卤键在化学传感和分子识别中的应用

卤键是一种新的分子间非共价作用力,它存在于卤素原子(路易斯酸)和具有孤电子对的原子或π-电子体系(路易斯碱)之间,在超分子化学、材料科学、生物识别和药物设计等领域已经显示出独特的优势。本文主要从卤键的特征和在化学传感和分子识别中的应用以及发展前景等几方面进行了介绍,期望引起人们对卤键的更多关注。     

基因识别及其算法研究

针对基因识别问题,基于DNA序列的3周期这一性质,首先给出了DNA序列功率和信噪比的快速算法并讨论了不同物种基因类型的阈值确定方法;在此基础上,建立了基于背景噪声抑制和频谱平滑的SNR频谱预处理模型,经过预处理后的频谱不仅大幅度抑制了背景噪声,同时保留了SNR频谱的模式特征.在编码序列识别上,对经典的EPND预测算法进行了改进,使用改进的EPND算法对经过预处理后频谱进行基因识别,实验结果显示这种基因识别模型具有优异的基因识别性能,比传统直接使用基于滑动窗口DFT的EPND识别算法在敏感度、特异性等评价指标上提高了2%-12%左右.     

Influences of aerodynamic loads on hunting stability of high-speed railway vehicles and parameter studies

The influences of steady aerodynamic loads on hunting stability of high-speed railway vehicles were investigated in this study.A mechanism is suggested to explain the change of hunting behavior due to actions of aerodynamic loads：the aerodynamic loads can change the position of vehicle system（consequently the contact relations）,the wheel/rail normal contact forces,the gravitational restoring forces/moments and the creep forces/moments.A mathematical model for hunting stability incorporating such influences was developed.A computer program capable of incorporating the effects of aerodynamic loads based on the model was written,and the critical speeds were calculated using this program.The dependences of linear and nonlinear critical speeds on suspension parameters considering aerodynamic loads were analyzed by using the orthogonal test method,the results were also compared with the situations without aerodynamic loads.It is shown that the most dominant factors a ff ecting linear and nonlinear critical speeds are different whether the aerodynamic loads considered or not.The damping of yaw damper is the most dominant influencing factor for linear critical speeds,while the damping of lateral damper is most dominant for nonlinear ones.When the influences of aerodynamic loads are considered,the linear critical speeds decrease with the rise of cross wind velocity,whereas it is not the case for the nonlinear critical speeds.The variation trends of critical speeds with suspension parameters can be significantly changed by aerodynamic loads.Combined actions of aerodynamic loads and suspension parameters also a ff ect the critical speeds.The effects of such joint action are more obvious for nonlinear critical speeds.     

面向鲁棒自动语音识别的一致性自监督学习方法

提出了一种基于一致性自监督学习的鲁棒自动语音识别方法。该方法通过使用语音信号仿真技术,模拟一条语音在不同声学场景下的副本;在通过自监督学习方式学习语音表征的同时,极大化一条语音在不同声学环境下对应语音表征的相似性,从而获取到与环境干扰无关的语音表征方式,提高下游语音识别模型的性能。在远讲数据集CHiME-4和会议数据集AMI上的实验表明,所提的一致性自监督学习算法能够取得相比已有的wav2vec2.0自监督学习基线算法30%以上的识别词错误率下降。这表明,所提方法是一种获取噪声无关语音表征、提升鲁棒语音识别性能的有效方法。     

Fatigue damage behavior of a surface-mount electronic package under different cyclic applied loads

This paper studies and compares the effects of pull-pull and 3-point bending cyclic loadings on the mechanical fa- tigue damage behaviors of a solder joint in a surface-mount electronic package. The comparisons are based on experimental investigations using scanning electron microscopy （SEM） in-situ technology and nonlinear finite element modeling, respec- tively. The compared results indicate that there are different threshold levels of plastic strain for the initial damage of solder joints under two cyclic applied loads; meanwhile, fatigue crack initiation occurs at different locations, and the accumulation of equivalent plastic strain determines the trend and direction of fatigue crack propagation. In addition, simulation results of the fatigue damage process of solder joints considering a constitutive model of damage initiation criteria for ductile materials and damage evolution based on accumulating inelastic hysteresis energy are identical to the experimental results. The actual fatigue life of the solder joint is almost the same and demonstrates that the FE modeling used in this study can provide an accurate prediction of solder joint fatigue failure.     

