一种“KooN”结构的马尔可夫简并建模和简化计算的方法

IEC61508等标准提出了几种计算安全仪表系统需求时平均失效概率的方法,但是,对于具有冗余配置的复杂系统,随着组件数量的增加,系统的中间状态数量快速增长,用户难以构建马尔可夫模型,即便借助计算机来建模运算也较为耗时。提出了一种同型“K oo N”简并状态的马尔可夫建模的通用方法,首先是根据降级状态进行判断,将符合条件的状态进行简并,然后对标记为危险失效状态的概率进行计算。通过严格的理论推导,该简并状态方法可以在不损失精度的前提下简化马尔可夫建模。     

机器翻译译文质量估计综述

机器翻译译文质量估计(Quality Estimation,QE)是指在不需要人工参考译文的条件下,估计机器翻译系统产生的译文的质量,对机器翻译研究和应用具有很重要的价值。机器翻译译文质量估计经过最近几年的发展,取得了丰富的研究成果。该文首先介绍了机器翻译译文质量估计的背景与意义;然后详细介绍了句子级QE、单词级QE、文档级QE的具体任务目标、评价指标等内容,进一步概括了QE方法发展的三个阶段: 基于特征工程和机器学习的QE方法阶段,基于深度学习的QE方法阶段,融入预训练模型的QE方法阶段,并介绍了每一阶段中的代表性研究工作;最后分析了目前的研究现状及不足,并对未来QE方法的研究及发展方向进行了展望。     

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。     

Effect of ultrasonic waveforms on gas–liquid mass transfer in microreactors

This study aims to investigate the effect of ultrasonic waveforms on the gas–liquid mass transfer process. For a given load power (P), continuous rectangular wave yielded stronger bubble oscillation and higher mass transfer coefficient (k_La) than continuous triangular and sinusoidal wave. For pulsed ultrasound, the k_La decreased monotonically with decreasing duty ratio (D), resulting in weak enhancement at low D (≤33%). For a given average load power (P_A), concentrating the P for a shorter period resulted in a higher k_La due to stronger cavitation behavior. For a given P_A and D, decreasing the pulse period (T) led to an increase in k_La, which reached a constant high level when the T fell below a critical value. By optimizing the D and T, a k_La equivalent to 92% of that under continuous ultrasound was obtained under pulsed ultrasound at a D of 67%, saving 33% in power consumption.     

Life cycle assessment of HFC-134a production by calcium carbide acetylene route in China

HFC-134a is a widely used environment-friendly refrigerant. At present, China is the largest producer of HFC-134a in the world. The production of HFC-134a in China mainly adopts the calcium carbide acetylene route. However, the production route has high resource and energy consumption and large waste emission, and few of the studies addressed on the environmental performance of its production process. This study quantified the environmental performance of HFC-134a production by calcium carbide route via carrying out a life cycle assessment (LCA) using the CML 2001 method. And uncertainty analysis by Monte-Carlo simulation was also carried out. The results showed that electricity had the most impact on the environment, followed by steam, hydrogen fluoride and chlorine, and the impact of direct CO₂ emissions in calcium carbide production stage on the global warming effect also could not be ignored. Therefore, the clean energy (e.g., wind, solar, biomass, and natural gas) was used to replace coal-based electricity and coal-fired steam in this study, showing considerable environmental benefits. At the same time, the use of advanced production technologies could also improve environmental benefits, and the environmental impact of the global warming category could be reduced by 4.1% via using CO₂ capture and purification technology. The Chinese database of HFC-134a production established in this study provides convenience for the relevant study of scholars. For the production of HFC-134a, this study helps to better identify the specific environmental hotspots and proposes useful ways to improve the environmental benefits.     

Influence of fluorine-containing side chain on the properties of waterborne polyurethane-urea

In order to prepare waterborne polyurethane with excellent water resistance and thermodynamic properties, a series of side chain fluorinated waterborne polyurethane-urea (FWPU-UA) was synthesized with polytetramethylene ether glycol, N-(2-methyl-1,3-propanediol-2′-)-perfluoro-1-butanesulfonyl amine (NPBA), isophorone diisocyanate, and isophoronediamine. With the increase of NPBA content, the weight loss temperature, glass transition temperature, and tensile strength of FWPU-UA were all improved. Gaussian fitting analysis of infrared data and density functional theory simulation proved that the introduction of fluorine side chains increased the interaction of hydrogen bonding in the FWPU-UA. X-ray photoelectron spectroscopy analysis indicated that the aggregation of fluorine atoms on the surface of film were caused by the migration and enrichment of fluorine side chains. Furthermore, the water resistance of polyurethane-urea film could be significantly improved by adding a small amount of NPBA, and the seven-day water absorption rate of polyurethane-urea film was reduced from 30.13% to 12.55%.