CLARO: modeling and processing uncertain data streams

Uncertain data streams, where data are incomplete and imprecise, have been observed in many environments. Feeding such data streams to existing stream systems produces results of unknown quality, which is of paramount concern to monitoring applications. In this paper, we present the claro system that supports stream processing for uncertain data naturally captured using continuous random variables. claro employs a unique data model that is flexible and allows efficient computation. Built on this model, we develop evaluation techniques for relational operators by exploring statistical theory and approximation. We also consider query planning for complex queries given an accuracy requirement. Evaluation results show that our techniques can achieve high performance while satisfying accuracy requirements and outperform state-of-the-art sampling methods.     

PP/PVC/PP-g-(St-co-MMA)的形态结构

以苯乙烯一甲基丙烯酸甲酯共聚物接枝聚丙烯[PP—g-（St—co—MMA）]为相容剂，利用扫描电子显微镜（SEM）和x射线能谱仪（EDS）对PP／PVC／PP—g-（St—co-MMA）的形态结构进行了研究。由x射线能谱微区分析得到了共混物中氯元素面分布图，对氯元素面分布进行了粒径统计。结果表明，在PP／PVC中加入PP-g-（St—co—MMA）相容剂，共混物亚微观相形态结构发生了变化。PP—g-（St—co—MMA）能够明显降低PP／PVC共混物中PVC分散相的分散尺寸，提高共混物两相之间的相容性。改变相容剂的用量和共混物的组成，都对体系的形态结构产生影响。     

基于循环神经网络的电信行业容量数据预测方法

在电信运维的容量预测过程中,存在容量指标和部署业务种类繁多的问题。现有研究未考虑指标数据类型的差异,对所有类型的数据使用同种预测方法,使得预测效果参差不齐。为了提升指标预测效率,提出一种指标数据类型分类方法,利用该方法将数据类型分为趋势型、周期型和不规则型。针对其中的周期型数据预测,提出基于双向循环神经网络（BiRNN）的周期型容量指标预测模型,记作BiRNN-BiLSTM-BI。首先,为分析容量数据的周期特征,提出一种忙闲分布分析算法;其次,搭建循环神经网络（RNN）模型,该模型包含一层BiRNN和一层双向长短时记忆网络（BiLSTM）;最后,充分利用系统忙闲分布信息,对BiRNN输出的结果进行优化。与传统的三次指数平滑、差分自回归移动平均（ARIMA）模型和反向传播（BP）神经网络模型进行比较的实验结果表明,在统一日志数据集和分布式缓存数据集上,提出的BiRNN-BiLSTM-BI模型的均方误差（MSE）分别比对比模型中表现最优的模型降低了15.16%和45.67%,可见预测准确率得到了很大程度的提升。     

一种引入冗余控制的特征排序模型

针对特征排序方法较少考虑特征之间的相关关系,导致选择的特征子集存在冗余的问题,提出一种引入冗余控制的特征排序模型。将特征子集判别能力最大且冗余程度最小作为模型的目标函数,以降低特征之间的冗余;使用贪心方法和非线性规划方法对模型进行求解。在9个开源数据上的实验及与特征排序方法比较表明,本模型在大部分数据上,所选择的特征子集能够获得更好的分类准确性且个数更少;使用非线性规划方法求解时,能够直接得到特征子集,有利于确定特征个数。本模型可用于特征之间存在冗余时的特征选择。     

基于时空消耗理论的城市快速路长度计算方法

随着城市交通压力的不断增大,我国许多大中型城市都展开了快速路建设的热潮,针对城市快速路网长度的研究,基于供需平衡的原理,文章提出一种基于时空消耗法的快速路网长度计算方法,使快速路的时空总量等于快速路上所有车辆的时空总量,进而得到快速路长度计算模型,并应用到济南市快速路计算实例中。计算结果与济南市快速路实际规划里程相对一致,此方法可以作为确定城市快速路规划里程的计算依据。     

Network Characterizations for Excluding Braess’s Paradox

Braess’s paradox exposes a counterintuitive phenomenon that when travelers selfishly choose their routes in a network, removing links can improve the overall network performance. Under the model of nonatomic selfish routing, we characterize the topologies of k-commodity undirected and directed networks in which Braess’s paradox never occurs. Our results strengthen Milchtaich’s series-parallel characterization (Milchtaich, Games Econom. Behav. 57(2), 321–346 (2006)) for the single-commodity undirected case.     

