5G核心网业务模型的智能化预测研究

摘 要：核心网业务模型的建立是5G网络容量规划和网络建设的基础，通过现有方法得到的理论业务模型是静态不可变的且与实际网络存在偏离。为了克服现有5G核心网业务模型与现网模型适配性较差以及规划设备无法满足用户实际业务需求的问题，提出了一种长短期记忆（long short-term memory，LSTM）网络与卷积LSTM （convolution LSTM，ConvLSTM）网络双通道融合的 5G 核心网业务模型预测方法。该方法基于人工智能（artificial intelligence，AI）技术以实现高质量的核心网业务模型的智能预测，形成数据反馈闭环，实现网络自优化调整，助力网络智能化建设。     

Mixing characteristics for the single and air-water two-phase flows in miniaturized parallel multichannel-based devices

Investigation on the miniaturized parallel multichannel-based devices packed with glass beads to improve the mass exchange execution is the critical focal point of the current study. One of the essential parameters to specify the miniaturized devices' flow distribution is the residence time distribution (RTD). In the present context, the RTDs of a liquid tracer were investigated for the air-water multiphase flows (concurrent) across the multichannel-based miniaturized devices (comprising of 11 similar dimensional parallel channels). The devices were variable in height and packed with glass beads. The conductivity estimations generated the RTD curves and were addressed by the axial dispersion model (ADM). The fluid-flow rates differed within the range of 5–23 ml min⁻¹. The axial dispersion coefficients and the rate of the specific energy dispersion were investigated. The effects of pressure difference and geometry on the hydrodynamic attributes and mixing properties were well-illustrated, and the new correlations were suggested.     

A kinetic model involving oxidation and volatilization in SiC-B4C-xAl2O3 ceramics system

In this paper, a new kinetic model considering both oxidation and volatilization kinetics is established and applied to analyze the oxidation of SiC-B₄C-xAl₂O₃ ceramics and other systems in various oxidation conditions. The effects of diffusion area and volume changes during the oxidation process are considered in this model. The physical meaning of each parameter in this model is explicit and simple. According to this model, the diffusion coefficient of species and the corresponding diffusion activation energy are easily available. The practicability of this model is well verified by the experimental data of SiC-B₄C-xAl₂O₃ and other systems oxidized under different conditions. In addition, the practice shows that the model is applicable not only to the systems where oxidation and volatilization coexist, but also to the system where only oxidation plays a major role. We hope the model proposed in this work can be used in other materials with more complex environments.     

Thermomechanical stability and inelastic energy dissipation as durability criteria for fuel cell gas diffusion media with pre-assembly effects

In this article, pre-assembly hot-press pressure and thermal expansion effects in gas-diffusion layers (GDLs) are addressed to explore the practicalities of the constitutive model reported in the companion article. A facile technique is proposed to include deformation history dependent residual strain effects. The model is implemented in the numerical environment and compared with widely followed conventional models such as isotropic and orthotropic material models. With the normal and accelerated thermal expansion effects no significant variation in stresses or strains is reported with the compressible GDL model in contrast to the conventional incompressible form of the GDL model. The present work identifies the critical differences with advanced and extended variants of the model along with conventional GDL material models in terms of planar stress/strain distribution and the membrane response. Finally, the model is simulated for micro-cyclic stress loads of varying amplitudes that imitate the real working conditions of fuel cell. The inelastic energy dissipation in GDLs is predicted using the proposed model, which is utilized further to distinguish the safe (elastic) and unsafe (inelastic shakedown) operating limits. The inelastic collapse of GDLs is shown to be a active function of high amplitude micro-cyclic load with high initial clamping load.     

Experimental study on the deformation and failure mechanism of overburden rock during coal mining using a comprehensive intelligent sensing method

Understanding the spatiotemporal evolution of overburden deformation during coal mining is still a challenge in engineering practice due to the limitation of monitoring techniques. Taking the Yangliu Coal Mine as an example, a similarity model test was designed and conducted to investigate the deformation and failure mechanism of overlying rocks in this study. Distributed fiber optic sensing (DFOS), high-density electrical resistivity tomography (HD-ERT) and close-range photogrammetry (CRP) technologies were used in the test for comprehensive analyses. The combined use of the three methods facilitates the investigation of the spatiotemporal evolution characteristics of overburden deformation, showing that the mining-induced deformation of overburden strata was a dynamic evolution process. This process was accompanied by the formation, propagation, closure and redevelopment of separation cracks. Moreover, the key rock stratum with high strength and high-quality lithology played a crucial role in the whole process of overburden deformation. There were generally three failure modes of overburden rock layers, including bending and tension, overall shearing, and shearing and sliding. Shear failure often leads to overburden falling off in blocks, which poses a serious threat to mining safety. Therefore, real-time and accurate monitoring of overburden deformation is of great significance for the safe mining of underground coal seams.     

