Effect of fiber orientation on Liquid–Gas flow in the gas diffusion layer of a polymer electrolyte membrane fuel cell

In this study, the lattice Boltzmann method was used to simulate the three-dimensional intrusion process of liquid water in the gas diffusion layer (GDL) of a polymer electrolyte membrane fuel cell (PEMFC). The GDL was reconstructed by the stochastic method and used to investigate fiber orientation's influence on liquid water transport in the GDL of a PEMFC. The fiber orientation can be described by the angle between a single fiber and the in-plane direction; three different samples were simulated for three different fiber orientation ranges. The simulated permeability correlated well with the anisotropic characteristics of reconstructed carbon papers. It was concluded that the fiber orientation had a significant effect on the liquid invasion pattern in the GDL by changing the pore shape and distribution of the GDL. The results indicated that the stochastically reconstructed GDL, taking into account the fiber orientation, better demonstrates the mass transport properties of the GDL.     

Fibrous porous ceramics with devisable phenolic resin reinforcing layer

Fibrous porous ceramics with devisable phenolic resin reinforcing layer were fabricated using low cost atmospheric impregnation technology at room temperature. In combination with additional sealing method, phenolic resin reinforcing layer with controllable thickness could be obtained on the surface of fibrous porous ceramics. Typical gradient profile was observed along the thickness direction of impregnation. The effects of the phenolic resin reinforcing layer on mechanical properties and thermal insulation properties were studied. The results revealed that compressive strength increased from 1.70?MPa to 2.61?MPa, tensile strength increased from 0.78?MPa to 0.91?MPa, and flexural strength increased from 9.55?MPa to 10.89?MPa with the phenolic resin layer increasing from 0?mm to 9?mm. Simultaneously, room-temperature thermal conductivity increased from 0.051?W/(m·K) to 0.055?W/(m·K). In addition, the impact resistance of the surface of the material was obviously improved. The contact angel of the surface of the material exceeded 125°, which effectively improved the environmental adaptability.     

DEM: Deep Entity Matching Across Heterogeneous Information Networks

Heterogeneous information networks, which consist of multi-typed vertices representing objects and multi-typed edges representing relations between objects, are ubiquitous in the real world. In this paper, we study the problem of entity matching for heterogeneous information networks based on distributed network embedding and multi-layer perceptron with a highway network, and we propose a new method named DEM short for Deep Entity Matching. In contrast to the traditional entity matching methods, DEM utilizes the multi-layer perceptron with a highway network to explore the hidden relations to improve the performance of matching. Importantly, we incorporate DEM with the network embedding methodology, enabling highly efficient computing in a vectorized manner. DEM’s generic modeling of both the network structure and the entity attributes enables it to model various heterogeneous information networks flexibly. To illustrate its functionality, we apply the DEM algorithm to two real-world entity matching applications: user linkage under the social network analysis scenario that predicts the same or matched users in different social platforms and record linkage that predicts the same or matched records in different citation networks. Extensive experiments on real-world datasets demonstrate DEM’s effectiveness and rationality.

     

结合预训练模型和语言知识库的文本匹配方法

针对文本匹配任务,该文提出一种大规模预训练模型融合外部语言知识库的方法。该方法在大规模预训练模型的基础上,通过生成基于WordNet的同义—反义词汇知识学习任务和词组—搭配知识学习任务引入外部语言学知识。进而,与MT-DNN多任务学习模型进行联合训练,以进一步提高模型性能。最后利用文本匹配标注数据进行微调。在MRPC和QQP两个公开数据集的实验结果显示,该方法可以在大规模预训练模型和微调的框架基础上,通过引入外部语言知识进行联合训练有效提升文本匹配性能。     

面向智能客服系统的情感分析技术

该文以阿里小蜜为例,对智能客服系统中的情感分析技术进行比较全面的介绍,包括情感分析算法模型的原理及其在智能客服系统的多个应用场景中的实际落地使用方式和效果分析。智能客服在解决客户高频业务问题的同时,也需要给客户提供多维度的、具有类人能力的助理、导购、语聊和娱乐等服务,提高客户对智能客服机器人的整体满意度。在此过程中,情感分析技术在机器人类人能力建设中起到了至关重要的作用。该文围绕智能客服系统中人机结合的服务形式,从六个维度总结和介绍了情感分析技术在智能客服系统中的应用场景,包括用户情感检测、用户情感安抚、情感生成式语聊、客服服务质检、会话满意度预估和智能人工入口。     

