Multivariate Time Series Anomaly Detection Based on Spatial-Temporal Network and Transformer in Industrial Internet of Things

In the Industrial Internet of Things(IIoT),sensors generate time series data to reflect the working state.When the systems are attacked,timely identification of outliers in time series is critical to ensure security.Although many anomaly detection methods have been proposed,the temporal correlation of the time series over the same sensor and the state(spatial)correlation between different sensors are rarely considered simultaneously in these methods.Owing to the superior capability of Transformer in learning time series features.This paper proposes a time series anomaly detection method based on a spatial-temporal network and an improved Transformer.Additionally,the methods based on graph neural networks typically include a graph structure learning module and an anomaly detection module,which are interdependent.However,in the initial phase of training,since neither of the modules has reached an optimal state,their performance may influence each other.This scenario makes the end-to-end training approach hard to effectively direct the learning trajectory of each module.This interdependence between the modules,coupled with the initial instability,may cause the model to find it hard to find the optimal solution during the training process,resulting in unsatisfactory results.We introduce an adaptive graph structure learning method to obtain the optimal model parameters and graph structure.Experiments on two publicly available datasets demonstrate that the proposed method attains higher anomaly detection results than other methods.

Comprehensive evaluation and spatial-temporal evolution characteristics of urban resilience in Chengdu-Chongqing Economic Circle

To clarify the connotations and extensions of urban resilience,this study focuses on the Chengdu-Chongqing Economic Circle with 16 cities as research subjects.A comprehensive evaluation index system was constructed to measure the resilience of each city from 2003 to 2020.The spatial-temporal evolution characteristics were analyzed using Kernel density estimation,standard deviation ellipse,and spatial Markov chain analysis,and the spatial Tobit model was introduced to discover the influencing factors.The results indicate the following:①Urban resilience in the Chengdu-Chongqing Economic Circle displays an upward trend,with the center of gravity moving to the southwest,and the polarization phenomenon intensifying.②The urban resilience level in a region has certain spatial and geographical dependence,while the probability of urban resilience transfer differs in adjacent cities with different resilience levels.③Urban centrality,economic scale,openness level,and financial development promote urban resilience,whereas government scale significantly inhibits it.Finally,this paper proposes countermeasures and suggestions to improve the urban resilience of the Chengdu-Chongqing Economic Circle.

AFSTGCN:Prediction for multivariate time series using an adaptive fused spatial-temporal graph convolutional network

The prediction for Multivariate Time Series(MTS)explores the interrelationships among variables at historical moments,extracts their relevant characteristics,and is widely used in finance,weather,complex industries and other fields.Furthermore,it is important to construct a digital twin system.However,existing methods do not take full advantage of the potential properties of variables,which results in poor predicted accuracy.In this paper,we propose the Adaptive Fused Spatial-Temporal Graph Convolutional Network(AFSTGCN).First,to address the problem of the unknown spatial-temporal structure,we construct the Adaptive Fused Spatial-Temporal Graph(AFSTG)layer.Specifically,we fuse the spatial-temporal graph based on the interrelationship of spatial graphs.Simultaneously,we construct the adaptive adjacency matrix of the spatial-temporal graph using node embedding methods.Subsequently,to overcome the insufficient extraction of disordered correlation features,we construct the Adaptive Fused Spatial-Temporal Graph Convolutional(AFSTGC)module.The module forces the reordering of disordered temporal,spatial and spatial-temporal dependencies into rule-like data.AFSTGCN dynamically and synchronously acquires potential temporal,spatial and spatial-temporal correlations,thereby fully extracting rich hierarchical feature information to enhance the predicted accuracy.Experiments on different types of MTS datasets demonstrate that the model achieves state-of-the-art single-step and multi-step performance compared with eight other deep learning models.