基于推理注意力机制的二阶段网络图像修复

现有的图像修复方法在处理大面积缺失或高度纹理化的图像时，通常会产生扭曲的结构或与周围区域不一致的模糊纹理，无法重建合理的图像结构。为此，提出了一种基于推理注意力机制的二阶段网络图像修复方法。首先通过边缘生成网络生成合理的幻觉边缘信息，然后在图像补全网络完成图像的重建工作。为了进一步生成视觉效果更逼真的图像，提高图像修复的精确度，在图像补全网络采用推理注意力机制，有效控制了生成特征的不一致性，从而生成更有效的信息。所提方法在多个数据集上进行了实验验证，结果表明该图像修复方法的结构相似性指数达到了88.9%，峰值信噪比达到了25.56 dB，与现有的图像修复方法相比，该方法具有更高的图像修复精确度，生成的图像更逼真。     

Dilated and soft attention-guided convolutional neural network for breast cancer histology images classification

Breast cancer is one of the most common types of cancer in women, and histopathological imaging is considered the gold standard for its diagnosis. However, the great complexity of histopathological images and the considerable workload make this work extremely time-consuming, and the results may be affected by the subjectivity of the pathologist. Therefore, the development of an accurate, automated method for analysis of histopathological images is critical to this field. In this article, we propose a deep learning method guided by the attention mechanism for fast and effective classification of haematoxylin and eosin-stained breast biopsy images. First, this method takes advantage of DenseNet and uses the feature map's information. Second, we introduce dilated convolution to produce a larger receptive field. Finally, spatial attention and channel attention are used to guide the extraction of the most useful visual features. With the use of fivefold cross-validation, the best model obtained an accuracy of 96.47% on the BACH2018 dataset. We also evaluated our method on other datasets, and the experimental results demonstrated that our model has reliable performance. This study indicates that our histopathological image classifier with a soft attention-guided deep learning model for breast cancer shows significantly better results than the latest methods. It has great potential as an effective tool for automatic evaluation of digital histopathological microscopic images for computer-aided diagnosis.     

Model of reactive precipitation process of preparation nano-ZnO in rotating packed bed: Contribution ratio

With excellent micromixing characteristic of rotating packed bed (RPB), many nanoparticles with small average size, narrower distribution and good morphology had been successfully and continuously prepared. To reveal complex crystal process, an empirical model were developed to simulate nano-ZnO by considering mass changed, population balance equation, growth rate G, nucleation rate B, drop sizes D_i, and resident time t. The predicted particle sizes were shown good agreement with experimental data with error of ±10%. Therefore, it was further adopted to predict the effects of rotating speed, liquid flow rate and reactant concentration on the mean particle size. To look more deeply insight in this process, their contribution ratios were further analyzed. The proposed empirical models were of great helpful to obtain suitable operation conditions for preparing much better properties of nanoparticles with fewer experiments. It was also beneficial to produce other nanoparticles in RPB.     

Trade-off study between risk and benefit in safety devices for hydrogen refueling stations using a dynamic physical model

As hydrogen refueling stations become increasingly common, it is clear that a high level of economic efficiency and safety is crucial to promoting their use. One way to reduce costs is to use a simple orifice instead of an excess flow valve, which Japanese safety regulations have identified as a safety device. However, there is concern about its effect on refueling time and on risk due to hydrogen leakage. To clarify the effect, we did a study of model-based refueling time evaluation and quantitative risk assessment for a typical refueling station. This study showed that an orifice is an effective alternative safety device. The increase in refueling time was less than 10%, based on simulations using a dynamic physical model of the station. Neither was there a significant difference in the risk between a configuration with excess flow valves and one with an orifice.