape of digital networks,the incidence of network intrusions has escalated alarmingly.Simultaneously,the crucial role of time series data in intrusion detection remains largely underappreciated,with most systems failing to capture the time-bound nuances of network traffic.This leads to compromised detection accuracy and overlooked temporal patterns.Addressing this gap,we introduce a novel SSAE-TCN-BiLSTM(STL)model that integrates time series analysis,significantly enhancing detection capabilities.Our approach reduces feature dimensionalitywith a Stacked Sparse Autoencoder(SSAE)and extracts temporally relevant features through a Temporal Convolutional Network(TCN)and Bidirectional Long Short-term Memory Network(Bi-LSTM).By meticulously adjusting time steps,we underscore the significance of temporal data in bolstering detection accuracy.On the UNSW-NB15 dataset,ourmodel achieved an F1-score of 99.49%,Accuracy of 99.43%,Precision of 99.38%,Recall of 99.60%,and an inference time of 4.24 s.For the CICDS2017 dataset,we recorded an F1-score of 99.53%,Accuracy of 99.62%,Precision of 99.27%,Recall of 99.79%,and an inference time of 5.72 s.These findings not only confirm the STL model’s superior performance but also its operational efficiency,underpinning its significance in real-world cybersecurity scenarios where rapid response is paramount.Our contribution represents a significant advance in cybersecurity,proposing a model that excels in accuracy and adaptability to the dynamic nature of network traffic,setting a new benchmark for intrusion detection systems.

Relationship Between Urban Road Traffic Characteristics and Road Grade Based on a Time Series Clustering Model： A Case Study in Nanjing, China

With the increasing number of vehicles in large-and medium-sized cities challenges in urban traffic management, control, and road planning are being faced. Taxi GPS trajectory data is a novel data source that can be used to study the potential dynamic traffic characteristics of urban roads, and thus identify locations that show a notable lack of road planning. Considering that road traffic characteristics on their own are insufficient for a comprehensive understanding of urban traffic, we develop a road traffic characteristic time series clustering model to analyze the relationship between urban road traffic characteristics and road grade based on existing taxi trajectory data. We select the main urban area of Nanjing as our study area and use the taxi trajectory data of a single month for evaluating our method. The experiments show that the clustering model exhibit good performance and can be successfully used for road traffic characteristic classification. Moreover, we analyze the correlation between traffic characteristics and road grade to identify road segments with planning designs that do not match the actual traffic demands.

A production prediction method of single well in water flooding oilfield based on integrated temporal convolutional network model

Since the oil production of single well in water flooding reservoir varies greatly and is hard to predict, an oil production prediction method of single well based on temporal convolutional network(TCN) is proposed and verified. This method is started from data processing, the correspondence between water injectors and oil producers is determined according to the influence radius of the water injectors, the influence degree of a water injector on an oil producer in the month concerned is added as a model feature, and a Random Forest(RF) model is built to fill the dynamic data of water flooding. The single well history is divided into 4 stages according to its water cut, that is, low water cut, middle water cut, high water cut and extra-high water cut stages. In each stage, a TCN based prediction model is established, hyperparameters of the model are optimized by the Sparrow Search Algorithm(SSA). Finally, the models of the 4 stages are integrated into one whole-life model of the well for production prediction. The application of this method in Daqing Oilfield, NE China shows that:(1) Compared with conventional data processing methods, the data obtained by this processing method are more close to the actual production, and the data set obtained is more authentic and complete.(2) The TCN model has higher prediction accuracy than other 11 models such as Long Short Term Memory(LSTM).(3) Compared with the conventional full-life-cycle models, the model of integrated stages can significantly reduce the error of production prediction.

Fusing multi-source data to map spatio-temporal dynamics of winter rape on the Jianghan Plain and Dongting Lake Plain, China

Mapping crop distribution with remote sensing data is of great importance for agricultural production, food security and agricultural sustainability. Winter rape is an important oil crop, which plays an important role in the cooking oil market of China. The Jianghan Plain and Dongting Lake Plain (JPDLP) are major agricultural production areas in China. Essential changes in winter rape distribution have taken place in this area during the 21st century. However, the pattern of these changes remains unknown. In this study, the spatial and temporal dynamics of winter rape from 2000 to 2017 on the JPDLP were analyzed. An artificial neural network (ANN)-based classification method was proposed to map fractional winter rape distribution by fusing moderate resolution imaging spectrometer (MODIS) data and high-resolution imagery. The results are as follows:(1) The total winter rape acreages on the JPDLP dropped significantly, especially on the Jianghan Plain with a decline of about 45% during 2000 and 2017.(2) The winter rape abundance keeps changing with about 20–30% croplands changing their abundance drastically in every two consecutive observation years.(3) The winter rape has obvious regional differentiation for the trend of its change at the county level, and the decreasing trend was observed more strongly in the traditionally dominant agricultural counties.

