Synthesis and slurry spray coating of barium strontium alumino silicate on SiC substrate

Barium strontium alumino silicate (BSAS); (Ba_0.6Sr_0.4Al₂Si₂O₈) was synthesized through solid state reaction between BaCO₃, SrCO₃, Al₂O₃ and SiO₂ subjected to wet milling in isopropanol for about 24 h. The sequence of the solid state reaction was studied by subjecting to DG/DTG from room temperature to 1550 °C. The crystallographic phase evolution was confirmed by X-ray diffraction of the powders calcined in the range 1000 to 1300 °C for 2 h. The monoclinic celsian phase obtained at 1300 °C, pelletized through uniaxial pressing was sinterable to 67 to 78% density in the temperature range of 1300 to 1500 °C. The density improved to 75 to 94% after ball milling for 76 h, while ZrO₂ addition further improved the density by 2%. The celcian phase of BSAS was dispersed in isopropyl alcohol, milled for about 24 h and spray coated on to plain SiC and mullite precoated SiC substrates. Sintering of coated samples and characterization for weight gain/loss, microstructure, scratch test prove that mullite + BSAS coating is more effective than single layer coating of BSAS on SiC substrates.     

一类分数阶微积分的中值定理及应用

Riemann-Liouville分数阶微积分算子是一类带有一个函数的分数阶微积分算子的特殊情形，以Riemann-Liouville分数阶微积分算子的积分中值定理和微分中值定理为基础，我们得到了一类带有一个函数的分数阶微积分算子的积分中值定理和微分中值定理，并给出其在计算方面的一些应用。     

High performance H2 sensor based on rGO-wrapped SnO2–Pd porous hollow spheres

Hydrogen (H₂) sensors based on metal oxide semiconductors (MOS) are promising for many applications such as a rocket propellant, industrial gas and the safety of storage. However, poor selectivity at low analyte concentrations, and independent response on high humidity limit the practical applications. Herein, we designed rGO-wrapped SnO₂–Pd porous hollow spheres composite (SnO₂–Pd@rGO) for high performance H₂ sensor. The porous hollow structure was from the carbon sphere template. The rGO wrapping was via self-assembly of GO on SnO₂-based spheres with subsequent thermal reduction in H₂ ambient. This sensor exhibited excellently selective H₂ sensing performances at 390 °C, linear response over a broad concentration range (0.1–1000 ppm) with recovery time of only 3 s, a high response of ～8 to 0.1 ppm H₂ in a minute, and acceptable stability under high humidity conditions (e. g. 80%). The calculated detection limit of 16.5 ppb opened up the possibility of trace H₂ monitoring. Furthermore, this sensor demonstrated certain response to H₂ at the minimum concentration of 50 ppm at 130 °C. These performances mainly benefited from the special hollow porous structure with abundant heterojunctions, the catalysis of the doped-PdO_x, the relative hydrophobic surface from rGO, and the deoxygenation after H₂ reduction.     

Time and space dependence of large-scale synthesis of colloidal particle with special morphology and tuneable polymer film architecture

This paper describes a facile method to control the morphology of polymer colloids and the architecture of polymer film via miniemulsion polymerization. By taking advantage of cyclization between the symmetrical diacrylate cross-linker hexamethylene diacrylate (HDDA) and the pendent vinyl in colloidal particles, the morphology of polymer colloids and the architecture of the after-formed polymer film were able to be well controlled by tuning the loading of cross-linker HDDA and crosslinking time. Four kinds of polymer colloid morphologies and four kinds of film architecture (honeycomb, close-packed, loose-packed, and enhanced-honeycomb) were characterized by TEM. The film formation mechanisms behind them were proposed based on the special and interesting results including Z-average size of the colloidal particles, Mc (molecular weight between crosslinking points) and mechanical properties of polymer film. Our results highly suggested that the morphology of polymer colloids and the polymer film architecture together determine the adhesive properties of the colloidal polymer film. The best of 180°-peel resistance, T-peel resistance and shear resistance of the polymer films were 138.12 N/25 mm, 40.98 N/25 mm and 25.72 N/cm² at 2.0 phm, 2.0 phm and 0.4 phm with the same crosslinking time of 90 min, respectively. The proposed method is promising to be scaled up for industrial production due to its well adaptability.     

Sequential alignment attention model for scene text recognition

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state-of-the-art solutions usually depend on the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. Unfortunately, there often exists severe misalignment between feature areas and text labels in real-world scenarios. To address this problem, this paper proposes a sequential alignment attention model to enhance the alignment between input images and output character sequences. In this model, an attention gated recurrent unit (AGRU) is first devised to distinguish the text and background regions, and further extract the localized features focusing on sequential text regions. Furthermore, CTC guided decoding strategy is integrated into the popular attention-based decoder, which not only helps to boost the convergence of the training but also enhances the well-aligned sequence recognition. Extensive experiments on various benchmarks, including the IIIT5k, SVT, and ICDAR datasets, show that our method substantially outperforms the state-of-the-art methods.     

