检索结果-南通市图书馆

Interdisciplinary Materials 2023年第1期2卷 53-90页

作者： qianli Ma Shichun Mu Fuel cell Materials Laboratory Foshan Xianhu LaboratoryFoshanChina State Key Laboratory of Advanced Technology for Materials Synthesis and Processing Wuhan University of TechnologyWuhanChina

As the most desirable hydrogen production device,the highly efficient acidic proton exchange membrane water electrolyzers(PEMWE)are severely limited by the sluggish kinetics of oxygen evolution reaction(OER)at the *** IrO2 is a commercial acid-stable OER catalyst with poor activity and high cost,which has motivated the development of ***,hitherto most of the designed acidic OER catalysts have disadvantages of low activity or stability,which cannot meet the requirement of industrial ***,exploring suitable strategies to enhance the activity and stability of cost-effective acidic OER catalysts is crucial for developing the PEMWE *** this review,the main OER mechanisms,different types of catalysts,and their activity and stability characteristics are summarized and discussed,and then possible strategies to improve activity and stability are ***,the problems and prospects of such catalysts are generalized to shed some light on the future research of advanced catalysts for acidic OER.

关键词： catalyst hydrogen energy oxygen evolution reaction proton exchange membrane water electrolyzer water oxidation

Characteristics of Hydrogen Storage Alloy Mg_2Ni Produced by Hydriding Combustion Synthesis

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Materials Science & Technology 2004年第2期20卷 209-212页

作者： qianli QinLI LijunJIANG Kou-chihCHOU FengZHAN QiangZHENG ResearchCenterofEnergyMaterialsandTechnology GeneralResearchInstituteforNonferrousMetalBeijing100088China DepartmentofPhysicalChemistry UniversityofScienceandTechnologyBeijingBeijing100083China

A high activity and large capacity of hydrogen storage alloy Mg2Ni by hydriding combustion synthesis was investigated by means of pressure composition isotherms, X-ray diffraction and scanning electron microscopy. The results showed that the maximum hydrogen absorption capacity of Mg2Ni is 3.25 mass fraction at 523 K, just after synthesis without any activation. The relationships between the equilibrium plateau pressure and the temperature for Mg2Ni were lgp (0.1 MPa)=-3026/T+5.814 (523 K≤T≤623 K) for hydriding and Igp (0.1 MPa)=-3613/T+6.715 (523 K≤T ≤623 K) for dehydriding. The kinetic equation is [-ln(1-a)]3/2 = kt and the apparent activation energy for the nucleation and growth-controlled hydrogen absorption and desorption were determined to be 64.3±2.31kJ/(mol.H2) and 59.9±2.99kJ/(mol.H2)respectively.

关键词： Hydriding combustion Mg2Ni Hydrogen storage property Apparent activation energy

Dynamic soil arching in piled embankment under train load of high-speed railways

在线全文

学校读者我要写书评

暂无评论

Earthquake Engineering and Engineering Vibration 2023年第3期22卷 719-730页

作者： Niu Tingting Yang Yule Ma qianli Zou Jiuqun Lin Bin School of Civil Engineering and Architecture Anhui University of Science and TechnologyHuainan 232001China College of Civil and Transportation Engineering Hohai UniversityNanjing 210098China

Piled embankments have many advantages that have been applied in high-speed railway construction ***,the load transfer mechanism of piled embankments,such as soil arching and tension membranes,is still unclear,especially under dynamic *** investigate the soil arching and tension membrane under dynamic train loads on high-speed railways,a large-scale piled embankment model test with X-shaped piles as vertical reinforcement was performed,in which twenty-eight earth pressure cells were installed in the piled embankment and an M-shaped wave was adopted to simulate the high-speed railway train *** results show that dynamic soil arching only occurs when two bogies of a carriage pass by and disappears at other *** dynamic soil arching and membrane effect are the most significant under the concrete *** arching height,stress concentration ratio and pile-soil load sharing ratio have a minimal value at 25 *** dynamic soil arching degrades severely at 25 Hz,whose height at 25 Hz is only 0.35 times that at 5 *** arching height fluctuates over a narrow range with increasing loading *** stress concentration ratio and the pile-soil load sharing ratio increase monotonically as the loading amplitude increases.