Spatial-temporal distribution and geochemistry of highly evolved Mesozoic granites in Great Xing’an Range,NE China:Discriminant criteria and geological significance

Highly evolved granite is an important sign of the mature continent crust and closely associated with deposits of rare metals.In this work,the authors undertake systematically zircon U-Pb ages and whole rock elemental data for highly evolved granitic intrusions from the Great Xing’an Range(GXR),NE China,to elucidate their discriminant criteria,spatial-temporal distribution,differentiation and geodynamic mecha-nism.Geochemical data of these highly evolved granites suggest that high w(SiO_(2))(>70%)and differentiation index(DI>88)could be quantified indicators,while strong Eu depletion,high TE_(1,3),lowΣREE and low Zr/Hf,Nb/Ta,K/Rb could only be qualitative indicators.Zircon U-Pb ages suggest that the highly evolved gran-ites in the GXR were mainly formed in Late Mesozoic,which can be divided into two major stages:Late Ju-rassic-early Early Cretaceous(162-136 Ma,peak at 138 Ma),and late Early Cretaceous(136-106 Ma,peak at 126 Ma).The highly evolved granites are mainly distributed in the central-southern GXR,and display a weakly trend of getting younger from northwest to southeast,meanwhile indicating the metallogenic potential of rare metals within the central GXR.The spatial-temporal distribution,combined with regional geological data,indicates the highly evolved Mesozoic granites in the GXR were emplaced in an extensional environ-ment,of which the Late Jurassic-early Early Cretaceous extension was related to the closure of the Mongol-Okhotsk Ocean and roll-back of the Paleo-Pacific Plate,while the late Early Cretaceous extension was mainly related to the roll-back of the Paleo-Pacific Plate.

Spatial-temporal Divergence Characteristics and Driving Factors of Green Economic Efficiency in the Yangtze River Economic Belt of China

The spatial and temporal variation of green economic efficiency and its driving factors are of great significance for the con-struction of high-efficiency and low-consumption green development model and sustainable socio-economic development.The research focused on the Yangtze River Economic Belt(YREB)and employed the miniumum distance to strong efficient frontier DEA(MinDs)model to measure the green economic efficiency of the municipalities in the region between 2008 and 2020.Then,the spatial autocorrel-ation model was used to analyze the evolution characteristics of its spatial pattern.Finally,Geodetector was applied to reveal the drivers and their interactions on green economic efficiency.It is found that:1)the overall green economic efficiency of the YREB from 2008 to 2020 shows a W-shaped fluctuating upward trend,green economic efficiency is greater in the downstream and smallest in the upstream;2)the spatial distribution of green economic efficiency shows clustering characteristics,with multi-core clustering based on‘city clusters-central cities'becoming more obvious over time;the High-High agglomeration type is mainly clustered in Jiangsu and Zheji-ang,while the Low-Low agglomeration type is clustered in the western Sichuan Plateau area and southwestern Yunnan;3)from input-output factors,whether it is the YREB as a whole or the upper,middle and lower reaches regions,the economic development level,labor input,and capital investment are the leading factors in the spatial-temporal evolution of green economic efficiency,with the com-prehensive influence of economic development level and pollution index being the most important interactive driving factor;4)from so-cio-economic factors,information technology drivers such as government intervention,transportation accessibility,information infra-structure,and Internet penetration are always high impact influencers and dominant interaction factors for green economic efficiency in the YREB and the three major regions in the upper,middle and lower reaches.Accordingly,the article puts forward relevant policy re-commendations in terms of formulating differentiated green transformation strategies,strengthening network leadership and informa-tion technology construction and coordinating multi-factor integrated development,which could provide useful reference for promoting synergistic green economic efficiency in the YREB.

Adaptive spatial-temporal graph attention network for traffic speed prediction

Considering the nonlinear structure and spatial-temporal correlation of traffic network,and the influence of potential correlation between nodes of traffic network on the spatial features,this paper proposes a traffic speed prediction model based on the combination of graph attention network with self-adaptive adjacency matrix(SAdpGAT)and bidirectional gated recurrent unit(BiGRU).First-ly,the model introduces graph attention network(GAT)to extract the spatial features of real road network and potential road network respectively in spatial dimension.Secondly,the spatial features are input into BiGRU to extract the time series features.Finally,the prediction results of the real road network and the potential road network are connected to generate the final prediction results of the model.The experimental results show that the prediction accuracy of the proposed model is im-proved obviously on METR-LA and PEMS-BAY datasets,which proves the advantages of the pro-posed spatial-temporal model in traffic speed prediction.