Modeling and Simulation of Time Series Prediction Based on Dynamic Neural Network

Molding and simulation of time series prediction based on dynamic neural network(NN) are studied. Prediction model for non-linear and time-varying system is proposed based on dynamic Jordan NN. Aiming at the intrinsic defects of back-propagation (BP) algorithm that cannot update network weights incrementally, a hybrid algorithm combining the temporal difference (TD) method with BP algorithm to train Jordan NN is put forward. The proposed method is applied to predict the ash content of clean coal in jigging production real-time and multi-step. A practical example is also given and its application results indicate that the method has better performance than others and also offers a beneficial reference to the prediction of nonlinear time series.

A PRELIMINARY STUDY OF FOURIER SERIES ANALYSIS ON CLOUD TRACKING WITH GOES HIGH TEMPORAL RESOLUTION IMAGES

Fourier series analysis is proposed as a new technique to address the problem of“sub-pixel motion”in deriving cloud motion winds(CMW)from high temporal resolution images.Based on a concept different from that of maximum correlation matching technique,the Fourier technique computes phase speed as an estimate of cloud motion.It is very effective for tracking small cellular clouds in 1-min interval images and more efficient for computation than the maximum correlation technique because only two templates in same size are involved in primary tracking procedure. Moreover it obtains not only CMW vectors but potentially also velocity spectrum and variance.A practical example is given to show the cloud motion winds from 1-min interval images with the Fourier method versus those from traditional 30-min interval images with maximum correlation technique.Problems that require further investigation before the Fourier technique can be regarded as a viable technique,especially for cloud tracking with high temporal resolution images,are also revealed.

Exploring and visualizing temporal relations in multivariate time series

This paper introduces an approach to analyzing multivariate time series(MVTS)data through progressive temporal abstraction of the data into patterns characterizing the behavior of the studied dynamic phenomenon.The paper focuses on two core challenges:identifying basic behavior patterns of individual attributes and examining the temporal relations between these patterns across the range of attributes to derive higher-level abstractions of multi-attribute behavior.The proposed approach combines existing methods for univariate pattern extraction,computation of temporal relations according to the Allen’s time interval algebra,visual displays of the temporal relations,and interactive query operations into a cohesive visual analytics workflow.The paper describes the application of the approach to real-world examples of population mobility data during the COVID-19 pandemic and characteristics of episodes in a football match,illustrating its versatility and effectiveness in understanding composite patterns of interrelated attribute behaviors in MVTS data.

Evaluation of temporal compositing algorithms for annual land cover classification using Landsat time series data

In this paper, four widely used temporal compositing algorithms, i.e.median, maximum NDVI, medoid, and weighted scoring-basedalgorithms, were evaluated for annual land cover classification usingmonthly Landsat time series data. Four study areas located in California,Texas, Kansas, and Minnesota, USA were selected for image compositingand land cover classification. Results indicated that images compositedusing weighted scoring-based algorithms have the best spatial fidelitycompared to other three algorithms. In addition, the weighted scoringbasedalgorithms have superior classification accuracy, followed bymedian, maximum NDVI, and medoid in descending order. However, themedian algorithm has a significant advantage in computational efficiencywhich was ~70 times that of weighted scoring-based algorithms, andwith overall classification accuracy just slightly lower (~0.13% onaverage) than weighted scoring-based algorithms. Therefore, werecommended the weighted scoring-based compositing algorithms forsmall area land cover mapping, and median compositing algorithm forthe land cover mapping of large area considering the balance betweencomputational complexity and classification accuracy. The findings of thisstudy provide insights into the performance difference between variouscompositing algorithms, and have potential uses for the selection ofpixel-based image compositing technique adopted for land covermapping based on Landsat time series data.