关键词： dynamic soil arching membrane effect piled embankment train load model test

Self-assembly of biomimetic light-harvesting complexes capable of hydrogen evolution

在线全文

学校读者我要写书评

暂无评论

Green Energy & Environment 2017年第1期2卷 58-63页

作者： Kai Liu Manzar Abass qianli Zou Xuehai Yan State Key Laboratory of Biochemical Engineering Institute of Process Engineering Chinese Academy of Sciences University of Chinese Academy of Sciences Center for Mesoscience Institute of Process EngineeringChinese Academy of Sciences

Biomimetics provides us a new perspective to understand complex biological process and strategy to fabricate functional materials. However,a great challenge still remains to design and fabricate biomimetic materials using a facile but effective method. Here, we develop a biomimetic light harvesting architecture based on one-step co-assembly of amphiphilic amino acid and porphyrin. Amphiphilic amino acid can self-assemble into nanofibers via π-stacking and hydrogen binding interactions. Negatively charged porphyrin adsorbs on the surface of the assembled nanofibers through electrostatic force, and the nanofibers further organize into porous urchin-like microspheres induced presumably by hydrophobic interaction. The assembled amphiphilic amino acid nanofibers work as a template to tune the organization of porphyrin with an architecture principle analogous to natural light harvesting complex. The co-assembled microspheres exhibit enhanced light capture due to the light reflection in the porous structure. Reaction center(platinum nanoparticles) can be effectively coupled with the light harvesting microspheres via photoreduction. After visible light illumination, hydrogen evolution occurs on the hybrid microspheres.

关键词： Light-harvesting Amino acid Porphyrin Co-assembly Hydrogen evolution

A survey on ensemble learning

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2020年第2期14卷 241-258页

作者： Xibin DONG Zhiwen YU Wenming CAO Yifan SHI qianli MA School of Computer Science and Engineering South China University of TechnologyGuangzhou 510006China Department of Computer Science City University of Hong KongHong Kong SAR 999077China

Despite significant successes achieved in knowledge discovery,traditional machine learning methods may fail to obtain satisfactory performances when dealing with complex data,such as imbalanced,high-dimensional,noisy data,*** reason behind is that it is difficult for these methods to capture multiple characteristics and underlying structure of *** this context,it becomes an important topic in the data mining field that how to effectively construct an efficient knowledge discovery and mining *** learning,as one research hot spot,aims to integrate data fusion,data modeling,and data mining into a unified ***,ensemble learning firstly extracts a set of features with a variety of *** on these learned features,multiple learning algorithms are utilized to produce weak predictive ***,ensemble learning fuses the informative knowledge from the above results obtained to achieve knowledge discovery and better predictive performance via voting schemes in an adaptive *** this paper,we review the research progress of the mainstream approaches of ensemble learning and classify them based on different *** addition,we present challenges and possible research directions for each mainstream approach of ensemble learning,and we also give an extra introduction for the combination of ensemble learning with other machine learning hot spots such as deep learning,reinforcement learning,etc.

关键词： ensemble learning supervised ensemble classification semi-supervised ensemble classification clustering ensemble semi-supervised clustering ensemble

Dynamics in Deep Classifiers Trained with the Square Loss:Normalization,Low Rank,Neural Collapse,and Generalization Bounds

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Research 2023年第3期2023卷 69-89页

作者： Mengjia Xu Akshay Rangamani qianli Liao Tomer Galantil Tomaso Poggio Center for Brains Minds and MachinesMassachusetts Institute of TechnologyCambridgeMAUSA Division of Applied Mathematics Brown UniversityProvidenceRIUSA

We overview several properties—old and new—of training overparameterized deep networks under the square *** first consider a model of the dynamics of gradient flow under the square loss in deep homogeneous rectified linear unit *** study the convergence to a solution with the absolute minimumρ,which is the product of the Frobenius norms of each layer weight matrix,when normalization by Lagrange multipliers is used together with weight decay under different forms of gradient descent.A main property of the minimizers that bound their expected error for a specific network architecture isρ.In particular,we derive novel norm-based bounds for convolutional layers that are orders of magnitude better than classical bounds for dense ***,we prove that quasi-interpolating solutions obtained by stochastic gradient descent in the presence of weight decay have a bias toward low-rank weight matrices,which should improve *** same analysis predicts the existence of an inherent stochastic gradient descent noise for deep *** both cases,we verify our predictions *** then predict neural collapse and its properties without any specific assumption—unlike other published *** analysis supports the idea that the advantage of deep networks relative to other classifiers is greater for problems that are appropriate for sparse deep architectures such as convolutional neural *** reason is that compositionally sparse target functions can be approximated well by“sparse”deep networks without incurring in the curse of dimensionality.