Spatial-temporal Variation Characteristics of Water Quality in the Lower Reaches of the Nenjiang River

As an important river in the western part of Jilin Province,the lower reach of the Nenjiang River is an important wetland water source conservation area in Jilin Province.Within the watershed,it governs the Momoge Wetland,the Xianghai Wetland,and the Danjiang Wetland in Jilin Province.The main problem in the lower reaches of the Nenjiang River is the uneven distribution of water resources in time and space,and the intensification of land salinization.Zhenlai County and Da an City in the Nenjiang River Basin have sufficient surface water resources,with surface water as the drinking water source.Baicheng City and Tongyu County have scarce surface water resources,and both use groundwater as their domestic water source.The main polluted section in the basin is the Xianghai Reservoir,and the annual water quality evaluation is Class V.However,the water quality of the Tao er River,the main stream of the Nenjiang River,is significantly better than that of the Xianghai Reservoir.In order to better study the water environmental pollution situation in the Nenjiang River basin,monitoring data from five sections of non seasonal rivers in the basin from 2012 to 2021 were selected for studying water quality.This in-depth exploration of the water pollution status and river water quality change trends in the Nenjiang River basin is of great significance for future rural development,agricultural pattern transformation,and the promotion of water ecological civilization construction.

基于Kriging插值代理模型的轴流式止回阀多目标优化

基于Kriging插值代理模型,对轴流式止回阀的各项性能进行综合优化,具体的优化目标为:降低正向流阻;减小阀芯振动;提升止回性能。通过试验设计方案找到影响轴流式止回阀正向流阻的主要结构参数因素,并以其作为优化对象,通过Kriging插值法拟合主要结构参数对应的性能样本点,在保证拟合精度的情况下,使用NSGA-II遗传算法同时对3个主要结构参数进行多目标优化,得到了三目标Pareto前沿,经过权衡不同因素对性能指标的影响大小,最终得到了综合性能最优的结构参数为:阀瓣丰满度系数α_(1)=2.199,阀芯长度L=386.322 mm,阀芯半径R=113.997 mm。其对应的止回阀性能为:正向出口流量Q=161.839 kg/s,正向出口流速v=3.30235 m/s,止回阀进口压力p=97632.2287 Pa。相比优化之前,阀门正向流阻减小了1.2%,阀芯振动降低了0.24%,止回性能提升了2.3%。

基于Kriging的多阶段加点策略及离心叶轮汽蚀优化

为了实现燃油离心泵汽蚀特性智能优化设计,提出一种基于Kriging的多阶段主动学习加点策略。重点是建立3个阶段的加点策略,并明确各阶段加点策略的切换判据。基于一维函数、多维函数等测试函数,与单加点策略对比以验证多阶段主动学习加点策略的有效性。进而,以某型燃油离心泵为例,完成所提出加点策略在其汽蚀特性的优化应用。测试函数算例结果表明:多阶段主动学习加点策略达到相同精度所需时间稍长于单MSE策略,优于单EI策略与单CV策略;但相比单MSE策略,全局最优解局部精度更高。离心泵汽蚀特性优化结果表明:优化泵进口处压力损失降低且进口处回流、旋涡等局部流动减弱,蜗壳内压力梯度增大且其出口处旋涡流动减弱;优化泵的汽蚀系数提高30.83%,汽蚀余量下降9.14%,汽蚀比转数下降7.17%,泵内气体含量下降17.27%,所提出的方法有效地提高了离心泵的抗汽蚀性能。

