FP-STE: A Novel Node Failure Prediction Method Based on Spatio-Temporal Feature Extraction in Data Centers

The development of cloud computing and virtualization technology has brought great challenges to the reliability of data center services.Data centers typically contain a large number of compute and storage nodes which may fail and affect the quality of service.Failure prediction is an important means of ensuring service availability.Predicting node failure in cloud-based data centers is challenging because the failure symptoms reflected have complex characteristics,and the distribution imbalance between the failure sample and the normal sample is widespread,resulting in inaccurate failure prediction.Targeting these challenges,this paper proposes a novel failure prediction method FP-STE(Failure Prediction based on Spatio-temporal Feature Extraction).Firstly,an improved recurrent neural network HW-GRU(Improved GRU based on HighWay network)and a convolutional neural network CNN are used to extract the temporal features and spatial features of multivariate data respectively to increase the discrimination of different types of failure symptoms which improves the accuracy of prediction.Then the intermediate results of the two models are added as features into SCSXGBoost to predict the possibility and the precise type of node failure in the future.SCS-XGBoost is an ensemble learning model that is improved by the integrated strategy of oversampling and cost-sensitive learning.Experimental results based on real data sets confirm the effectiveness and superiority of FP-STE.

Features of seismicity in Xinjiang and its possible reason after the Yutian M_S7.4 earthquake,2008

The paper discusses quantitatively the influence of the Yutian Ms7.4 earthquake of March 21, 2008 and Wuqia Ms6.9 earthquake of October 5, 2008 on regional seismicity in Xinjiang, and explains primarily the possible reason of earthquake activity feature in Xinjiang after the Yutian Ms7.4 earthquake by analyzing the static Coulomb failure stress change produced by the Yutian Ms7.4 earthquake and Wuqia Ms6.9 earth-quake, and the seismicity feature of Ms≥3 earthquakes in the positive Coulomb stress change region of Kashi-Wuqia joint region, the central segment of Tianshan Mountain and Kalpin block. The result shows that the Yutian Ms7.4 earthquake of March 21, 2008, may encourage the Wuqia Ms6.9 earth-quake of October 5, 2008, and the Yutian Ms7.4 earthquake and Wuqia Ms6.9 earthquake may change the seismicity state in the central segment of Tianshan Mountain, Kalpin block and Kashi-Wuqia joint region, and encourage the subsequent Ms≥3 earthquakes.

Clinical characteristics and risk factors of 27 liver failure patients complicated by invasive fungal infections

To investigate the clinical feature, risk factors and outcome of treatment in patients with liver failure complicated by invasive fungal infections. Retro-spective analysis of the clinical data and related factors of 27 patients with liver failure com-plicated by invasive fungal infections was per-formed. These patients were admitted from January 2007 to August 2009 in our department. Among them, Candida albicans accounted for 17cases (54.84%), albicans tropicals for 4 cases (12.90%). Fungal infection in respiratory tract and alimentary tract accounted for 58.06% and 11% respectively. 81.25% of them had fever fluctuating from 37.4oC to 40oC. 81.25% had elevated white blood cell counts .All had the usage of broad-spectrum of antibiotics, whereas some of them used corticosteroids and had invasive medical manipulation for the treatment. Most patients deteriorated after invasive fungal infections. 21 cases accepted with the treatment of antifungal drugs and mortality rate was 63.00%. It was found that the invasive fungal infection possibility of patients with liver failure significantly increased. To prevent the occurrence of invasive fungal infection, promptly early treatment of liver failure, proper use of antibiotics, cautious use or disuse of corticosteroids, reduction of invasive medical manipulation should be well done. Early detection and treatment of fungal infection are vital to decrease in mortality rate.