钻井机器人的连续油管钻压和钻速控制模型

连续油管钻井机器人利用机身内外的钻井液压力差作为动力源，可在牵引连续油管的同时加载钻压。以钻井机器人为基础，建立连续油管钻柱动力学模型，并推导出通过钻井液排量控制钻压和钻速的单参数控制数学模型；对钻井机器人引入调速回路，建立具有调速功能的钻柱动力学模型；在溢流阀调定压力大于机身内外压差时，推导出利用钻井液排量和节流阀流通面积两种参数控制钻压、钻速的数学模型，在溢流阀调定压力小于机身内外压差时，推导出利用钻井液排量、节流阀流通面积和溢流阀调定压力3种参数控制钻压、钻速的数学模型；以11.43 cm（4.5英寸）井眼为例，对上述3种数学模型进行了分析。分析结果表明：钻压、钻速随钻井液排量的增加基本呈线性增加，在钻井液排量大于0.005 m³/s时，钻井机器人能够向前爬行，在钻井液排量大于0.005 7 m³/s时，钻头能够正常钻进；调节节流阀流通面积和溢流阀调定压力，可以在一定范围内无级调钻压和钻速；3种控制方法相结合，可以实现小排量、大钻压，及大排量、小钻压等钻井参数的控制。以控制模型为基础，针对不同井下工况建立钻进工艺的专家数据库，以钻井机器人为"大脑"，结合井下随钻测量数据就能够实现闭环控制，自动钻进。     

Continuum model of two-dimensional crystal lattice of metamaterials

The continuum models of two-dimensional crystal lattice of metamaterial are investigated in this paper. The equivalent classical continuum requires the introduction of frequency-dependent mass density that becomes negative or infinite near the resonance frequency. In order to avoid the frequency-dependent mass density and make the dispersive characteristic of the elastic waves propagating in the equivalent continuum approximating that of lattice wave in two-dimensional crystal lattice of metamaterial, three kinds of continuum models, namely, the multiple displacements continuum model, the strain gradient continuum model and the nonlocal strain gradient continuum model, are investigated. It is found that the nonlocal gradient continuum model may better represent the dispersive relation and the bandgap characteristics of the metamaterial by the appropriate selection of nonlocal parameters.     

ATSP:时态网络可视化的自适应时间片划分方法

现有的时态网络可视化方法大多采用等量时间片来可视化网络的演变，不利于时态模式的快速挖掘和发现。为此，根据时态网络固有的特征提出自适应时间片划分方法（Adaptive Time Slice Partition method，ATSP）。在时态网络的两种表示方式（基于事件的表示方式和基于快照的表示方式）的基础上，构建了ATSP的基础模型，同时提出了一种改进模型用来描述事件间隔时间服从长尾分布的时态网络。为了实现时间片的不等量划分，针对探索任务的不同提出了基于时态模式的ATSP规则和基于中心节点的ATSP规则，并提出了实现算法--层次划分算法（Hierarchical Partition algorithm，HP）和增量划分算法（Incremental Partition algorithm，IP）。实验结果表明，ATSP方法比传统的时间片划分方法更能准确地表示网络的时态特征，且该方法应用于可视化时，能有效归纳并展示网络的特征，明显提高了视觉分析的效率。     

时态神经网络模型及其在股票分类预测上应用

对具有时间属性的数据进行数据挖掘称为时态数据挖掘，用以发现数据在时间上的知识，当数据变化不规律时，如股票交易数据,就很难发现有价值的规律与规则。而神经网络具有并行、容错、可以硬件实现以及自我学习的优点,可作为股票分类预测应用的一种方法。通过将股票数据与时态型相结合，将股票数据转换成时态型股票数据，提出时态神经网络模型的分类方法，对收集的若干上市公司十年内的股票数据进行分析，构建了时态股票数据神经网络分类器对股票进行分类预测。经过实验验证，相比改进前的神经网络和支持向量机方法，该分类器具有更高的分类准确率。结果证明，这种时态数据神经网络模型对于多只股票的分类预测是非常有效的，可以很好地运用到股票市场的分类预测中。     

Cloud resource prediction model based on triple exponential smoothing method and temporal convolutional network

The container cloud represented by Docker and Kubernetes has the advantages of less additional resource overhead and shorter start-up and destruction time.However there are still resource management issues such as over-supply and under-supply.In order to allow the Kubernetes cluster to respond “in advance” to the resource usage of the applications deployed on it,and then to schedule and allocate resources in a timely,accurate and dynamic manner based on the predicted value,a cloud resource prediction model based on triple exponential smoothing method and temporal convolutional network was proposed,based on historical data to predict future demand for resources.To find the optimal combination of parameters,the parameters were optimized using TPOT thought.Experiments on the CPU and memory of the Google dataset show that the model has better prediction performance than other models.