基于改进生成对抗网络的时序数据异常检测

针对时序数据复杂的时间相关性,以及现有异常检测模型存在准确性低、训练不稳定等问题,提出一种结合BiLSTM和WGAN-GP的无监督时序数据异常检测模型。使用BiLSTM作为生成器和判别器的基础网络来捕获时序数据的时间相关性;为保证训练过程的稳定性,使用Wasserstein距离取代原有的衡量方法,在判别器损失中加入梯度惩罚项;将重构损失与判别损失相结合定义异常函数,采用局部自适应阈值方法判别异常,提高异常检测的准确性。为验证模型性能,在涉及多个领域的5类数据集上进行实验,其结果表明,该模型相比于Arima、LSTM等模型具有最高的平均F1分数。

基于VMD-TCN-GRU模型的水质预测研究

为充分挖掘水质数据在短时震荡中的变化特征,提升预测模型的精度,提出一种基于VMD(变分模态分解)、TCN(卷积时间神经网络)及GRU(门控循环单元)组成的混合水质预测模型,采用VMD-TCN-GRU模型对汾河水库出水口高锰酸盐指数进行预测,并与此类研究中常见的SVR(支持向量回归)、LSTM(长短期记忆神经网络)、TCN和CNN-LSTM(卷积神经网络-长短期记忆神经网络)这4种模型预测结果对比表明:VMD-TCN-GRU模型能更好挖掘水质数据在短时震荡过程中的特征信息,提升水质预测精度;VMD-TCN-GRU模型的MAE(平均绝对误差)、RMSE(均方根误差)下降,R^(2)(确定系数)提高,其MAE、RMSE、R^(2)分别为0.0553、0.0717、0.9351;其预测性能优越,预测精度更高且拥有更强的泛化能力,可以应用于汾河水质预测。

基于时间卷积网络的机床齿轮箱轴承剩余寿命预测

基于深度神经网络的RUL预测模型结构比较复杂,不能很好地满足中长期预测任务的要求。为了更好地利用时间信息,设计一种基于时间卷积网络(TCN)的轴承RUL预测模型。以振动信号的频谱特征作为输入,利用因果膨胀卷积结构提取频域特征并捕获长期依赖,从而实现对轴承准确的RUL预测。为了进一步说明所提方法的优越性,将所提方法与卷积神经网络(CNN)、门控循环单元(GRU)进行了对比。结果表明:所提出的TCN模型的RUL预测精度优于其他现有方法,具有较高的精度。

环渤海地区空气质量时空变化特征及动态预测

基于环渤海地区2017—2021年各城市空气质量指数(AQI)、污染物浓度与社会经济数据,利用数理统计、克里金插值法对环渤海地区AQI与污染物浓度的时空变化特征进行分析,运用皮尔逊相关性分析方法探讨AQI与污染物浓度、社会经济因素的相关关系,采用时间序列预测模型对2022年6月—2023年12月空气质量及污染物浓度进行预测。结果表明:环渤海地区AQI及污染物浓度大致呈逐年降低的趋势。AQI的逐月变化呈“W”形,O_(3)浓度的年内变化呈倒“V”形,其余污染物则呈现与O_(3)相反的变化趋势。AQI大致呈现西南高、东北低的空间分布特点,而污染物浓度分布具有明显的空间差异。环渤海地区5个代表性城市的AQI类别以良好为主,冬季首要污染物主要为PM_(2.5)、PM10,夏季首要污染物以O_(3)为主。人口数量是影响AQI的主要因素,城市园林绿地面积对AQI具有一定影响。预测结果显示,未来环渤海地区AQI、主要污染物浓度(O_(3)除外)均呈现出随时间的推移逐渐下降的变化趋势。

基于深度学习的公交行驶轨迹预测研究综述

公交行驶轨迹预测是对公交车到达线路上的重要轨迹点,如站点和道路交叉口等,进行到达时间预测。准确预测公交车到达站点和道路交叉口的时间,可以提高城市公交系统的运行效率和服务质量,对于城市公共交通规划和公交调度至关重要。从公交行驶轨迹预测方法的发展现状入手,分析了影响公交运行的相关因素,归纳并探讨了不同类型的相关数据集以及数据预处理方法。依照其发展脉络将公交行驶轨迹预测方法分为基于历史数据的模型、以时间序列模型为代表的参数模型以及包括机器学习和深度学习方法的非参数模型三大类,并总结分析了不同方法的优势和局限性。由于基于深度学习的相关模型在时间序列预测任务中表现出了优越性能,因此越来越多的学者开始采用基于深度学习的模型来解决公交行驶轨迹预测问题,同时考虑将城市道路所展现的空间特征与时间特征相结合以进一步提高预测精度。最后,阐述了公交行驶轨迹预测研究领域中面临的挑战,并对该领域未来的发展进行总结分析与趋势展望。