关键词： bounds stochastic decay

Characteristics of Particulate Matter Emissions for Low-Sulfur Heavy Oil Used in Low-Speed Two-Stroke Diesel Engines of Ocean-Going Ships

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Thermal Science 2024年第2期33卷 739-750页

作者： WU Gang MA qianli WEI Lijiang JIANG Guohe WANG Tengfei LI Tie Merchant Marine College Shanghai Maritime UniversityShanghai201306China State Key Laboratory of Ocean Engineering Shanghai Jiao Tong UniversityShanghai200240China

In this work,particulate matter(PM) emissions from a large two-stroke,low-speed marine diesel engine were investigated when the engine was operated with low-sulfur heavy fuel oil(HFO) at various *** samples were collected in situ from the engine exhaust to determine the detailed physical and chemical *** nanostructure and morphology of the nanoparticles were analyzed using transmission electron microscopy images(TEM).The results show that volatile organic carbon(OC) accounts for more than 80% in the HFO particles and leads to an increase in particle *** thermodynamic conditions of a low-speed engine favor the behavior of capturing the soluble organic components.A large number of spherical char HFO particles with aerodynamic diameters of 0.2 μm-0.5 μm and a suspected inner metal core were *** two peak aerodynamic diameters of the HFO nanoparticles are 15 nm and 86 *** morphological differences among the HFO nanoparticles in varied engine conditions represent the formation process from primary nascent particles to mature graphitized particles caused by *** above study will be valuable for understanding the characteristics of PM emissions from low-sulfur HFO to achieve the ship PM emissions reduction target.

关键词： low-speed diesel engine maritime regulations heavy fuel oil TEM particles size and distribution

The Key Technology on Chinese Word Segmentation Based on Bi-LSTM-CRF Model

在线全文

学校读者我要写书评

暂无评论

The Key Technology on Chinese Word Segmentation Based on Bi-...

作者： qianli Ma 华中师范大学

学位级别：硕士

自然语言处理是人工智能领域的核心技术之一,而自然语言处理中的汉语分词是自然语言处理的基础。目前主流的分词方法是基于传统的机器学习模型。近年来,随着人工智能浪潮的再次兴起,无论是短期记忆还是长期记忆的LSTM神经网络模型都对... 详细信息

自然语言处理是人工智能领域的核心技术之一,而自然语言处理中的汉语分词是自然语言处理的基础。目前主流的分词方法是基于传统的机器学习模型。近年来,随着人工智能浪潮的再次兴起,无论是短期记忆还是长期记忆的LSTM神经网络模型都对传统的循环神经网络模型进行了改进,该模型能长期依赖于信息的不足,并广泛应用于自然语言处理各种任务,取得了良好的效果。作为自然语言处理模块的重要组成部分,中文分词的好坏会影响到基于自然语言处理的相关研究,中文分词盼段将会影响相关的精度,因此中文分词阶段的精度将会其他例如语义和语法层次以及应用层次,分词的精度是衡量这一阶段的重要指标。在中文分词的精度提升上,必须解决如下几个关键问题:1.在中文分词中3如何拥有更高的识别精度是该自然语言处理子方向的首要解决的问题。不仅仅通过该词在句子中的位置来预測分词结果往往是不够的,而对上下文进行依赖,通过句子位置和上下文之间的关系进行分词更加的具有研究价值和重要意义。2.传统的神经模型LSTM模型已经证明在处理信息学习有着相当大的优势,但是LSTM神经网络的前向传导的特点,在训练过程中会导致部分后文的关系并没有被考虑学习,解决全文的上下文关系的学习,防止文本依赖不足所导致的问题也是需要研究的方向。3.对于根据上下文之间的潜在关系进行研究所得出来的结果是基于非规则的研究,但这些上下文关系有的时候并不以一定具有普遍的确定性,这样会导致尽管上下文有联系但是会出现一些前词和后词逻辑性并没有关联的结果,如何增加中文分词的逻辑性对提升中文分词精度有相当的理论研究价值。4.对每一个通过神经网络进行模型都有者自己最佳的参数学习方式,以此来达到局部最优解,对模型相关的参数需要通过不断调整来达到较优以接近最优值。通过上述对问题的阐述,本文在解决上述问题的主要贡献是（1）首先对经典的单向LSTM模型进行了改进,通过增加正向LSTM层并设针双向LSTM模型。这样通过对文本进行前向和后向的传导,以此来改善单向LSTM对以下文本的依赖不足。（2）在神经网络通过添加CRF层,增加了单词间的约束。通过使用CRF所具有的转移矩阵特性来学习单词之间的逻辑,提高因逻辑错误所导致的分词结果精度降低。（3）通过验证词向量嵌入的必要性和设置最佳Dropout防止过拟合,以及动态调整学习率等参数设置来达到模型最佳的精度。在整体的设计上,首先在传统神经网络的基础上,对增加的字向量进行改进,利用字嵌入预处理,然后扩展到LSTM层的双向LSTM,并通过整个连接增加了CRF层的密集度。本文利用Bakeoff提供的SIGAN2006会议进行了主要对比实验。通过实验,基于本文的BI-LSTM-CRF模型取得了较好的效果。通过北京大学的数据集PKU和微软的数据集MSA，准确率和查全率都达到了很好的效果。在机器学习方面,条件随机场可以达到很好的分词效果,在深度学习方面,分词效果得到了更明显的提高。本文主要通过结合双向记忆相关神经网络模型的优化,原LSTM单元结构的神经网络每时每刻、每一步都在执行一种记忆机制,但每一个计算量随着句子长度的增加而激增,导致分词阶段的任务效果较差,逻辑问题较多。本文的改进是词向量用于维度预处理,然后双向段记忆神经深度学习神经网络模型是用来进行隐藏的多个循环传输层,从而达到最好的中文分词的效果。该方法较好地解决了单隐层、防止过拟合以及卷积神经网络不能处理长依赖序列信息的缺点。同时,双向短记忆和长记忆依赖神经网络解决了简单循环神经网络长期依赖和单向LSTM无法利用方向序列信息的问题。经过softmax和dense层处理后,输出将是每个单词所携带的标签的概率。为了解决一些单词没有逻辑的问题,将输出放入CRF层进行传输矩阵处理。最后,合并后的BI-LSTM-CRF模型可以更好地从语义角度进行分词。在最终的实验中,在预实验中人民日报1998语料库的数据集P值为95.3359%,R值为95.1864%,F1值为95.3359%。人民日报2014语料库的数据集P值为96.7524%,R值为94.7981%,F1值为95.7653%。国家数据集P值为93.1640%,R值为93.2534%,F1值为93.2087%。在最终对比实验数据中,在微软的MSRA数据集,通过F1值的比较,通过F1值的比较,可以看出bi-lstm-crf模型的F1值约为2.9%高于CRF++模型,和高于LSTM模型0.9%,bilstm模型高出0.8%,高于Istm-crf模型0.8%。从p值和r值可以看出,基于MSRA数据集的bi-lstm-crf模型比其他模型工作得更好。在PKU数据集中,在P值比较中,bi-lstm-CRF模型比LSTM-CRF模型低0.2%,但比CRF++模型高0.7%左右,比LSTM模型高0.4%左右,比bilstm模型高0.4