SHPB冲击作用下层状千枚岩多尺度破坏机理研究

通过分离式霍普金森杆对层状千枚岩施加动态载荷，得到不同层理倾角下层状千枚岩的动态抗压强度与宏观破坏模式。采用三维激光仪获得断裂面细观形貌，引入分形几何定量计算断口面粗糙度；结合SEM观察到的微观尺度下不同层理倾角断口破坏机理，分析了不同层理倾角下层状岩石的动态破坏机制。研究结果表明:动态压缩下层理弱面对岩石的抗压强度影响较大；不同层理倾角千枚岩的断口形貌分形维数随层理倾角增大呈U型变化；从强度与裂纹扩展两方面考虑层理弱面对层状岩石破坏特征的影响，对于层理倾角为0°的试样，强度由岩石基质控制，但层理弱面仍对岩石破坏的裂纹分布与走向产生较大影响；对于层理倾角为22.5°的试样，强度与裂纹走向受岩石基质与层理弱面共同控制；对于层理倾角为45°～67.5°的试样，强度与裂纹走向受层理弱面控制；而对于层理倾角为90°的试样，动态抗压强度受岩石基质的影响较大，在层理弱面较早形成纵向宏观裂纹，导致该层理弱面角度下裂纹受层理弱面的影响较大。     

乙腈溶剂中芳酰基硫脲对F~-识别作用的理论研究

采用混合密度泛函B3LYP方法,在6-31G**水平上和URHF溶剂半径中优化乙腈溶剂中四种芳酰基硫脲及其氟离子复合物的几何构型,从几何结构参数、电荷布居、前线轨道、结合能以及热力学参数等角度探讨复合物形成过程中主客体间的相互作用,结果表明,芳酰基硫脲在乙腈溶液中识别氟离子是一个自发过程,识别作用主要是硫脲氮氢与氟离子的氢键作用。受体分子识别氟离子能力与取代基密切相关,以N-对硝基苯基-N’-对硝基苯甲酰基硫脲氟离子复合物最稳定。在基态结构的基础上用相同的理论水平计算了乙腈溶剂中四种受体分子及氟离子复合物的电子吸收光谱。结果表明,有取代基的芳酰基硫脲识别氟离子后的光谱行为变化大,有利于使用比色法检测阴离子。光谱理论计算值很好地反映了实测值,这为实验室开展工作提供了依据。     

基于深度学习的人体行为识别算法

目前,针对深度学习的人体行为识别研究,往往采用视频中的全局信息对人体行为进行分析.然而,局部信息缺失造成的特征提取不完备,同样会导致识别精度急剧下降.由此,提出了基于多流深度学习的人体行为识别方法,将人体局部信息与全局信息相结合,通过局部不同特征的精确识别,使人体行为识别更加准确.实验表明,与现有深度学习方法相比,提出的方法在数据集UCF101和HMDB51上识别精度分别平均提高了约4.0%和6.2%.     

非正态变异下的非线性轮廓异常点识别方法研究

在质量控制领域,非线性轮廓异常点识别问题是重点研究问题之一。本文综合运用了小波分析、数据深度、聚类分析等数据分析处理技术,提出了一种新的非正态变异的异常点识别方法。文章通过仿真分析技术,将新方法χ²与控制图方法进行性能对比,结果证实新方法能够以更高的准确率和稳定性识别异常点,表现出更好的异常点识别性能。最后将新方法应用于木板垂直密度轮廓实例对新方法进行验证,分析结果表明本方法能够有效识别出异常轮廓数据。