Nonlinearly correlated failure analysis and autonomic prediction for distributed systems

In order to achieve failure prediction without manual intervention for distributed systems, a novel failure feature analysis and extraction approach to automate failure prediction is proposed. Compared with the traditional methods which focus on building heuristic rules or models, the autonomic prediction approach analyzes the nonlinear correlation of failure features by recognizing failure patterns. Failure data are sorted according to the nonlinear correlation and failure signature is proposed for autonomic prediction. In addition, the Manifold Learning algorithm named supervised locally linear embedding is applied to achieve feature extraction. Based on the runtime monitoring of failure metrics, the experimental results indicate that the proposed method has better performance in terms of both correlation recognition precision and feature extraction quality and thus it can be used to design efficient autonomic failure prediction for distributed systems.

Analysis of 610 cases inpatients with chronic heart failure

Objective： To analyze the clinical characteristics and medical treatment of hospitalized patients with chronic heartfailure, and provide medication for the patients. Methods： According to medical records of inpatients, demographicinformation, etiology, clinical features and treatment information were collected. Results： A total of 610 cases withheart failure from our hospital between July 2010 and June 2016 were analyzed. The average age of all the patientswas 63. Males accounted for 50.49%. There were 82.7% patients with NYHA functional classification at III-IV.31.3% patients with coronary artery disease, 28.4% with rheumatic heart diseases and 21.8% with expansion ofheart disease. 27.4 percent of the patients with the left ventricular ejection fraction lower than 40%. There were78.9% patients received aldosterone antagonist treatment, 63.4% received ACEI/ARB, 62.1% received digoxin,59.8% received beta blockers, and 75.4% patients received decoction, and 80.8% received Chinese patent medicine.Conclusion： The study has shown that high utilization of traditional Chinese medicine is to highlight theadvantages of the combination of traditional Chinese and western medicine in the treatment of heart failure. Theusage of spironolactone, ACEI/ARB, beta blockers in our hospital was lower than the ratio reported, higher thanthat of the national average. However, the use of diureticis is lower than that of national average, which may relateto the use of damp-clearing herbs. Clinicians should pay attention to the high utilization rate of digoxin. Patientswith the left ventricular ejection fraction 〉 50% accounted for 61.1%, which required more attention should be paidto the diagnosis and treatment of patients with heart failure of the ejection fraction remains.

Feature of resistivity response of slope from steady to unsteady

Using resistivity as index and referring to the law about effect of slope to resistivity,the apparent resistivities of geophysical model concerned with unsteady rock type slope failure were calculated systematically by using the boundary integral equation method.After studying the feature of resistivity response of slope failure,the variety of resistivity during evolution of slope from steady to unsteady was found and the characteristics of resistivity response about slope failure was concluded.These make electrical exploring method for detecting the slip plane or structural plane of slope failure,evaluating the stability of the slope,and forecasting slope failure become true.

土工格栅加筋粉煤灰改良黄土力学特性研究

随着西部黄土地区城镇化进程的加快,填挖结合的新增用地方式要求地基处理技术需要重点解决填方区的强度与变形问题。论文将粉煤灰作为改性填料对黄土进行改良,通过直剪试验、三轴试验、湿陷试验与扫描电镜试验,分析不同粉煤灰掺入比条件下加筋与未加筋黄土试样的强度变化规律、宏观破坏特征和微结构特征。结果表明:采用粉煤灰改良和土工格栅加筋后,黄土的最优含水率增大,最大干密度降低。当粉煤灰掺入比λ=20%时,加筋改良效果最好,黄土的峰值强度、残余强度和等效内摩擦角均有大幅提高,侧向变形显著减小,且能有效降低其湿陷性。微结构分析表明:粉煤灰对黄土的改良主要体现在颗粒直接填充和化学结晶体交织成的网状填充两种类型的强化作用,改良后黄土的孔隙率与孔隙尺寸均有所下降。研究结果可为黄土地区填方边坡与填方路基的改良加固提供参考。

