基于注意力机制与文本信息的用户关系抽取

随着社交媒体的发展,用户之间的关系网络对于社交媒体的分析有很大的帮助。因此,该文主要研究用户好友关系检测。以往的关于用户好友关系抽取的研究主要基于社交媒体上的结构化信息,比如其他好友关系,用户的不同属性等。但是,很多时候用户本身并没有大量的好友信息存在,同时也不一定有很多确定的属性。因此,我们希望基于用户发表的文本信息来对用户关系进行预测。不同于以往的潜在好友推荐算法,该文提出了一种基于注意力机制以及长短时记忆网络(long short-term memory,LSTM)的好友关系预测模型,将好友之间的评论分开处理,通过分析用户之间的评论来判断是否具备一定的好友关系。该模型将好友双方信息拼接后的结果作为输入,并将注意力机制应用于LSTM的输出。实验表明,用户之间的评论对于好友关系预测确实有较大的实际意义,该文提出的模型较之于多个基准系统的效果,取得了明显的提升。在不加入任何其它非文本特征的情况下,实验结果的准确率达到了77%。     

储层泥浆侵入深度预测方法研究

泥浆侵入半径的确定尚缺乏严格的验证标准.以油水两相渗流理论和离子扩散方程为基础,结合储集层特点,研究了不同储层参数下泥浆滤液对地层的侵入特性.数值模拟侵入时间选取10 d和20 d.数值模拟结果表明,侵入半径在渗透率不变的情况下随孔隙度的增大而减小,在孔隙度不变的情况下随渗透率的增大而增大;当渗透率和孔隙度都发生变化时,泥浆侵入半径一般随孔隙度的增加呈幂函数增加.依据这种关系对测井资料约束处理,得到的泥浆侵入半径较客观地反映了地层的真实情况.     

An original approximate method for estimating the invariant probability distribution of a large class of multi-dimensional nonlinear stochastic oscillators

This paper proposes an original method for obtaining analytical approximations of the invariant probability density function of multi-dimensional Hamiltonian dissipative dynamic systems under Gaussian white noise excitations, with linear non-conservative parts and nonlinear conservative parts. The method is based on an exact result and a heuristic argument. Its pertinence is attested by numerical tests.     

掺气减蚀保护作用的新概念

利用针式掺气流速仪测量原型和模型掺气浓度场、流速场，气泡尺寸及其概率分布的研究成果表明，原型水流韦伯数高，形成微小气泡的能力比模型强，气泡上浮慢，接近底部的小尺寸气泡概率及掺气浓度比模型大。初步研究认为：0.2mm或0.5mm以下的微小气泡在掺气减蚀中起着主要作用，可能只要很小掺气浓度即可达到掺气减蚀的效果。因此以小尺寸气泡的掺气浓度，作为判断掺气减蚀保护作用的指标将更为准确。     

火工品压药机装配误差的研究

针对军用雷管生产线中压药机结构,分别用极值法和概率法对压药机冲子和雷管的配合精度,进行了装配尺寸链的描述和计算,并对所得的结果进行了比较。     

δ－Al_2O_3纤维／Al－12Si复合材料室温拉伸强度分析

对δ－Ａｌ２Ｏ３纤维／Ａｌ－１２Ｓｉ复合材料室温拉伸强度的分析表明，在实验条件下该复合材料存在δ－Ａｌ２Ｏ３纤维的最小体积分数Ｖｍｉｎ和临界体积分数Ｖｃｒｉｔ，并求出其基体强度σ＇ｍ和室温强度σｃ－δ－Ａｌ２Ｏ３纤维体积分数Ｖｆ直线方程及纤维的临界长度ｌｃ和复合材料的剪切应力τｉ．确定复合材料的ＲＯＭ预测曲线，应首先判断σ＇ｍ是否等于未增强合金的强度σｕｍ才能得出正确的结论     

测井新技术在四川油气勘探开发中的应用

综述随着四川盆地油气勘探深入所面临的油气领域复杂和困难的技术问题，介绍应用测井新技术为四川油气开发解决了大量的技术难题，特别是在复杂缝，洞型储层的评价，井旁地质构造形态精细分析和地质，测井，地震多学科结合储层横向预测等3个方面所取得的良好应用效果。     

等偏移距自动初至静校正

茅金根 ,梁秀文 ,杨午阳 ,冯有奎 .等偏移距自动初至静校正 .石油地球物理勘探 ,2 0 0 2 ,37(5 ) :4 6 9～ 4 72　　本文提出一种自动计算静校正量的新方法。它首先在等偏移距道集内对相邻道的初至波采用高精度相关构建一个关于炮、检时差的超大矩阵方程 ,然后采用优化统计迭代算法求解炮点和检波点的静校正量 ,并用概率分布函数确定精确的炮、检静校正量 ,最后用误差校正函数消除炮、检静校正误差。实际资料处理结果表明 ,本方法自动、省时、精度高。     

Effects of solidification structure on fatigue crack initiation and fatigue strength in Al−Si−Mg cast alloys

Rotating-bending uniaxial fatigue tests and micro-fatigue crack initiation tests were carried out using a permanent mold cast (PMC) and semi-solid die cast (SDC) with Al−7%Si−0.35%Mg composition in order to examine the relationship between solidification structures and fatigue behaviors. The crack length was measured using a replication method. Fatigue strength was improved in SDC, which was almost consistent with the predicted fatigue strength using the size of Si particle cluster. Resistance to fatigue crack initiation and fatigue strength were improved in SDC owing to the finer Si cluster and to higher ultimate tensile strength. Fatigue crack in PMC was preferentially initiated at pores. For SDC, the fatigue crack was initiated at the Si particle/matrix interface, and then sucessively grew along eutectic cell boundaries.     

A probabilistic model for cleavage fracture with a length scale--parameter estimation and predictions of stationary crack experiments

This study presents a large experimental investigation in the transition temperature region on a modified A508 steel. Tests were carried out on single-edge-notch-bend specimens with three different crack depth over specimen width ratios to capture the strong constraint effect on fracture toughness. Three test temperatures were considered, covering a range of 85 °C. All specimens failed by cleavage fracture prior to ductile tearing. A recently proposed probabilistic model for the cumulative failure by cleavage was applied to the comprehensive sets of experimental data. This modified weakest link model incorporates a length scale, which together with a threshold stress reduce the scatter in predicted toughness distributions as well as introduces a fracture toughness threshold value. Model parameters were estimated by a robust procedure, which is crucial in applications of probabilistic models to real structures. The conformity between predicted and experimental toughness distributions, respectively, were notable at all the test temperatures.