The chain/wire rope/chain combination is a common choice for mooring offshore floating platforms. However, data of the drag coefficients of chain links are rather limited, resulting in uncertainties with the calculati...The chain/wire rope/chain combination is a common choice for mooring offshore floating platforms. However, data of the drag coefficients of chain links are rather limited, resulting in uncertainties with the calculations of the drag force, and hence the damping of the mooring system. In this paper, the importance of the selection of the drag coefficient is first investigated. The computational fluid dynamics(CFD) method is then used to determine the drag coefficients of a studless chain under steady flows. Numerical model validation is first completed by simulating a smooth circular cylinder under steady flows. In particular, the performance of different turbulence models is assessed through the comparisons between the calculations and the experimental results. The large eddy simulation(LES) model is finally selected for the simulation of steady flows past a chain. The effects of the Reynolds number on the drag coefficient of a stud-less chain is also studied. The results show that the calculated drag coefficients of a stud-less chain are fairly consistent with the available experimental data.展开更多
In October,the national consumption market continued to maintain the momentum of rapid growth.Retail sales of national social consumer goods in October amounted to RMB 3.11 trillion,up10%year on year.From January to O...In October,the national consumption market continued to maintain the momentum of rapid growth.Retail sales of national social consumer goods in October amounted to RMB 3.11 trillion,up10%year on year.From January to October,retail sales of national social consumer goods increased 10.3%,the same growth as that of the first three quarters.According to the monitoring of the MOFCOM,in October,5,000major retail businesses increased 4.6%展开更多
To investigate the chemical structure of cell wall mannan obtained from pathogenic yeast, Candida tropicalis NBRC 1400 (former antigenic standard strain, IFO 1400). As a result of two-dimensional NMR analysis, it was ...To investigate the chemical structure of cell wall mannan obtained from pathogenic yeast, Candida tropicalis NBRC 1400 (former antigenic standard strain, IFO 1400). As a result of two-dimensional NMR analysis, it was shown that the mannan of this strain is composed of α-1,6-, α-1,3-, α-1,2- and β-1,2-linked mannose residues. In this research, the mannan was subjected to three degradation procedures, acid-treatment, α-mannosidase, and acetolysis under two conditions in order to determine the chemical structure of the antigenic oligomannosyl side chains in this molecule. The 1H-nuclear magnetic resonance spectra of resultant oligosaccharides, pentaose and hexaose, demonstrated the existence of the oligomannosyl side chains corresponding to Manα1-3Manα1-2Manα1-2Manα1-2Man and Manα1-3Manα1-2Manα1-2Manα1-2Manα1-2Man, respectively, which have previously also been found in Candida albicans serotype A strain mannans. These findings indicate that C. tropicalis and C. albicans serotype A have no significant difference in the chemical structure of these cell wall mannans. Therefore, it can be interpreted that it is extremely difficult to distinguish both species by targeting the antigenic group in these mannans.展开更多
The isomorphism of planar kinematic chains is studied. The original methods for de- tecting isomorphism is discussed. A new matrix, the link -link path matrix, is presented to show the structures of planar kinematic c...The isomorphism of planar kinematic chains is studied. The original methods for de- tecting isomorphism is discussed. A new matrix, the link -link path matrix, is presented to show the structures of planar kinematic chains. It is simple and reliable to detect isomorphism of planar kinematic chains by using the link-link path matrix.展开更多
研究旨在让机器自动生成符合人类提问习惯的阶梯式问句链,辅助智能阅读提问、智能客服、智能医疗问诊等领域对于自然语言理解与生成任务的实际需求,提出一种基于知识图谱关系预测的阶梯式问句链生成模型(stepped question chain generat...研究旨在让机器自动生成符合人类提问习惯的阶梯式问句链,辅助智能阅读提问、智能客服、智能医疗问诊等领域对于自然语言理解与生成任务的实际需求,提出一种基于知识图谱关系预测的阶梯式问句链生成模型(stepped question chain generator based on knowledge graph link prediction,SQCG-KGLP)。SQCG-KGLP模型包括编码和解码两部分,在编码部分,对问句链的问句实体进行特征初始化,并通过特征融合方法获得问句链的融合头实体和待测尾实体的初始向量。将融合头实体和待测尾实体的初始向量送入SQCG-KGLP模型的Graph Attention图表示学习模块中,从而获得融合头实体和待测尾实体的表示向量。在解码部分,将融合头实体以及待测尾实体的表示向量输入到SQCG-KGLP模型的convKB模块中进行链接预测。在自建问句链数据集上进行不同跳数的实验,以MRR、Hits@1、Hits@3、Hits@10作为评测指标。结果表明,SQCG-KGLP算法均优于基线模型。当问句链生成迭代次数增加时,支持关系预测的知识图谱跳数随之增加,下一个节点问句语义信息表达的自然性会随之下降。此外,在现实应用场景中较难获得足量的高跳数问句训练数据集。目前,该研究仅能支持三跳以下的问句链生成。利用知识图谱的语义关联及知识图谱关系预测机制,生成的阶梯式问句链具有层层递进、由易到难、环环相扣的特点,使问句链的阶梯式逻辑质量更优。展开更多
基金financial support for the PhD study from GL-Nobel Denton based in London
文摘The chain/wire rope/chain combination is a common choice for mooring offshore floating platforms. However, data of the drag coefficients of chain links are rather limited, resulting in uncertainties with the calculations of the drag force, and hence the damping of the mooring system. In this paper, the importance of the selection of the drag coefficient is first investigated. The computational fluid dynamics(CFD) method is then used to determine the drag coefficients of a studless chain under steady flows. Numerical model validation is first completed by simulating a smooth circular cylinder under steady flows. In particular, the performance of different turbulence models is assessed through the comparisons between the calculations and the experimental results. The large eddy simulation(LES) model is finally selected for the simulation of steady flows past a chain. The effects of the Reynolds number on the drag coefficient of a stud-less chain is also studied. The results show that the calculated drag coefficients of a stud-less chain are fairly consistent with the available experimental data.