基于DGLPP-SVDD算法的化工过程故障检测

为解决传统全局局部保留投影算法(GLPP)不能充分利用已有故障数据进行特征提取的缺点,提出了判别全局局部保留投影算法(DGLPP)。在数据降维处理后,为应对高斯和非高斯混合分布的过程数据特性,通过支持向量数据描述算法(SVDD)构建故障检测统计量。将两种算法相结合提出基于DGLPP-SVDD的故障检测方法。将DGLPP-SVDD算法应用于TE过程仿真,并与GLPP算法对比,结果表明:DGLPP-SVDD算法具有更短的故障检测滞后时间和更高的故障检测率。

关家崖煤矿重复采动巷道变形特征及控制对策研究

针对重复采动巷道围岩变形严重、无法复用,重复采动巷道在服务期内具有明显的叠加演化特征的问题,以关家崖煤矿13092巷道为研究背景,采用现场实测、数值模拟和理论分析的方法,对重复采动巷道变形的叠加扩展特征和控制对策进行了研究。重复采动巷道变形特征分析结果表明:①一次回采扰动下,重复采动巷道变形呈现分区和非对称破坏特征,可划分为快速变形区、强烈变形区和缓慢变形区;裂纹破坏主要在煤壁帮和煤柱帮,而顶底板较少,表现为巷道两帮显著片帮和内移;煤壁帮与顶板、煤柱帮与底板交汇处变形严重。②二次采动巷道在一次破坏基础上叠加扩展,使得非对称破坏更加显著,形成巷道围岩蝶形叠加塑性破坏区。③重复采动巷道围岩控制的重点时间为一次回采阶段,重点区域为强烈变形区和缓慢变形区的巷道煤柱帮一侧。通过分析采动巷道蝶形变形特征和破坏分区规律,提出了重复采动巷道多层次耦合控制技术,采用浅低压−深高压注浆提高煤柱支撑力,采用锚索补强提高支护体支撑力,实现耦合控制。通过加固前后变形量对比分析验证了多层次耦合控制满足巷道复用要求。

基于深度学习的集群系统故障预测方法

在面对集群系统故障预测时,长时间序列预测中存在因关键特征信息丢失而导致梯度消失或爆炸问题,从而影响了故障预测模型的准确性。基于此,提出一种新的基于深度学习的集群系统故障预测方法。该方法采用双向门控循环网络(bidirectional gate recurrent unit, BiGRU)来捕捉局部时序特征,同时采用Transformer来提高全局特征提取能力。通过BiGRU层中双向的信息传递获得集群系统日志上时序特征的动态变化,以获取集群事件中的潜在因果关系和局部时间特征,使用Transformer层并行处理BiGRU层输出的时间序列,得到全局的时间依赖性,继而由全连接神经网络层得到预测结果。通过由Blue Gene/L系统产生的真实日志所构建的公共数据集来验证方法的有效性,结果表明,所提方法优于对比方法,其最佳正确率和F1值分别达到91.69%和92.74%。

供水管网故障预测的机器学习方法

近年来我国各地供水管网运行安全问题频发,造成了巨大的经济损失与严重的社会影响。因此,准确地预测管道故障,精准地维护管道,经济高效地保障供水管网的运行安全具有十分重要的意义。然而,目前关于前沿算法在管道故障预测方面的应用有待探索,并且关于系统性比较机器学习算法的研究尚不多见。为此,首先明确了供水管网故障预测问题,介绍了逻辑回归、随机森林、人工神经网络和一维卷积神经网络四种机器学习算法的基本原理。以我国南方某市工业园区供水管网为例,检验比较了四种模型的故障预测性能,结果显示一维卷积神经网络准确性最好但随机森林效率最高。同时分析了管道特征对故障概率的影响,管径、管长、道路等级与施工企业资质是最重要的四个特征,故障概率与管径、道路等级呈负相关,而与管长、施工企业资质呈正相关。

基于低压直流断路器的机械故障机理分析与特征提取技术研究

直流断路器是轨道交通直流牵引供电系统的核心保护设备,是直流供电系统安全运行的保证,确保其稳定运行至关重要。文中针对直流牵引断路器机械结构特点,进行了断路器机械故障模拟实验,选取分合闸线圈电流及振动信号作为监测物理量,研究其机械故障机理与特征提取方法。研制信号采集硬件装置,搭建直流断路器机械故障模拟实验平台,模拟6种常见的机械故障。采集不同故障下的线圈电流及振动信号进行特征提取后进行故障机理分析,为低压直流断路器机械故障诊断提供理论支撑。

