基于改进MobileFaceNet的人脸识别方法北大核心

针对目前大多数人脸识别算法参数多、计算量大,难以部署到移动端和嵌入式设备中的问题,提出了一种基于改进MobileFaceNet的人脸识别方法。通过对MobileFaceNet模型结构的调整,将bottleneck模块优化为sandglass模块,改良深度卷积和逐点卷积的相对位置,适当增大sandglass模块的输出通道数,从而减少特征压缩时的信息丢失,增强人脸空间特征的提取。实验结果表明:改进后的方法在LFW测试数据集上准确率达99.15%,模型大小和计算量分别仅为原算法的61%和45%,验证了所提方法的有效性。     

爆破位置对深凹露天矿山炮烟扩散的影响研究

深凹露天矿山由于其特殊的结构,爆破产生的炮烟扩散稀释较为困难,严重危害生产作业人员的生命安全与健康。基于实际矿山构建了深凹露天矿山的二维物理及数学模型,采用非稳态数值分析方法研究了不同爆破位置下,深凹露天矿山采坑内爆破炮烟的扩散规律。研究结果表明：不同爆破位置下,露天采坑内均出现复环流,爆破点位置是影响露天采坑内风流结构特征的重要因素；露天采坑内的炮烟最高浓度均随着时间变化而逐渐下降,但下降的速率逐步减小,呈现三个阶段的下降趋势；爆破位置位于背风侧时露天采坑内的炮烟最高浓度和降至安全浓度所需时间远高于迎风侧三个爆破位置；随着背风侧爆破点距采坑底部距离的减小,炮烟最高浓度及降至安全浓度所需时间先降低后增加,炮烟最高浓度及降至安全浓度所需时间随着迎风侧爆破位置距采坑底部距离的减小而增加。研究结果对于指导深凹露天矿山企业合理组织爆破后的生产作业和保障作业人员安全具有重要意义。     

Influence of high energy ball milling on structural,microstructural and optical properties of TiO2 nanoparticles

The influence of high-energy ball milling on structural, microstructural, and optical properties of TiO₂ by modifying the nanoparticle size was studied. Five samples were extracted at different milling times (0, 2, 4, 8, and 13 h). The average particle sizes estimated by dynamic light scattering (DLS) were 205, 155.8, 116.8, 82.9, and 82.7 nm at 0, 2, 4, 8, and 13 h, respectively. X-ray diffraction analysis confirmed progressive broadening of the peaks as the milling time elapsed. Besides, a correlation was found between d spacing and the average crystal size. The UV–Vis diffuse reflectance spectra of TiO₂ revealed a decrease in reflectance due to particle size reduction. Similarly, an alteration of the bandgap transition energy was presented, whose values gradually decreased from 2.966 eV to 2.861 eV for the sample without and with the maximum duration milling performed (13 h), respectively. Likewise, the SEM analysis showed a distribution in nanoparticle size that became more homogeneous and smaller average grain size as the milling duration was longer.     

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。     

高耸桥墩保护性爆破拆除关键技术

为解决高耸钢砼桥墩上部倾斜导致的工程质量不合格问题,采用保护性控制爆破将上部倾斜部分钢砼桥墩予以拆除,同时保留下部未偏斜部分桥墩不受破坏。采用开设组合形状切口,预开定向窗、导向窗;设置组合高差卸荷槽,预伤钢筋弱化抗压抗拉能力为关键措施的控制爆破方案。方案实施取得了十分满意的效果,上部拟拆除部分桥墩顺利倾倒下坠,下部保留部分桥墩完好无损。采用保护性控制爆破技术能高效、安全地解决类似高耸钢砼桥墩(构筑物)部分拆除、部分保留利用的问题,并能取得可观的经济、社会效益。     

Anchor plate bearing capacity in flexible mesh facings

This work addresses the problem of the loading capacity of an anchor plate coupled with a steel wire mesh in soil retaining applications. The interaction mechanism between the flexible mesh facing, the underlying soil layer and the plate is studied starting from the results of several laboratory punch tests involving both the plate and the mesh only, and the whole soil-mesh-plate system. The experimental tests have been reproduced by adopting a 3D discrete element model where also the wire mesh is discretized as an assembly of interconnected nodal particles. The interaction between these particles is ruled by elasto-plastic tensile force–displacement laws in which a distortion is introduced in a stochastic manner to account for the wires’ geometrical irregularities. The mesh model is then validated with reference to a set of punch tests in which the shape and size of the punching element as well as the nominal wire diameter were varied. Subsequently, the model is extended to a punch against soil test configuration permitting an insight into the nontrivial local mechanism between the mesh facing and the underlying granular layer. The good agreement between the numerical predictions and the experimental observations at the laboratory scale allowed us to extend the model towards more realistic field conditions for which the role of the mesh panel boundary conditions, the mesh mechanical properties, the soil mechanical properties and the anchor plate geometry is investigated.     