文摘In October,the national consumption market continued to maintain the momentum of rapid growth.Retail sales of national social consumer goods in October amounted to RMB 3.11 trillion,up10%year on year.From January to October,retail sales of national social consumer goods increased 10.3%,the same growth as that of the first three quarters.According to the monitoring of the MOFCOM,in October,5,000major retail businesses increased 4.6%
文摘To investigate the chemical structure of cell wall mannan obtained from pathogenic yeast, Candida tropicalis NBRC 1400 (former antigenic standard strain, IFO 1400). As a result of two-dimensional NMR analysis, it was shown that the mannan of this strain is composed of α-1,6-, α-1,3-, α-1,2- and β-1,2-linked mannose residues. In this research, the mannan was subjected to three degradation procedures, acid-treatment, α-mannosidase, and acetolysis under two conditions in order to determine the chemical structure of the antigenic oligomannosyl side chains in this molecule. The 1H-nuclear magnetic resonance spectra of resultant oligosaccharides, pentaose and hexaose, demonstrated the existence of the oligomannosyl side chains corresponding to Manα1-3Manα1-2Manα1-2Manα1-2Man and Manα1-3Manα1-2Manα1-2Manα1-2Manα1-2Man, respectively, which have previously also been found in Candida albicans serotype A strain mannans. These findings indicate that C. tropicalis and C. albicans serotype A have no significant difference in the chemical structure of these cell wall mannans. Therefore, it can be interpreted that it is extremely difficult to distinguish both species by targeting the antigenic group in these mannans.
文摘The isomorphism of planar kinematic chains is studied. The original methods for de- tecting isomorphism is discussed. A new matrix, the link -link path matrix, is presented to show the structures of planar kinematic chains. It is simple and reliable to detect isomorphism of planar kinematic chains by using the link-link path matrix.
文摘研究旨在让机器自动生成符合人类提问习惯的阶梯式问句链,辅助智能阅读提问、智能客服、智能医疗问诊等领域对于自然语言理解与生成任务的实际需求,提出一种基于知识图谱关系预测的阶梯式问句链生成模型(stepped question chain generator based on knowledge graph link prediction,SQCG-KGLP)。SQCG-KGLP模型包括编码和解码两部分,在编码部分,对问句链的问句实体进行特征初始化,并通过特征融合方法获得问句链的融合头实体和待测尾实体的初始向量。将融合头实体和待测尾实体的初始向量送入SQCG-KGLP模型的Graph Attention图表示学习模块中,从而获得融合头实体和待测尾实体的表示向量。在解码部分,将融合头实体以及待测尾实体的表示向量输入到SQCG-KGLP模型的convKB模块中进行链接预测。在自建问句链数据集上进行不同跳数的实验,以MRR、Hits@1、Hits@3、Hits@10作为评测指标。结果表明,SQCG-KGLP算法均优于基线模型。当问句链生成迭代次数增加时,支持关系预测的知识图谱跳数随之增加,下一个节点问句语义信息表达的自然性会随之下降。此外,在现实应用场景中较难获得足量的高跳数问句训练数据集。目前,该研究仅能支持三跳以下的问句链生成。利用知识图谱的语义关联及知识图谱关系预测机制,生成的阶梯式问句链具有层层递进、由易到难、环环相扣的特点,使问句链的阶梯式逻辑质量更优。