Vibration analysis of cable-driven parallel manipulators

A cable-driven parallel manipulator is a manipulator whose end-effector is driven by a number of parallel cables instead of rigid links. Since cables always have more flexibility than rigid links, a cable manipulator bears a concern of possible vibration. Thus, investigation of vibration of cable manipulators caused by cable flexibility is important for applications requiring high system stiffness or bandwidth. This paper provides a vibration analysis of general 6-DOF cable-driven parallel manipulators. Based on the analysis of the natural frequencies of the multibody system, the study demonstrates that a cable manipulator can be designed stiff enough for special applications like the cable-manipulator based hardware-in-the-loop simulation of contact dynamics. Moreover, under an excitation, a cable may vibrate not only in its axial direction, but also in its transversal direction. The paper also analyzes the vibration of cable manipulators caused by cable flexibilities in both axial and transversal directions. It is shown that the vibration of a cable manipulator due to the transversal vibration of cables can be ignored comparing to that due to the axial flexibility of cables.     

Two-dimensional wavelet transform and application to holographic particle velocimetry

The goal of holographic particle velocimetry is to infer fluid velocity patterns from images reconstructed from doubly exposed holograms of fluid volumes seeded with small particles. The advantages offered by in-line holography in this context usually make it the method of choice, but seeding densities sufficient to achieve high spatial resolution in the sampling of the velocity fields cause serious degradation, through speckle, of the signal-to-noise ratio in the reconstructed images. The in-line method also leads to a great depth of field in paraxial viewing of reconstructed images, making it essentially impossible to estimate particle depth with useful accuracy. We present here an analysis showing that these limitations can be circumvented by variably scaled correlation, or wavelet transformation. The shift variables of the wavelet transform are provided automatically by the optical correlation methodology. The variable scaling of the wavelet transform derives, in this case, directly from the need to accommodate varying particle depths. To provide such scaling, we use a special optical system incorporating prescribed variability in spacings and focal length of lenses to scan through the range of particle depths.

Calculation shows, among other benefits, improvement by approximately two orders of magnitude in depth resolution. A much higher signal-to-noise ratio together with faster data extraction and processing should be attainable.

     

Coordination control method to block cascading failure of a renewable generation power system under line dynamic security

The large application of renewable energy generation (REG) has increased the risk of cascading failures in the power system. At the same time REG also provides the possibility of new approaches for the suppression of such failures. However, the capacity and position of the synchronous generator (SG) involved in regulation limit the power regulation speed (PRS) of REG to the overload line which is the main cause of cascading failures, while the PRS of SG is related to the position and shedding power. REG and SGs have difficulty in achieving effective cooperation under constraints of system power balance. Particularly, the dynamic variation of line flow during power regulation causes new problems for the accurate evaluation of line thermal safety under overload. Therefore, a new strategy for quantitatively coordinating shedding power and power regulation to block cascading failures in the dynamic security domain is proposed in this paper. The control capability and dynamic security domain of the overload line are modeled, and the coordination control method based on power regulation is then proposed to minimize shedding power. The algorithm for the optimal control scheme considers the constraints of load capacity, power source capacity and bus PRS. The correctness of the proposed method is verified using case studies.     

身份保持约束下的面部图像超分辨率重建方法

低分辨率是影响人脸识别精度的重要因素。一种有效方法是使用图像超分辨率技术对低分辨率图像重建,生成超分辨率图像后再对其作人脸识别,从而克服低分辨率面部图像对人脸识别的限制。但是,现有超分辨率方法在重建过程中往往忽略了保持其原始身份信息,这直接影响生成图像的人脸识别结果。针对上述问题,提出了一种身份保持约束下的面部超分辨率重建方法IPNet,在提高低分辨率面部图像质量的同时,能保持重建后的面部图像身份。IPNet方法将语义分割网络和面部生成器相结合,通过语义分割网络提取低维隐码和多分辨率空间特征,进而指导面部生成器输出接近于原图的真实面部图像。在此基础上引入人脸识别网络,将身份信息整合到超分辨率方法中,从而约束重建前后的面部图像身份保持一致。实验结果表明,IPNet方法在超分辨率图像质量和身份保持上均优于其他对比方法。