基于3D scSE-UNet的肝脏CT图像半监督学习分割方法

针对分割神经网络需要大量的高质量标签但较难获取的问题,提出基于3D scSE-UNet的半监督学习分割方法. 该方法使用自训练的半监督学习框架,将包含改进的并行空间/特征通道压缩和激励模块（scSE-block+）的3D scSE-UNet作为分割网络. scSE-block+可以从图像空间和特征通道2个方面自动学习图像的有效特征,抑制无用冗余特征,更好地保留图像边缘信息. 在自训练过程中加入全连接条件随机场,对分割网络产生的伪标签进行边缘细化,提升伪标签的精确度. 在LiTS17 Challenge和SLIVER07数据集上验证所提出方法的有效性. 当有标签图像占训练集总图像的30%时,所提方法的Dice相似系数(dice score)为0.941. 结果表明,所提出的半监督学习分割方法可以在仅使用少量标注数据的情况下,取得与全监督分割方法相当的分割效果,有效减轻肝脏CT图像分割对专家标注数据的依赖.     

电能变换领域中随机脉宽调制技术研究概况及展望

相较于传统的脉宽调制技术,随机脉宽调制具备有效的谐波频谱分散能力,从而实现机电负载噪声的削弱。随机脉宽调制技术包括起步的随机开关调制、后续的随机脉冲位置与随机开关频率等3个基本方法。随着电力电子技术的整体发展,随机脉宽调制已从传统的载波调制逐渐拓展到空间矢量调制之中,应用范围日趋广泛,谐波频谱分散与功率谱峰值抑制的效果不断提升。同时,新的优化方案不断被提出,对谐波频谱及功率谱的分散效果亦在不断提高。梳理了随机脉宽调制的研究历程,总结了数种典型的随机脉宽调制的基本原理,并对未来随机脉宽调制的研究方向作出展望,为随机脉宽调制技术的深入研究提供参考。     

The Effects of Random Porosities in Resonant Frequencies of Graphene Based on the Monte Carlo Stochastic Finite Element Model

With the distinguished properties in electronics, thermal conductivity, optical transparence and mechanics, graphene has a powerful potential in nanosensors, nano-resonators, supercapacitors, batteries, etc. The resonant frequency of graphene is an important factor in its application and working environment. However, the random dispersed porosities in graphene evidently change the lattice structure and destroy the integrity and geometrical periodicity. This paper focuses on the effects of random porosities in resonant frequencies of graphene. Monte Carlo simulation is applied to propagate the porosities in the finite element model of pristine graphene. The statistical results and probability density distribution of porous graphene with atomic vacancy defects are computed based on the Monte Carlo finite element model. The results of porous graphene with atomic vacancy defects are compared and discussed with the results of graphene with bond vacancy defects. The enhancement effects of atomic vacancy defects are confirmed in porous graphene. The influences of atomic vacancy defects on displacement and rotation vector sums of porous graphene are more concentrated in local places.     

电网故障处置预案文本中的命名实体识别研究

电网故障处置预案是电网故障处置的重要参考,对电网故障处置预案文本中各类电力设备、名称编号等细粒度的关键实体信息进行抽取,是实现计算机学习理解预案内容并进一步支撑故障处置智能化的重要基础。文中提出一种基于深度学习的电网故障处置预案文本命名实体识别技术,首先采用字向量表征预案文本,然后将注意力机制以及双向长短期记忆网络相结合,有所侧重地提取实体词深层字符特征,最后采用条件随机场求解最优序列化的标注。算例表明:文中所提预案文本命名实体识别模型不依赖人工特征,能够自动高效地提取文本特征,准确识别预案文本中细粒度的实体词,满足预案文本中关键实体信息精确定位和识别的要求。     

磁性纤维随机混合媒质等效电磁参数的计算

为计及强散射偶极子间的相互作用,引入参量εh和μh,利用广义多重散射理论及MATLAB计算机软件对在2 GHz～18 GHz频率下的体积占空比为8%的铁纤维随机混合媒质等效磁导率的实验值进行曲线的拟合,导出了耦合系数εh和μh,并在此基础上,对体积占空比分别为30%和20%的铁纤维随机混合媒质等效磁导率进行了计算,计算结果与实验吻合良好.     

汽车盘式制动衬片热压模设计

分析了汽车盘式制动嚣零件的工艺性，对其成形制造工艺、衬片模具结构及其设计、衬片模具的选材及主要零部件的制造加工要点、衬片模具的工作过程及使用情况等作了系统地介绍。     

湿式离合器/制动器及其在机械压力机中的应用

根据多年在汽车制造业中规划，使用和维修机械压力机的经验，介绍了湿式离合器／制动器在机械压力机中的应用现状及其与气动离合器／制动器在使用性能方面的比较。     

全级配混凝土单轴疲劳特性细观数值模拟

为研究全级配混凝土在等幅、变幅及阶梯重复荷载作用下的疲劳特性,揭示材料特性与疲劳应力水平对其疲劳寿命的影响规律,通过自编程序建立二维随机骨料模型,运用有限元软件和耐久性疲劳分析软件进行联合仿真,对全级配混凝土细观结构进行应力分析和疲劳分析,探讨了不同骨料分布、骨料形状及疲劳应力水平对混凝土疲劳性能的影响,并估算了疲劳寿命。结果表明:在阶梯重复荷载作用下,骨料形状和分布对混凝土疲劳寿命影响较小,而疲劳应力水平对疲劳寿命影响较大;圆形和椭圆形骨料混凝土疲劳寿命较高,多边形骨料寿命较低;在疲劳加载破坏过程中,骨料界面先起裂,然后裂纹逐渐扩展贯通,最后导致整个试件完全破坏,可见骨料界面是混凝土中的薄弱部位,其寿命直接影响整个混凝土试件的寿命。研究成果为后续混凝土疲劳性能研究提供了参考。     

The properties of equimolar copolycarbonates of bisphenol A with bisphenol S and 1,4-bis(hydroxymethyl)cyclohexane

Ranges of both alternating and random 1:1 copolycarbonates of Bisphenol-A (2,2-bis(4-hydroxyphenyl)propane) and Bisphenol-S (4,4′-dihydroxydiphenyl sulphone) and of Bisphenol-A and 1,4-dimethanoylcyclohexane have been synthesized in order to test the hypothesis that, other things being equal, a more random structured polymer should exhibit superior thermal ageing characteristics. Studies on these materials' mechanical and thermal properties establish that within each pair they are extremely similar and hence the original hypothesis is disproved.     

Synthesis of tertiary poly(amido-amine)s with amido- and amino-groups randomly arranged along the macromolecular chain

A new synthesis of poly(amido-amine)s from bis (secondary amine)s and acryloyl chloride has been studied, leading to polymers in which the amido- and amino-groups are randomly arranged along the macromolecular chain. A poly(amido-amine) based on piperazine, chosen as the mother compound, has been fully characterized by ¹³C n.m.r. as well as by identification and quantitative evaluation of its hydrolysis products.