基于多峰优化Kriging模型与距离相关系数的高速列车动力学参数多输出灵敏度分析

高速列车动力学参数较多,有多个动力学性能的评价指标。综合考虑这些动力学指标进行参数灵敏度分析,对高速列车动力学参数的恰当匹配具有重要意义。现有的单输出灵敏度分析技术难以准确度量参数对车辆动力学综合性能的影响。为避免这一缺陷,本文引入了一种新的基于距离相关系数的多输出灵敏度分析技术。该方法通过求解单个输入与多个输出变量之间的距离相关系数,实现多输出全局灵敏度分析的目的。为提高灵敏度分析的效率,建立了多个动力学指标与输入参数的Kriging模型;为提高近似模型的精度,引入了一种新颖的多峰优化算法对Kriging模型的超参数进行全局寻优。与传统单目标智能优化算法不同,该优化算法通过引入种群差异指标,将单目标优化问题转化为双目标优化问题,从而增加了子代种群的多样性,避免了传统优化算法易于陷入局部最优解的问题。基于所提方法,以和谐号动车组动车为例,研究了多个动力学指标的多输出全局灵敏度。结果表明:由于增强了种群多样性,所提的超参数优化方法对Kriging模型精度和稳定性提升效果明显;基于距离相关系数的灵敏度分析方法能够更加合理地识别出对车辆动力学综合性能影响较大的参数,排除几乎没有影响的参数。

应用时空复合趋势面Kriging方法插值地表沉降

通过布设监测点可反映地表沉降变化趋势,确保施工和运营期间安全,但沉降监测点数量有限,且易受施工作业、天气等因素干扰导致数据缺失或污染,因此需对监测区域数据进行插值处理。针对采用传统地统计插值方法估计地铁地表沉降趋势精度较低的问题,提出了一种基于时空复合趋势面的Kriging插值方法。以乌鲁木齐地铁3号线2017年6月24日—9月3日共15期地铁地表沉降观测值为实验数据,分别采用两种方法对沉降监测点缺失数据进行插补。结果表明,相较于传统方法,基于时空复合趋势面的Kriging插值方法的拟合精度更高,RMSE可降低约35%。

基于Kriging代理模型的无刷双馈发电机转子优化设计

无刷双馈发电机由于其自身没有电刷、滑环等结构特点,因此被认为是最有希望成为风力发电系统的主流发电机之一。无刷双馈发电机是依靠转子的磁场调制机理实现机电能量转换,其转子的耦合能力是影响电机性能的重要因素。因此本文提出了一种串联笼条辅助磁障转子的新型转子结构,分析了该种转子不同结构参数对电机性能的影响,并提出一种基于Kriging代理模型的多目标差分进化算法的无刷双馈发电机转子优化设计方法。首先建立该种新型转子的参数化模型,通过Kriging代理模型建立优化变量与优化目标之间的响应关系,再利用多目标差分进化算法找到最优的变量组合,最后通过有限元仿真对电机的性能进行分析。

基于Kriging模型的多次尝试差分进化贝叶斯有限元模型修正

标准差分进化自适应Metropolis(differential evolution adaptive Metropolis,DREAM)算法需进行多条马氏链并行计算,存在收敛效率低和计算成本高的问题。为此,提出一种基于Kriging模型的多次尝试差分进化贝叶斯有限元模型修正(multiple-try differential evolution adaptive Metropolis with“ZS”,MT-DREAM(ZS))框架。该框架在DREAM的基础上引入历史向量差分采样、斯诺克更新以及多次尝试Metropolis抽样,并利用Kriging模型代替有限元模型进行随机抽样,实现利用极少数并行链便可快速探索多维修正参数后验分布。利用固结钢板梁模型试验,比较了DREAM和MT-DREAM(ZS)的修正效果。结果表明:MT-DREAM(ZS)可实现马尔科夫链的快速收敛,其收敛效率较DREAM提升了3.42倍,且修正结果精度和稳定性有提升;Kriging模型可大幅度降低计算成本。所提框架为解决多参数不确定模型修正中的收敛效率低和计算成本高等问题提供了一种新思路。