发动机高速滚动轴承磨损故障信号特征识别

发动机高速滚动轴承在恶劣的工作环境下长时间使用,会导致其出现磨损故障,严重影响发动机的正常运行,且故障类型多样,每种故障类型都有其独特的特征,使得故障信号特征识别困难。为了有效解决这一问题,提出了一种发动机高速滚动轴承磨损故障信号特征识别方法。通过对发动机高速滚动轴承的振动信号进行EEMD分解和重建,获得其固有振动模态函数IMF,根据所得的IMF构建Hankel矩阵,获得拼接的奇异值样本特征矢量。通过划分样本空间和设置迭代阈值的方式,采用模糊聚类算法对故障样本聚类,计算出每个样本在不同聚类中的隶属度,以获得其贴近程度和故障特征。通过实验证明,所提算法能够较好的判别机械故障,准确性高,误报错报概率小,可确保设备的安全运行。

基于改进SqueezeNet算法的VBE设备电路板元件失效识别研究

在直流输电系统中,换流阀阀基电子VBE(valve base electronics)设备的稳定运作对维护直流系统安全至关重要。传统的阀基电子设备电路板(VBE板)元件失效检测方法依赖于耗时的人工检查或基于规则的自动化系统,这些方法通常检测效率低下且准确性有限。针对该问题,提出一种基于改进的SqueezeNet深度学习模型的VBE板元件失效区域识别方法。通过引入深度可分离卷积和残差连接,所提改进SqueezeNet模型旨在提高元件失效检测的准确性,同时降低计算资源的需求。在VBE板元件失效数据集上的实验结果表明,所提方法在元件失效检测准确率和运算效率方面均优于传统方法和标准SqueezeNet模型,准确率达到了95.27%,比原模型高出4.45%。不仅提升了VBE板元件失效检测的效率和准确性,而且为电力系统中类似设备的元件失效诊断提供了新的技术参考。

循环加卸载下大理岩渐进损伤的能量特性研究

为探究锦屏大理岩渐进损伤过程中的能量特性,开展了5种围压下岩样的三轴循环加卸载试验,分析了循环加卸载下大理岩渐进损伤的弹性模量演化、能量演化及分配规律、能量演化的围压效应、损伤演进特性,并系统剖析探讨了基于能耗比和能耗差定义的两种岩石强度失效的能量判据。研究结果表明:大理岩从初始屈服到进入残余阶段的整个变形破坏过程中,加载段弹性模量和卸载段弹性模量均呈现单调退化的弹塑性耦合特征,应避免采用弹性模量不变假定计算弹性应变能带来的偏差;围压升高不仅能够有效提升大理岩的吸能和储能效率,也能强化大理岩的吸能和储能极限,同时还能抑制大理岩的能量耗散效率和耗散能的提升幅度;岩样从峰前到峰后的能量演化是一个弹性能占比逐渐降低,耗散能占比逐渐升高的过程,残余阶段能量占比趋于稳定,其残余弹性能和残余耗散能均随着围压的增加呈现近似线性的增长,而最大弹性能和最大耗散能则随着围压的增加呈现非线性的增长。考虑初始损伤和残余承载力修正的损伤演化方程能很好地捕捉不同围压下岩样基于耗散能定义的损伤变量的变化特征。进而,基于能量相对比例定义的能耗比判据可视为大理岩的材料失效判据,基于能量绝对差值定义的能耗差判据可视为大理岩的结构失效判据。

