Soils with spatial variability are the product of natural history.The mechanical properties tested by soil samples from boreholes in the same soil layer may be different.Underground structure service in surrounding so...Soils with spatial variability are the product of natural history.The mechanical properties tested by soil samples from boreholes in the same soil layer may be different.Underground structure service in surrounding soils,their seismic response is controlled by the deformation of the surrounding soils.The variability of soil mechanical parameters was not considered in the current research on the seismic response of underground structures.Therefore,a random field model was established to describe the spatial variability of surrounding soils based on the random field theory.Then the seismic response of underground structures in the random field was simulated based on the time-domain explicit global FEM analysis,and the soil mechanical parameters and earthquake intensity influencing the seismic response of surrounding soils and underground structures were studied.Numerical results presented that,the randomness of soil parameters does not change the plastic deformation mode of surrounding soils significantly.The variation coefficients of inter-story deformation of structures and lateral deformation of columns are much smaller than that of mechanical parameters,and the randomness of soil parameters has no obvious effect on the structural deformation response.展开更多
A fast method for phrase structure grammar analysis is proposed based on conditional ran- dom fields （CRF）. The method trains several CRF classifiers for recognizing the phrase nodes at dif- ferent levels, and uses ...A fast method for phrase structure grammar analysis is proposed based on conditional ran- dom fields （CRF）. The method trains several CRF classifiers for recognizing the phrase nodes at dif- ferent levels, and uses the bottom-up to connect the recognized phrase nodes to construct the syn- tactic tree. On the basis of Beijing forest studio Chinese tagged corpus, two experiments are de- signed to select the training parameters and verify the validity of the method. The result shows that the method costs 78. 98 ms and 4. 63 ms to train and test a Chinese sentence of 17. 9 words. The method is a new way to parse the phrase structure grammar for Chinese, and has good generalization ability and fast speed.展开更多
Tunnel horizontal convergence monitoring is essential to ensure the operation safety.However,only a few representative tunnel sec-tions are chosen for monitoring due to the cost limitation.It is difficult to capture t...Tunnel horizontal convergence monitoring is essential to ensure the operation safety.However,only a few representative tunnel sec-tions are chosen for monitoring due to the cost limitation.It is difficult to capture the horizontal convergence of each tunnel ring with limited measurements.Confronted with this difficulty,the paper proposes a horizontal convergence reconstruction method based on the measurements of deployed sensors.The tunnel horizontal convergence along the longitudinal direction is seen as a one-dimensional sta-tionary and ergodic random field.The reconstruction problem is then transformed into the generation of conditional random fields.Monte Carlo simulation is adopted to generate possible realizations and the mean of realizations is considered as the maximum likeli-hood reconstruction.Error analysis proves the effectiveness of the proposed reconstruction method.The proposed method is proved to be applicable in reconstructing the time-variant horizontal convergence and is verified by the monitoring results of the shield tunnel of Shanghai Metro Line 2.The effect of sensor numbers is parametrically studied,and an optimal sensor placement scheme is decided.Additional sensors placed at the deformation drastically changed location can significantly improve the performance of the proposed method.展开更多
