The interaction of bubbles is the key to understand gas–liquid bubbling flow. Two-dimensional axis-symmetry computational fluid dynamics simulations on the interactive bubbles were performed with VOF method,which was...The interaction of bubbles is the key to understand gas–liquid bubbling flow. Two-dimensional axis-symmetry computational fluid dynamics simulations on the interactive bubbles were performed with VOF method,which was validated by experimental work. It is testified that several different bubble interactive behaviors could be acquired under different conditions. Firstly, for large bubbles(d: 4, 6, 8, 10 mm), the trailing bubble rising velocity and aspect ratio have negative correlations with liquid viscosity and surface tension. The influences of viscosity and surface tension on leading bubble are negligible. Secondly, for smaller bubbles(d: 1, 2 mm), the results are complicated. The two bubbles tend to move together due to the attractive force by the wake and the potential repulsive force. Especially for high viscous or high surface tension liquid, the bubble pairs undergo several times acceleration and deceleration. In addition, bubble deformation plays an important role during bubble interaction which cannot be neglected.展开更多
The dynamics of the scattering processes of diatomic molecules from metal surfaces has been studied with different theoretical approaches. Modified LEPS (London-Eyring-Polanyi-Sato) potential surfaces for several diat...The dynamics of the scattering processes of diatomic molecules from metal surfaces has been studied with different theoretical approaches. Modified LEPS (London-Eyring-Polanyi-Sato) potential surfaces for several diatomie molecule-surface systems have been constructed and examined for the dynamic study. The surfaces are treated as rigid but corrugated. The potential parameters are adjusted to produce reliable potential hypersurfaces. Molecular dissociation, diffraction, adsorption and consequent desorption in the scattering processes have been observed through quasiclassieal trajectory calculations. The significance of the effective corrugation of the potential surfaces has been evaluated in calculating the dissociation and adsorption probabilities. Vibration-rotation-translation energy transfer in the inelastic scattering is investigated to understand the mechanism of selective adsorptions mediated through vibrational or rotational degrees of freedom. We have carried out quantum mechanical calculations to obtain the rotational and vibrational transition probabilities. Relative importance of rotational and vibrational transitions for each adsorbed state with respect to incidence energy has been carefully examined to determine the dominant factor which causes the adsorbed state. The results show that vibration mediation is an essential factor to the selective adsorption especially in the ease of higher incidence energies.展开更多
现有的方面级情感分析方法对句法依存树蕴含信息使用不足,忽略多方面词之间的关联,并且缺少对外部知识的使用。针对这些问题,提出一种知识增强的方面词交互图神经网络(KEAIG)模型。首先利用融合领域知识的BERT-PT (Bidirectional Encode...现有的方面级情感分析方法对句法依存树蕴含信息使用不足,忽略多方面词之间的关联,并且缺少对外部知识的使用。针对这些问题,提出一种知识增强的方面词交互图神经网络(KEAIG)模型。首先利用融合领域知识的BERT-PT (Bidirectional Encoder Representation from Transformers with Post-Train)编码文本,并利用知识图谱增加句法树的情感信息。模型分两部分对句法依存树蕴含的信息进行提取:第一部分利用句法依存树中的关联关系和每个单词的词性标签提取句子特征,第二部分对融入知识图谱的句法依存树进行特征提取。之后使用融合门控单元将多方面词关联特征融合进提取到的特征中。最后将两部分句子表示拼接起来作为最终分类依据。在4个数据集上的实验结果表明,所提模型相较于基准模型关系图注意力网络(RGAT),在准确率上分别提升了2.17%、5.54%、2.60%和2.83%,在F1值(Macro-F1)上分别提升了2.69%、6.87%、8.77%和14.70%,充分表明了利用句法树、引入外部知识和提取多方面词关联的有效性。展开更多
目前大多数方面级情感分类研究都忽略了方面词的建模,以及方面词与上下文之间的交互信息,并且难以体现语法上与方面词有直接联系上下文单词的重要程度。针对上述问题,提出基于方面词交互(aspect word interaction,AWI)和图卷积网络(grap...目前大多数方面级情感分类研究都忽略了方面词的建模,以及方面词与上下文之间的交互信息,并且难以体现语法上与方面词有直接联系上下文单词的重要程度。针对上述问题,提出基于方面词交互(aspect word interaction,AWI)和图卷积网络(graph convolutional network,GCN)的方面级情感分类模型(AWI-GCN)。使用双向长短期记忆网络(bi-directional long short-term memory,Bi-LSTM)分别提取方面词和上下文的特征;采用GCN根据句法依存树进一步提取与方面词有直接语法联系的上下文情感特征;利用注意力机制学习方面词与上下文的交互信息,同时提取上下文中为方面词情感分类做出重要贡献的情感特征。针对3个公开数据集上的仿真实验结果表明,AWI-GCN模型相比当前代表模型取得了更好的情感分类效果。展开更多
基金国家自然科学基金 Grant No.61305079湖北省教育厅科学技术研究项目 Grant No.B20111607+5 种基金福建省自然科学基金 Grant No.2012J01250福建省教育厅科技项目 Grant Nos.JA12077JA12080武汉大学软件工程国家重点实验室开放基金 Grant No.SKLSE2012-09-28福建省高校杰出青年科研人才培育计划项目 Grant No.JA12471福建师范大学青年骨干教师培育计划 Grant No.fjsdjk2012083~~
基金Supported by the National Natural Science Foundation of China(91334105).