基于Kriging模型与FORM的钢板组合梁桥结构优化设计

考虑材料参数的不确定性以及设计变量多重约束条件的影响,提出了基于可靠性分析的钢板组合梁桥结构优化设计方法。首先,利用拉丁超立方试验设计(Latin hypercube design,LHD)构建试验组合方案,并通过有限元分析获取试验设计组合下的最大挠度响应值。基于获取的训练样本,构建能够表征结构最大挠度值与随机输入之间映射关系的Kriging代理模型。基于建立的代理模型,以钢结构截面积最小化为目标函数,建立符合规范要求与可靠度约束的优化数学模型,并采用一阶可靠性方法(first order reliability method,FORM),实现设计变量优化。通过比较具有相同设计变量上限,但下限分别为初始值的80%和70%两种优化设计方案,评估了两者在相同阈值以及不同阈值下设计变量的优化结果。最后对方案二在阈值为17 mm下的优化结果进行极限状态验算。结果表明:两个优化设计方案在阈值为17 mm下的优化结果更具合理性,并且方案二在同一阈值下具有更高的优化程度。随着阈值增加,优化程度增加,且腹板高度在较大阈值下对结果影响更大,而翼缘宽度在较小阈值下影响更大。

基于PSO-Kriging算法的三维地质建模技术研究

三维地质模型的构建对于理解和预测地下结构至关重要。地质钻孔数据能够反映岩体空间分布和地质构造特征,本研究以小保当一号煤矿11盘区内的23个地质钻孔数据为基础,采用添加虚拟地层的方法解决了地层缺失与地层重复现象,构建共计27层地层的三维地质模型以及二维剖面模型。此外,针对传统的克里金方法在处理复杂地质数据参数选择困难的问题,采用粒子群算法对传统克里金插值方法中的块金值(C 0)、偏基台值(C)和变程(a)三个关键参数进行寻优,从而克服普通克里金插值参数选择的主观性和不确定性,采用实际验证法选取了研究区内四个钻孔来对比插值结果,结果表明经过PSO优化的Kriging算法在X3-1、X3-2、K3-4、K3-5四个钻孔的RMSE值分别降低至1.184、1.267、1.606、1.560,相比于Kriging的RMSE平均降低了31%,且PSO-Kriging算法在四个钻孔处对2-2煤层的插值结果与实际值相比较误差分别为1.00 m、0.01 m、0.11 m和0.03 m,比Kriging插值结果更接近实际值,表明了所提方法的有效性和优越性。

一种基于自适应Kriging集成模型的结构可靠性分析方法

基于Kriging模型的复杂结构可靠性分析结果高度依赖于Kriging模型的拟合精度,在构建Kriging模型的过程中,不同相关函数和回归函数的选择均会影响模型精度。为解决模型的不确定性对可靠性分析结果的影响,同时兼顾计算效率和精度,基于Kriging模型和蒙特卡罗模拟(MCS)方法,提出了一种结合自适应集成策略和主动学习函数的结构可靠度计算方法。该方法考虑Kriging模型的建模不确定性,将多种Kriging模型组合,构建了一种综合考虑样本点贡献和样本点距离的主动学习函数,通过主动学习函数迭代更新集成Kriging模型直至满足收敛条件,最后通过构建的集成Kriging模型和MCS方法进行结构可靠性分析。数值算例和工程算例结果验证了所提方法的有效性,该方法与其他主要方法相比稳健性更好,在保证计算精度的同时,计算效率更高。