面向多维时间序列异常检测的时空图卷积网络

针对现有多维时间序列异常检测模型对局部和全局时空依赖性捕获能力不足的问题,提出一种基于时空图卷积网络的多维时间序列异常检测模型。首先,在时间维度上利用扩张因果卷积和多头自注意力机制,分别捕获短期和长期时间依赖性,并且引入通道注意力来学习不同通道的重要性权重;其次,在空间维度上利用静态图学习层根据节点嵌入构建静态图邻接矩阵,旨在捕获多维时间序列数据的全局空间依赖性,同时利用动态图学习层构建一系列演化的图邻接矩阵,旨在建模局部动态的空间依赖性;最后,联合优化重构模型和预测模型,通过重构误差和预测误差计算异常分数,然后比较阈值和异常分数的关系,进而检测异常。在MSL、SMAP和SWaT三个公开数据集上的实验结果表明,该模型在异常检测性能指标F1分数方面优于OmniAnomaly、MTAD-GAT和GDN等相关的基线模型。

基于动态自适应时空图的多元时序预测模型

深度学习模型在多元时间序列预测、智能驾驶、图像识别等多个领域广泛应用,其中多元时间序列预测是学者们关注的重点之一,多元时间序列预测是典型的回归任务,旨在通过海量的历史数据构建模型以预测未来状态,被广泛运用于交通、电力、金融等领域.多元时间序列数据具有复杂的时空依赖性,现有模型大多仅能捕获序列数据中的时间特征,难以捕获空间特征,而图神经网络解决了这一问题.图神经网络能够自然地建模实体间的复杂关系,可以很好地处理拓扑数据,而多元时序数据大多可以构造为拓扑图,因此图神经网络可以很好地学习多元时序数据中的空间特征.基于图神经网络的多元时间序列预测模型受到广泛关注并取得了一定的成果,但现有基于图神经网络的模型仍存在诸多不足.首先,现有方法大多分别捕获和建模多元时间序列数据中的空间特性和时间特性,未充分考虑多元时间序列的时空统一性,导致模型的次优建模;其次,现有方法主要基于静态预定义图或动态自适应图,其中静态预定义图通常根据监测节点之间的空间相关性进行构造且不会随着时间而改变,基于预定义图的研究忽略了时间序列数据中的时间特征,即忽略了数据模式随时间发生的改变;而自适应图通常由模型自主学习并不包含监测节点间的固有属性,基于自适应图的研究忽略了大量有效的领域知识,如道路的连通性和道路间的属性.为了解决上述问题,提出基于动态自适应时空图的多元时序预测模型MTP-Graph(Multivariate Time series Prediction model based on dynamic adaptive spatio-temporal Graph),利用时空融合模块将时空信息进行统一处理,避免了分开捕获时间特性与空间特性而导致的次优建模问题,提出图结合模块将静态预定义图和动态自适应图进行动态融合,获取时空信息的同时充分考虑领域知识,使模型可以更好地学习多元时间序列中的时空特性.在PeMSD3、PeMSD7和PeMSD8数据集上的大量实验结果表明,MTP-Graph预测性能优于其他基准方法,验证了MTP-Graph的可用性和有效性.

基于改进TCN的上扣扭矩序列数据分类

在油气开发领域,油套管安装后的密封性能检测尤为重要。其中,上扣过程中产生的扭矩序列数据可以作为油套管密封性的评判依据,用来判断上扣是否合格。为了利用上扣扭矩序列数据信息进行油套管密封性的识别分类,首先基于TCN网络模型结构,再融入位置编码机制和自注意力机制,搭建了一种新的网络模型,即PSE-TCN网络。通过比较不同策略下的结果准确率,展示了模型学习的过程,通过与其他网络模型进行对比,验证了本方法的有效性。实验结果表明,PSE-TCN相较于其他经典网络模型和一些改进后的TCN网络模型,扭矩序列识别精度有较大提升,在自制UCR_whorl数据集上,模型识别准确率达到93.41%。

全国森林火灾形势及其时空聚集性特征分析

为进一步研究全国森林火灾发生的时间规律与空间聚集性特征,基于全国2004—2022年度、2015—2021年月度的森林火灾发生次数及2017年31个省(自治区、直辖市)水文地理条件数据,利用Prophet时间序列模型与模糊聚类分析法对森林火灾特征进行分析。结果表明:Prophet时间序列模型在森林火灾预测领域具有良好的适用性;时间规律方面,2004—2022年森林火灾次数呈先下降后上升再下降趋势,且季节因素影响突出,节假日影响因素不显著;空间聚集性方面,31个省(自治区、直辖市)可大致分为7个类别,其分类基本与地域特点有关。研究结果可为挖掘森林火灾时空分布及聚集性特征提供参考。