基于卷积神经网络的5G蜂窝网络无线定位方法

5G蜂窝网络发展迅猛，其覆盖面积将逐渐增大，因此使用5G蜂窝网络进行定位是有研究潜力的研究方向。本文提出一种新的深度学习技术来实现高效、高精度和低占用的定位，以代替传统指纹定位过程中繁重的指纹库生成以及距离计算。该方法建立了一个特殊的卷积神经网络，并根据5G天线信号的接收信号强度指示、相位和到达角等特征量，选择合适的输入数据格式构造样本组建训练集，对该卷积神经网络进行训练。训练得到的卷积神经网络可以替代指纹定位中的庞大指纹库，非常有利于直接在5G移动设备端实现定位。虽然卷积神经网络在训练过程中需要大量时间，但在训练完毕后直接进行分类定位的速度非常快，可以保障定位实现的实时性。本文所实现的卷积神经网络权重与偏置所占内存不到0.5 MB，且能够在实际应用环境中以95%的定位准确率以及0.1 m的平均定位精度实现高精度定位。     

疫情下电磁场理论课程的网络教学

本文主要总结了新冠疫情期间作者的电磁场理论课程在线教学经验。对比分析了录播和直播的优缺点后,选择录播教学方式。基于超星网络教学平台,展示了录播网络教学的具体措施,包括网上答疑和学习效果检查以及在线批改作业等。给出了网络教学可以为线下教学继续使用的方法和手段,为疫情结束后的正常教学提供了新的网络教学补充措施。     

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。     

Ion migration behavior and its interaction with sintering environment in cation conductor during flash sintering

Flash sintering has been reported in various ceramics. Nevertheless, anion and cation conductors exhibit different flash-sintering behaviors, and the interaction mechanism between the conductive species and the sintering environment has remained unclear. Herein, we report the flash-sintering phenomena of a typical cation conductor, Na₃Zr₂(SiO₄)₂(PO₄) with anode region surrounded by air and NaNO₃ environments. The results prove that the ionic behavior and joule heating distribution can be controlled by changing the electrode environment. Four possible scenarios describing the ion migration behavior and interaction with the environment are proposed for providing a guidance for controlling the ion interaction behavior during flash sintering.     

Motion trajectory prediction model of hydrogen leak and diffusion in a stable thermally-stratified environment

The motion trajectory of hydrogen leakage is an essential safe issue for the application of hydrogen energy. A dimensionless fast-running motion trajectory prediction model is proposed to predict the dispersion characteristics of the buoyant jet of hydrogen leakage for the accident. The impact of different leakage angles, leakage velocity and thermal stratification of ambient air on hydrogen leakage behavior was analyzed. The new developed model was verified by experimental results in literatures. Leakage hydrogen can flow upwards freely in a uniform environment. However, it shows an oscillating trajectory at a certain height in a thermally stratified environment, which is so called “locking phenomenon”. The trajectory of hydrogen leakage is upward and hydrogen gathers at the top of the space to form stratification in a uniform environment, while the hydrogen leakage shows an oscillating trajectory at a certain height in a thermal stratification environment. With the increase of Froude number Fr, it shows that the stable height and maximum height of the leakage airflow have a trend of rising first and then falling in a thermally stratified environment. The findings are expected to give guidance in real-world situations, for example, a larger Fr value and a larger temperature gradient can lead to a decrease in the stable height in the thermally stratified environment. It is found that the fitting of the stable height with different temperature gradients satisfies the power function relationship. This work is expected to be helpful for reducing hydrogen leakage accumulation and explosion risk.     

Machine fault diagnosis with small sample based on variational information constrained generative adversarial network

In actual engineering scenarios, limited fault data leads to insufficient model training and over-fitting, which negatively affects the diagnostic performance of intelligent diagnostic models. To solve the problem, this paper proposes a variational information constrained generative adversarial network (VICGAN) for effective machine fault diagnosis. Firstly, by incorporating the encoder into the discriminator to map the deep features, an improved generative adversarial network with stronger data synthesis capability is established. Secondly, to promote the stable training of the model and guarantee better convergence, a variational information constraint technique is utilized, which constrains the input signals and deep features of the discriminator using the information bottleneck method. In addition, a representation matching module is added to impose restrictions on the generator, avoiding the mode collapse problem and boosting the sample diversity. Two rolling bearing datasets are utilized to verify the effectiveness and stability of the presented network, which demonstrates that the presented network has an admirable ability in processing fault diagnosis with few samples, and performs better than state-of-the-art approaches.     

使能未来工厂的5G能力综述

5G系统将移动通信服务从移动电话、移动宽带和大规模机器通信扩展到新的应用领域，即所谓对通信服务有特殊要求的垂直领域。对使能未来工厂的5G能力进行了全面的分析总结，包括弹性网络架构、灵活频谱、超可靠低时延通信、时间敏感网络、安全和定位，而弹性网络架构又包括对网络切片、非公共网络、5G局域网和边缘计算的支持。希望从广度到深度，对相关的理论及技术应用做透彻、全面的梳理，对其挑战做清晰的总结，从而为相关研究和工程技术人员提供借鉴。     

基于全U网络的混凝土路面裂缝检测算法

针对现有的混凝土裂缝检测算法在各种复杂环境中检测精度不够、鲁棒性不强的问题,根据深度学习理论和U-net模型,提出一种全U型网络的裂缝检测算法。首先,依照原U-net模型路线构建网络;然后,在每个池化层后都进行一次上采样,恢复其在池化层之前的特征图规格,并将其与池化之前的卷积层进行融合,将融合之后的特征图作为新的融合层与原U-net网络上采样之后的网络层进行融合;最后,为了验证算法的有效性,在测试集中进行实验。结果表明,所提算法的平均精确率可达到83.48%,召回率为85.08%,F1为84.11%,相较于原U-net分别提升了1.48%,4.68%和3.29%,在复杂环境中也能提取完整裂缝,保证了裂缝检测的鲁棒性。     

Learn decision trees with deep visual primitives

In this paper, we strive to propose a self-interpretable framework, termed PrimitiveTree, that incorporates deep visual primitives condensed from deep features with a conventional decision tree, bridging the gap between deep features extracted from deep neural networks (DNNs) and trees’ transparent decision-making processes. Specifically, we utilize a codebook, which embeds the continuous deep features into a finite discrete space (deep visual primitives) to distill the most common semantic information. The decision tree adopts the spatial location information and the mapped primitives to present the decision-making process of the deep features in a tree hierarchy. Moreover, the trained interpretable PrimitiveTree can inversely explain the constituents of the deep features, highlighting the most critical and semantic-rich image patches attributing to the final predictions of the given DNN. Extensive experiments and visualization results validate the effectiveness and interpretability of our method.