随着大量科研论文全文本的出现,如何从中挖掘相应的知识不仅有利于学术文献的深度知识组织而且有益于学术文献的精准检索。而识别学术文本的结构是进行上述探究的基础,因为结构的识别有助于从更深层次或者偏重语义的角度理解学术文本,...随着大量科研论文全文本的出现,如何从中挖掘相应的知识不仅有利于学术文献的深度知识组织而且有益于学术文献的精准检索。而识别学术文本的结构是进行上述探究的基础,因为结构的识别有助于从更深层次或者偏重语义的角度理解学术文本,从而促进学术文本挖掘研究的发展。本文以学术文本的不同结构功能为研究对象,以Journal of the Association for Information Science and Technology(JASIST)上发表的1579篇论文为数据集,进行双向长短时记忆神经网络、支持向量机和条件随机场三种模型上的预实验,并对比实验结果的性能,最终确定利用条件随机场模型做进一步探究。利用条件随机场模型,本文将学术文本结构功能识别问题转化为对句子单元的序列标注问题,寻找最优识别模型并探究不同特征对结构功能识别的影响,最终获得开放测试的调和平均值为92.88%的结构整体识别效果。实验结果表明,章节标题中词汇信息和章节内容的特征词汇信息对学术文本的功能结构识别起到巨大作用,可以达到令人满意的效果,而结构的长度特征则干扰条件随机场方法的性能。在最后,本文对学术文本结构功能识别出错原因进行总结,指出进一步探讨的问题和方向。展开更多
定义在单一空间分辨率上的树结构马尔可夫场(Tree-Structured Markov Random Field,TS-MRF)模型能够表达图像的分层结构信息,但难以描述图像的非平稳性.针对该问题,提出小波域的TS-MRF图像建模方法—WTS-MRF模型.按照图像分类层次树的...定义在单一空间分辨率上的树结构马尔可夫场(Tree-Structured Markov Random Field,TS-MRF)模型能够表达图像的分层结构信息,但难以描述图像的非平稳性.针对该问题,提出小波域的TS-MRF图像建模方法—WTS-MRF模型.按照图像分类层次树的结构形式,该模型将一系列的MRF嵌套定义在多分辨率的小波域中:每一个树节点对应于定义在不同分辨率上的一个MRF集合,并通过条件概率的形式将相邻分辨率上的MRF间的作用关系考虑进来;同时相同分辨率的父子节点对应的MRF通过区域约束嵌套定义.基于WTS-MRF模型,给出了一个监督图像分割的递归算法,通过给定的分类层次树表示先验信息,并通过训练数据给出叶子节点在各分辨率上的统计参数.它在尺度内和尺度间两个层次上进行递归:首先,在最低分辨率上执行尺度内递归,即采用ICM算法从树的根节点到叶子节点依次对MRF进行递归估计;然后执行尺度间递归,即在相邻的更高分辨率尺度上,通过直接投影的方式依次获取每一MRF的初始估计,并采用ICM算法递归优化;最后,原始分辨率的MRF估计完成,获取最终分割结果.两组实验从视觉效果和定量指标(整体分类正确率和Kappa系数)两个方面验证了算法的有效性.展开更多
基金supported by the Beijing Natural Science Foundation(8212007)the Pyramid Talent Training Project of Beijing University of Civil Engineering and Architecture(JDYC20200311)。
文摘Soils with spatial variability are the product of natural history.The mechanical properties tested by soil samples from boreholes in the same soil layer may be different.Underground structure service in surrounding soils,their seismic response is controlled by the deformation of the surrounding soils.The variability of soil mechanical parameters was not considered in the current research on the seismic response of underground structures.Therefore,a random field model was established to describe the spatial variability of surrounding soils based on the random field theory.Then the seismic response of underground structures in the random field was simulated based on the time-domain explicit global FEM analysis,and the soil mechanical parameters and earthquake intensity influencing the seismic response of surrounding soils and underground structures were studied.Numerical results presented that,the randomness of soil parameters does not change the plastic deformation mode of surrounding soils significantly.The variation coefficients of inter-story deformation of structures and lateral deformation of columns are much smaller than that of mechanical parameters,and the randomness of soil parameters has no obvious effect on the structural deformation response.
基金Supported by the Science and Technology Innovation Plan of Beijing Institute of Technology(2013)
文摘A fast method for phrase structure grammar analysis is proposed based on conditional ran- dom fields （CRF）. The method trains several CRF classifiers for recognizing the phrase nodes at dif- ferent levels, and uses the bottom-up to connect the recognized phrase nodes to construct the syn- tactic tree. On the basis of Beijing forest studio Chinese tagged corpus, two experiments are de- signed to select the training parameters and verify the validity of the method. The result shows that the method costs 78. 98 ms and 4. 63 ms to train and test a Chinese sentence of 17. 9 words. The method is a new way to parse the phrase structure grammar for Chinese, and has good generalization ability and fast speed.