心脏磁共振组织特征追踪技术评估左心衰合并肺高压的临床价值

目的基于心脏磁共振特征追踪(cardiac magnetic resonance feature tracking,CMR-FT)技术定量分析左心衰合并肺高压(pulmonary hypertension due to left heart failure,PH-LHF)患者的心肌应变参数并探讨心脏磁共振(cardiac magnetic resonance,CMR)在评估PH-LHF患者中的应用价值。材料与方法回顾性分析2018年9月至2020年9月期间215例确诊为左心衰(left heart failure,LHF)患者的临床和CMR参数,根据心脏超声测量的收缩期肺动脉压将患者分为两组,即LHF组(n=129)和PH-LHF组(n=86)。比较两组的基线资料和CMR参数包括心血管形态参数、心室容积功能参数以及心室心房应变参数。通过单因素及多因素logistic回归分析CMR参数中PH-LHF的独立预测因子,并绘制受试者工作特征(receiver operating characteristics,ROC)曲线评估CMR参数的诊断价值。结果PH-LHF患者组左右心室舒张末期容积指数、收缩末期容积指数、右心室心肌质量指数、室间隔角、左心房最大直径及面积、初始T1值均高于LHF组,而左右心室射血分数、右心室每搏输出量指数、右心室心脏指数低于LHF组。PH-LHF组左心室整体纵向应变、整体周向应变以及应变率、右心室整体周向应变、左心房主动应变、被动应变、总应变均不同程度下降。多因素logistic回归分析显示CMR参数中右心室射血分数,左心室整体周向应变,左心房主动应变和左心房最大直径是PH-LHF的独立预测因子。ROC曲线分析结果显示临床模型、CMR模型、联合模型诊断PH-LHF的AUC值分别为0.773、0.777、0.828,并且DeLong检验显示当加入CMR参数后,临床模型诊断效能提升(0.773 vs.0.828,P<0.05)。结论基于CMR-FT技术的心肌应变参数可以定量评估心室心房应变,反映PH-LHF患者的心肌功能以及运动情况,多参数CMR在诊断评估LHF是否合并PH中具有较好的临床增益价值。

射血分数改善的心力衰竭患者临床特征及预后分析

目的分析射血分数改善的心力衰竭患者的临床特征、预测因素及预后特点。方法回顾性分析2019年6月1日至2022年6月1日南京医科大学附属南京医院心内科因心力衰竭住院治疗的327例患者的临床资料。根据院内及复查时LVEF值将患者分为2组:射血分数持续性减低组(pHFrEF组,n=224):入院时LVEF≤40%,复查时LVEF≤40%;射血分数改善组(HFimpEF组,n=103):入院时LVEF≤40%,复查时LVEF>40%。通过电子病例、门诊系统及心衰随访数据库收集入组患者的一般资料、临床检验及检查数据,随访终点事件(心衰再住院、全因死亡)发生情况。采用Logistic回归模型分析HFimpEF发生的预测因素。绘制2组的Kaplan-Meier生存曲线,通过Log-rank法检验2组终点事件发生情况的差异。结果与pHFrEF组比较,HFimpEF组患者较为年轻,初诊心衰的比例更大,病程更短,血压更高、心率更快、整体心功能分级较好,扩张型心肌病的比例更大,同时肾功能更好、NT-proBNP较低,并且使用ACEI/ARB/ARNI及β受体阻滞剂的比例更高(P<0.05)。pHFrEF组较HFimpEF组合并脑血管病、冠心病及心肌梗死史的比例更高,并且PCI及CABG史的比例更高(P<0.05)。超声心动图方面,与pHFrEF组比较,HFimpEF组患者入院时左心更小,左心及右心功能均更好,治疗后心脏结构及功能的改善更显著(P<0.05)。多因素Logistic回归模型分析结果显示,初诊心衰、较短的病程、较高的平均动脉压、不合并陈旧性心肌梗死、较小的左室舒张末期内径以及较高的三尖瓣瓣环收缩期位移距离是HFimpEF发生的预测因素(P<0.05)。HFimpEF组的患者发生心衰再入院及全因死亡的比例显著低于pHFrEF组(P<0.05)。HFimpEF组患者的心衰再住院率、全因死亡率及复合终点发生率均显著低于pHFrEF组(P<0.05)。结论HFimpEF患者初始状况较好,预后更好,及时诊断、早期规范治疗心衰有助于左室射血分数的改善,并且血压更高、不合并陈旧性心肌梗死、左室更小以及右心功能更好的HFrEF患者更有发展为HFimpEF的潜力。