文摘The interaction of bubbles is the key to understand gas–liquid bubbling flow. Two-dimensional axis-symmetry computational fluid dynamics simulations on the interactive bubbles were performed with VOF method,which was validated by experimental work. It is testified that several different bubble interactive behaviors could be acquired under different conditions. Firstly, for large bubbles(d: 4, 6, 8, 10 mm), the trailing bubble rising velocity and aspect ratio have negative correlations with liquid viscosity and surface tension. The influences of viscosity and surface tension on leading bubble are negligible. Secondly, for smaller bubbles(d: 1, 2 mm), the results are complicated. The two bubbles tend to move together due to the attractive force by the wake and the potential repulsive force. Especially for high viscous or high surface tension liquid, the bubble pairs undergo several times acceleration and deceleration. In addition, bubble deformation plays an important role during bubble interaction which cannot be neglected.
基金The projcct supportcd by National Natural Science Foundation of China
文摘The dynamics of the scattering processes of diatomic molecules from metal surfaces has been studied with different theoretical approaches. Modified LEPS (London-Eyring-Polanyi-Sato) potential surfaces for several diatomie molecule-surface systems have been constructed and examined for the dynamic study. The surfaces are treated as rigid but corrugated. The potential parameters are adjusted to produce reliable potential hypersurfaces. Molecular dissociation, diffraction, adsorption and consequent desorption in the scattering processes have been observed through quasiclassieal trajectory calculations. The significance of the effective corrugation of the potential surfaces has been evaluated in calculating the dissociation and adsorption probabilities. Vibration-rotation-translation energy transfer in the inelastic scattering is investigated to understand the mechanism of selective adsorptions mediated through vibrational or rotational degrees of freedom. We have carried out quantum mechanical calculations to obtain the rotational and vibrational transition probabilities. Relative importance of rotational and vibrational transitions for each adsorbed state with respect to incidence energy has been carefully examined to determine the dominant factor which causes the adsorbed state. The results show that vibration mediation is an essential factor to the selective adsorption especially in the ease of higher incidence energies.
文摘现有的方面级情感分析方法对句法依存树蕴含信息使用不足,忽略多方面词之间的关联,并且缺少对外部知识的使用。针对这些问题,提出一种知识增强的方面词交互图神经网络(KEAIG)模型。首先利用融合领域知识的BERT-PT (Bidirectional Encoder Representation from Transformers with Post-Train)编码文本,并利用知识图谱增加句法树的情感信息。模型分两部分对句法依存树蕴含的信息进行提取:第一部分利用句法依存树中的关联关系和每个单词的词性标签提取句子特征,第二部分对融入知识图谱的句法依存树进行特征提取。之后使用融合门控单元将多方面词关联特征融合进提取到的特征中。最后将两部分句子表示拼接起来作为最终分类依据。在4个数据集上的实验结果表明,所提模型相较于基准模型关系图注意力网络(RGAT),在准确率上分别提升了2.17%、5.54%、2.60%和2.83%,在F1值(Macro-F1)上分别提升了2.69%、6.87%、8.77%和14.70%,充分表明了利用句法树、引入外部知识和提取多方面词关联的有效性。
文摘目前大多数方面级情感分类研究都忽略了方面词的建模,以及方面词与上下文之间的交互信息,并且难以体现语法上与方面词有直接联系上下文单词的重要程度。针对上述问题,提出基于方面词交互(aspect word interaction,AWI)和图卷积网络(graph convolutional network,GCN)的方面级情感分类模型(AWI-GCN)。使用双向长短期记忆网络(bi-directional long short-term memory,Bi-LSTM)分别提取方面词和上下文的特征;采用GCN根据句法依存树进一步提取与方面词有直接语法联系的上下文情感特征;利用注意力机制学习方面词与上下文的交互信息,同时提取上下文中为方面词情感分类做出重要贡献的情感特征。针对3个公开数据集上的仿真实验结果表明,AWI-GCN模型相比当前代表模型取得了更好的情感分类效果。