基金supported by the National Key R&D Program of China(Grant No.2022YFC3801000)Natural Science Foundation Committee Program of China(Grant No.51978530,52022070)and China Scholarship Council.
文摘Tunnel horizontal convergence monitoring is essential to ensure the operation safety.However,only a few representative tunnel sec-tions are chosen for monitoring due to the cost limitation.It is difficult to capture the horizontal convergence of each tunnel ring with limited measurements.Confronted with this difficulty,the paper proposes a horizontal convergence reconstruction method based on the measurements of deployed sensors.The tunnel horizontal convergence along the longitudinal direction is seen as a one-dimensional sta-tionary and ergodic random field.The reconstruction problem is then transformed into the generation of conditional random fields.Monte Carlo simulation is adopted to generate possible realizations and the mean of realizations is considered as the maximum likeli-hood reconstruction.Error analysis proves the effectiveness of the proposed reconstruction method.The proposed method is proved to be applicable in reconstructing the time-variant horizontal convergence and is verified by the monitoring results of the shield tunnel of Shanghai Metro Line 2.The effect of sensor numbers is parametrically studied,and an optimal sensor placement scheme is decided.Additional sensors placed at the deformation drastically changed location can significantly improve the performance of the proposed method.
文摘随着大量科研论文全文本的出现,如何从中挖掘相应的知识不仅有利于学术文献的深度知识组织而且有益于学术文献的精准检索。而识别学术文本的结构是进行上述探究的基础,因为结构的识别有助于从更深层次或者偏重语义的角度理解学术文本,从而促进学术文本挖掘研究的发展。本文以学术文本的不同结构功能为研究对象,以Journal of the Association for Information Science and Technology(JASIST)上发表的1579篇论文为数据集,进行双向长短时记忆神经网络、支持向量机和条件随机场三种模型上的预实验,并对比实验结果的性能,最终确定利用条件随机场模型做进一步探究。利用条件随机场模型,本文将学术文本结构功能识别问题转化为对句子单元的序列标注问题,寻找最优识别模型并探究不同特征对结构功能识别的影响,最终获得开放测试的调和平均值为92.88%的结构整体识别效果。实验结果表明,章节标题中词汇信息和章节内容的特征词汇信息对学术文本的功能结构识别起到巨大作用,可以达到令人满意的效果,而结构的长度特征则干扰条件随机场方法的性能。在最后,本文对学术文本结构功能识别出错原因进行总结,指出进一步探讨的问题和方向。
文摘定义在单一空间分辨率上的树结构马尔可夫场(Tree-Structured Markov Random Field,TS-MRF)模型能够表达图像的分层结构信息,但难以描述图像的非平稳性.针对该问题,提出小波域的TS-MRF图像建模方法—WTS-MRF模型.按照图像分类层次树的结构形式,该模型将一系列的MRF嵌套定义在多分辨率的小波域中:每一个树节点对应于定义在不同分辨率上的一个MRF集合,并通过条件概率的形式将相邻分辨率上的MRF间的作用关系考虑进来;同时相同分辨率的父子节点对应的MRF通过区域约束嵌套定义.基于WTS-MRF模型,给出了一个监督图像分割的递归算法,通过给定的分类层次树表示先验信息,并通过训练数据给出叶子节点在各分辨率上的统计参数.它在尺度内和尺度间两个层次上进行递归:首先,在最低分辨率上执行尺度内递归,即采用ICM算法从树的根节点到叶子节点依次对MRF进行递归估计;然后执行尺度间递归,即在相邻的更高分辨率尺度上,通过直接投影的方式依次获取每一MRF的初始估计,并采用ICM算法递归优化;最后,原始分辨率的MRF估计完成,获取最终分割结果.两组实验从视觉效果和定量指标(整体分类正确率和Kappa系数)两个方面验证了算法的有效性.