基于主动学习Kriging模型的改进一次可靠度方法

结构可靠度分析的一次可靠度方法在每一迭代中均涉及迭代点的函数值及梯度值计算,但后续迭代过程不能充分利用前期迭代过程中迭代点的计算结果,因此计算效率有待于进一步提升。考虑到迭代后期迭代点在局部区域内波动,若以已有迭代结果为基础建立代理模型进行迭代后期迭代点的函数值及梯度值计算将有助于改善一次可靠度方法的计算效率。为此,该文在将一次可靠度方法的迭代过程分为全局搜索阶段与局部搜索阶段的基础上,针对两个阶段分别采用不同的计算策略,即全局搜索阶段沿用已有一次可靠度方法的迭代过程,在局部搜索阶段则基于全局搜索阶段的迭代结果建立Kriging模型,并引入可评估Kriging模型在迭代点处精度的学习函数,实现局部搜索阶段迭代点的高效率计算,从而提出了具有更高计算效率的改进一次可靠度方法。数值算例和工程算例的计算结果表明建议方法在保持精度不变的情况下,可显著提高一次可靠度方法的计算效率。

基于自适应Kriging模型的双螺旋轴可靠性分析

双螺旋轴是高速水稻插秧机的关键部分,其回转槽面在强轴向力作用下容易破坏失效,因此对双螺旋轴进行可靠性分析十分重要.针对传统方法获得其可靠度耗资大、时间长的问题,采用自适应Kriging模型结合Monte Carlo模拟的可靠性分析方法(AK-MCS)对双螺旋轴进行可靠性分析,以减少计算成本、提高计算效率.根据双螺旋轴的工况条件,考虑材料密度、工作载荷、工作温度3个参数对其所受挤压应力的影响,进行可靠性分析.分析结果表明:在许用应力值为613.21 MPa的条件下,计算得出双螺旋轴的可靠度为0.9986.与单一的Monte Carlo模拟法进行对比,本文提出的评估方法在仿真效率上有明显优势,为双螺旋轴可靠性分析提供了一种思路.

基于Kriging模型的船舶主机隔离系统抗冲击优化

为了提高系统的抗冲击性能和优化效率,本文以限位隔离系统的刚度阻尼等7个参数为设计变量,以系统冲击响应中的相对位移响应和绝对加速度响最大峰值为响应值建立Kriging代理模型。通过NSGA-II多目标优化算法对构建的代理模型进行多目标优化,优化目标为相对位移响应和绝对加速度响最大峰值两者和的最小值。与初始方案相比,优化后系统相对位移响应降低了27.139%,绝对加速度响应降低了37.846%。结果表明:建立的代理模型对隔离系统的冲击响应有较精确的预报作用,优化后船舶主机隔离系统的抗冲击得到明显提高。

Spatial-temporal difference between nitrate in groundwater and nitrogen in soil based on geostatistical analysis

The study of temporal and spatial variations of nitrate in groundwater under different soil nitrogen environments is helpful to the security of groundwater resources in agricultural areas.In this paper,based on 320 groups of soil and groundwater samples collected at the same time,geostatistical analysis and multiple regression analysis were comprehensively used to conduct the evaluation of nitrogen contents in both groundwater and soil.From May to August,as the nitrification of groundwater is dominant,the average concentration of nitrate nitrogen is 34.80 mg/L;The variation of soil ammonia nitrogen and nitrate nitrogen is moderate from May to July,and the variation coefficient decreased sharply and then increased in August.There is a high correlation between the nitrate nitrogen in groundwater and soil in July,and there is a high correlation between the nitrate nitrogen in groundwater and ammonium nitrogen in soil in August and nitrate nitrogen in soil in July.From May to August,the area of low groundwater nitrate nitrogen in 0-5 mg/L and 5-10 mg/L decreased from 10.97%to 0,and the proportion of high-value area(greater than 70 mg/L)increased from 21.19%to 27.29%.Nitrate nitrogen is the main factor affecting the quality of groundwater.The correlation analysis of nitrate nitrogen in groundwater,nitrate nitrogen in soil and ammonium nitrogen shows that they have a certain period of delay.The areas with high concentration of nitrate in groundwater are mainly concentrated in the western part of the study area,which has a high consistency with the high value areas of soil nitrate distribution from July to August,and a high difference with the spatial position of soil ammonia nitrogen distribution in August.