ZerNet: Convolutional Neural Networks on Arbitrary Surfaces Via Zernike Local Tangent Space Estimation

In this paper, we propose a novel formulation extending convolutional neural networks (CNN) to arbitrary two-dimensional manifolds using orthogonal basis functions called Zernike polynomials. In many areas, geometric features play a key role in understanding scientific trends and phenomena, where accurate numerical quantification of geometric features is critical. Recently, CNNs have demonstrated a substantial improvement in extracting and codifying geometric features. However, the progress is mostly centred around computer vision and its applications where an inherent grid-like data representation is naturally present. In contrast, many geometry processing problems deal with curved surfaces and the application of CNNs is not trivial due to the lack of canonical grid-like representation, the absence of globally consistent orientation and the incompatible local discretizations. In this paper, we show that the Zernike polynomials allow rigourous yet practical mathematical generalization of CNNs to arbitrary surfaces. We prove that the convolution of two functions can be represented as a simple dot product between Zernike coefficients and the rotation of a convolution kernel is essentially a set of 2 × 2 rotation matrices applied to the coefficients. The key contribution of this work is in such a computationally efficient but rigorous generalization of the major CNN building blocks.     

基于多尺度结构特征的快速异源图像匹配

针对异源图像提出一种基于多尺度密集结构特征的快速匹配算法。算法首先利用Gabor滤波器逐像素提取图像中的结构响应,再根据主方向响应对多尺度结构特征融合,然后使用快速傅里叶变换在频域计算各特征分量图像之间的卷积,最后将卷积生成的系数矩阵求和计算出图像之间的相似性并选择相似性最大位置作为匹配结果输出。本文算法能有效适应异源图像间的非线性灰度变化和噪声干扰问题。测试使用可见光、红外、雷达图像组成的异源图像数据集对本文算法和现有算法进行测试比较,结果表明:本文算法的平均误匹配率最低,并且计算速度有明显优势。     

Damage evolution around white etching layer during uniaxial loading

Rolling contact fatigue cracks and thermally induced defects are common problems in the railway industry especially as demands for increasing loads, speeds, and safety continue to rise. Often, the two types of defects are found together in the field, however, whether one causes the other to occur is not completely agreed upon. The effect of thermal damage, in the form of a martensite spot on pearlitic steel test bars, on the fatigue life in uniaxial low cycle fatigue experiments was investigated by the authors. However, the focus of the current work was to characterize the damage evolution from the low cycle fatigue (LCF) tests and correlate the crack initiation and propagation with the initial thermal damage. Residual stress measurements, digital image correlation, and X‐ray tomography were used to characterize the effects of the thermal damage before, during, and after fatigue testing, respectively. It was found that the thermal damage causes strain accumulation and crack initiation at the interface between the two materials. The strain evolution was visualized using digital image correlation (DIC), clearly showing the strain concentrations at the top and bottom of the white etching layers (WEL), where the residual stresses are also most tensile. X‐ray tomography confirmed the planar crack growth from the martensite spot.     

Evaluation of electrochemical properties of antibacterial ZnO layers deposited to 316LVM steel using atomic layer deposition

The presence of a biofilm can lead to the disappearance of the surrounding bone tissue and, as a result, disturb the osseointegration process. Unfortunately, both in the case of instability of the implant and long-term bacterial infections, there is often a need for reoperation as well as replacement of the implant, which in turn is associated with huge costs, but most of all discomfort for the patient associated with long-term hospitalization. In order to limit this unfavorable process, the physicochemical properties of the surface layer of implants are indicated. Therefore, the paper proposes applying a layer with antibacterial effect on the surface of 316LVM steel used in bone surgery. As part of the work, the ZnO layer was applied using the atomic layer deposition method with different parameters of the application process (different number of cycles at constant temperature). In the first stage, pitting corrosion resistance tests were carried out using the potentiodynamic method and studies using electrochemical impedance spectroscopy. The surface morphology tests using a scanning electron microscope were also complemented. Obtained results may form the basis for the development of more detailed criteria for the assessment of the final quality of medical devices used in the skeletal system.