Adaptive investigation of the optical properties of polymer fibers from mixing noisy phase shifting microinterferograms using deep learning algorithms

In this article, an adaptive denoising method is suggested to accurate investigate the optical and structural features of polymeric fibers from noisy phase shifting microinterferograms. The mixed class of noise that may produce in the phase-shifting interferometric techniques is established. To our knowledge, this is an early study considered the mixing noises that may occur in microinterferograms. The suggested method utilized the convolution neural networks to detect the noise class and then denoising, it according to its class. Four convolution neural networks (Googlenet, VGG-19, Alexnet, and Alexnet–SVM) are refined to perform the automatic classification process for the noise class in the established data set. The network with the highest validation and testing accuracy of these networks is considered to apply the suggested method on realistic noisy microinterferograms for polymeric fibers, polypropylene and antimicrobial polyethylene terephthalate)/titanium dioxide, recoded using interference microscope. Also, the suggested method is applied on noisy microinterferograms include crazing and nanocomposite material. The demodulated phase maps and the three-dimensional birefringence profiles are calculated for tested fibers according to the suggested method. The obtained results are compared with the published data for these fibers and found to be in good agreements.     

Deep Learning Based Process Analytics Model for Predicting Type 2 Diabetes Mellitus

Process analytics is one of the popular research domains that advanced in the recent years. Process analytics encompasses identification, monitoring, and improvement of the processes through knowledge extraction from historical data. The evolution of Artificial Intelligence (AI)-enabled Electronic Health Records (EHRs) revolutionized the medical practice. Type 2 Diabetes Mellitus (T2DM) is a syndrome characterized by the lack of insulin secretion. If not diagnosed and managed at early stages, it may produce severe outcomes and at times, death too. Chronic Kidney Disease (CKD) and Coronary Heart Disease (CHD) are the most common, long-term and life-threatening diseases caused by T2DM. Therefore, it becomes inevitable to predict the risks of CKD and CHD in T2DM patients. The current research article presents automated Deep Learning (DL)-based Deep Neural Network (DNN) with Adagrad Optimization Algorithm i.e., DNN-AGOA model to predict CKD and CHD risks in T2DM patients. The paper proposes a risk prediction model for T2DM patients who may develop CKD or CHD. This model helps in alarming both T2DM patients and clinicians in advance. At first, the proposed DNN-AGOA model performs data preprocessing to improve the quality of data and make it compatible for further processing. Besides, a Deep Neural Network (DNN) is employed for feature extraction, after which sigmoid function is used for classification. Further, Adagrad optimizer is applied to improve the performance of DNN model. For experimental validation, benchmark medical datasets were used and the results were validated under several dimensions. The proposed model achieved a maximum precision of 93.99%, recall of 94.63%, specificity of 73.34%, accuracy of 92.58%, and F-score of 94.22%. The results attained through experimentation established that the proposed DNN-AGOA model has good prediction capability over other methods.     

Physical and electrochemical behavior of affordable Na3SbS4 solid electrolyte at different heat treatment temperatures

High-capacity and affordable all-solid-state Na-ion batteries have gathered increasing interest in recent years owing to low-cost sodium, which contributes to reducing the price of these Na-ion batteries to approximately 70% of that in lithium batteries. However, in terms of electrolyte performance and battery cost, the complete replacement of lithium batteries has a long way to go. In this work, low-cost and high-safety Na₂S·9H₂O materials are used in synthesizing Na₃SbS₄ solid electrolyte, the price of which is only one-fifth that of high-purity Na₂S. The structure and electrochemical properties are studied through X-ray diffraction analysis, Raman spectroscopy, scanning electron microscopy, and electrochemical tests. Results indicate that a multiphase Na₃SbS₄ structure containing cubic and tetragonal phases formed after heat treatment at 300 °C. In addition, a third phase transition of Na₃SbS₄ is inferred after further heating at 600 °C. This phase structure contributes to the improvement of electrochemical performance by promoting increasing ionic conductivity to 0.54 mS cm^?1 at room temperature (25 °C) and reducing activation energy to 0.076 eV. This work provides an affordable material with good electrochemical properties and not only simplifies the preparation but also greatly reduces the risk of the process.     

低共熔溶剂用于CO2捕集的研究进展

低共熔溶剂(DESs)具有原料廉价易得、化学稳定性好、可设计、合成工艺简单、可循环使用和绿色环保等优点,在CO₂捕集领域受到广泛关注。重点综述了近年来DESs用于CO₂捕集的研究进展,总结了DESs捕集CO₂的能力,分析了DESs捕集CO₂的影响因素和DESs的循环使用性能,归纳了DESs捕集CO₂的机理(包括物理吸收、化学吸收和物理化学协同吸收),并总结了CO₂在DESs中的溶解度计算模型。分析发现,DESs捕集CO₂的影响因素中,温度、压力和水含量(质量分数)均对CO₂的捕集有影响,且DESs的结构是重要的影响因素;大部分DESs可循环使用;CO₂在DESs中溶解度计算模型的建立有效推动了DESs捕集CO₂的进一步发展。最后,指出了DESs捕集分离CO₂所面临的主要问题并对进一步的研究工作进行了分析讨论。