关键词： BI-LSTM-CRF模型中文分词机器学习神经网络条件随机场

Interconnected Ni(OH)_(2)and polyether amine on carbon fiber surface for simultaneously improving interfacial adhesion of epoxy composites and electrochemical properties

同方学位论文库评论

在线全文

同方学位论文库

学校读者我要写书评

暂无评论

Chinese Journal of Aeronautics 2022年第5期35卷 419-428页

作者： Qing WU Qinqin WAN qianli LIU Fen WANG Xin LIU Jianfeng ZHU School of Materials Science and Engineering Shaanxi Key Laboratory of Green Preparation and Functionalization for Inorganic MaterialsShaanxi University of Science&TechnologyXi’an 710021China Shanghai Composites Science&Technology Co. Ltd.Shanghai 201112China

Interconnected Ni(OH)_(2)nanoflakes and polyether amine(PEA)were deposited on carbon fiber tows via a facial and effective process of chemical bath deposition and dip *** on this,a win–win benefit of simultaneously improvements in interfacial shear strength(IFSS)of carbon fiber/epoxy composites and the electrochemical activity has been *** with CF and CF-Ni(OH)_(2)composites,the IFSS of CF-Ni(OH)_(2)-PEA/epoxy composite respectively increased 7.9%and 45.4%,which was put down to the covalent bonding of Ni(OH)_(2)-PEA coating with fiber and epoxy matrix,as well as the effective stress transfer by the uniform honeycomb structure of Ni(OH)_(2).In aqueous KOH electrolyte,the CF-Ni(OH)_(2)-PEA electrode presented the maximum specific capacitance of 689.98 F·g^(-1)at 5 m V·s^(-1),572.28 F·g^(-1)at a current density of 0.5 A·g^(-1)due to the strong adhesion of carbon fiber with Ni(OH)_(2)by PEA,the reservation of the threedimensional hollow honeycomb structure of Ni(OH)_(2)for easy ion-transport and–NH_(2)functional groups from PEA for providing more active *** excellent performance of CF-Ni(OH)_(2)-PEA reinforcement demonstrates its promising potential for application in high performance composites with integrated structure and function,which shows great advantages in various fields of aerospace,energy,electronics,automobile,civil engineering,sports,etc.

关键词： Carbon fiber Electrochemical properties Interface/interphase Polymer-matrix composites Surface modification