基于变分时频分解与随机森林的柴油机气门故障诊断方法

柴油机气门间隙异常作为常见的柴油机故障,对其进行状态监测和诊断对柴油机的高效正常运行具有重要意义。针对柴油机气门间隙异常典型故障,提出一种基于变分时频分解与随机森林算法的气门故障诊断方法。首先,由于不同气门故障所引起的冲击信号存在冲击差异,利用变分时频分解方法可以自适应提取冲击信号中的单频率单冲击成分的特点,实现振动信号的分解并进行特征值提取;然后,利用随机森林算法对特征值进行训练;最后,在柴油机试验台上通过调整气门间隙开展模拟试验。实际试验结果表明,所提方法在气门间隙异常故障诊断中的整体准确率为96.42%,验证了所提方法的有效性。

深埋矿井沿空留巷切顶卸压底板变形控制

目前有关巷道底鼓的研究与实践主要探讨巷道底板的变形机理及控制技术,对沿空留巷切顶卸压前后底板力学分析不全面。针对该问题,基于煤体分区破坏特征构建了切顶前后巷道围岩和底板力学模型,分析实体煤、巷旁支护及采空区对底板的作用,获得切顶前后巷道底鼓解析解,得出巷旁煤帮弹塑性区、巷道支护体及顶板下沉区底板所受载荷共同影响巷道底鼓量大小。采用数值模拟验证切顶卸压前后沿空留巷围岩破坏特征、应力分布及底鼓量变化,结果表明:切顶卸压技术可有效缩小巷道实体煤侧面及顶部的破坏区域,维持巷道围岩结构稳定;巷道底板最大应力、巷旁支护阻力、巷道底鼓量均下降,平均降幅分别为25.78%,56.14%,54.07%。现场应用结果表明,厚硬顶板沿空留巷底鼓量由709.3451 mm降至320.9658 mm,切顶卸压技术可以优化巷道围岩应力结构,抑制巷道底鼓,有效改善底板破坏情况。

Characterization of acute-on-chronic liver diseases: A multicenter prospective cohort study

BACKGROUND Acute-on-chronic liver disease(AoCLD)accounts for the majority of patients hospitalized in the Department of Hepatology or Infectious Diseases.AIM To explore the characterization of AoCLD to provide theoretical guidance for the accurate diagnosis and prognosis of AoCLD.METHODS Patients with AoCLD from the Chinese Acute-on-Chronic Liver Failure(ACLF)study cohort were included in this study.The clinical characteristics and outcomes,and the 90-d survival rate associated with each clinical type of AoCLD were analyzed,using the Kaplan-Meier method and the log-rank test.RESULTS A total of 3375 patients with AoCLD were enrolled,including 1679(49.7%)patients with liver cirrhosis acute decompensation(LC-AD),850(25.2%)patients with ACLF,577(17.1%)patients with chronic hepatitis acute exacer-bation(CHAE),and 269(8.0%)patients with liver cirrhosis active phase(LC-A).The most common cause of chronic liver disease(CLD)was HBV infection(71.4%).The most common precipitants of AoCLD was bacterial infection(22.8%).The 90-d mortality rates of each clinical subtype of AoCLD were 43.4%(232/535)for type-C ACLF,36.0%(36/100)for type-B ACLF,27.0%(58/215)for type-A ACLF,9.0%(151/1679)for LC-AD,3.0%(8/269)for LC-A,and 1.2%(7/577)for CHAE.CONCLUSION HBV infection is the main cause of CLD,and bacterial infection is the main precipitant of AoCLD.The most common clinical type of AoCLD is LC-AD.Early diagnosis and timely intervention are needed to reduce the mortality of patients with LC-AD or ACLF.