摘要：MoS_(2)is a promising electrocatalyst for hydrogen evolution reaction and a good candidate for cocatalyst to enhance the photoelectrochemical(PEC)performance of Si-based photoelectrode in aqueous *** main challenge lies in the optimization of the microstructure of MoS_(2),to improve its catalytic activity and to construct a mechanically and chemically stable cocatalyst/Si *** this paper,a highly-ordered mesoporous MoS_(2)was synthesized and decorated onto a TiO_(2)protected p-silicon *** additional TiO_(2)necking was introduced to strengthen the bonding between the MoS_(2)particles and the TiO_(2)*** meso-MoS_(2)/TiO_(2)/p-Si hybrid photocathode exhibited significantly enhanced PEC performance,where an onset potential of+0.06 V(versus RHE)and a current density of-1.8 mA/cm^(2)at 0 V(versus RHE)with a Faradaic efficiency close to 100%was achieved in 0.5 mol/L H_(2)SO_(4).Additionally,this meso-MoS_(2)/TiO_(2)/p-Si photocathode showed an excellent PEC ability and durability in alkaline *** paper provides a promising strategy to enhance and protect the photocathode through high-performance surface cocatalysts.

摘要：边坡失稳每年在全球造成重大的经济损失,为了更快速精确地判断边坡的稳定状态,保障边坡工程的安全。针对此问题,本文提出采用GA-SVM算法构建边坡稳定性预测模型。选取6个典型边坡参数:容重、黏聚力、内摩擦角、坡角、坡高、孔隙水压力比作为输入端要素,边坡状态作为输出端要素,结合收集到740个工程实际案例数据构建完成模型训练样本集。结果表明:GA-SVM模型受试者工作特征曲线下面积(AUC)值为0.942,准确率为0.976,召回率为0.976,精确率0.979,F1 score为0.976,反映出该模型在识别边坡失稳状态时的预测精度高、泛化能力强、预测结果可靠。再结合工程实例验证反映出边坡状态的评估结果与实际情况一致。说明可将GA-SVM模型应用于实际的边坡稳定性预测,可为边坡的设计施工提供依据,在实际的工程应用中具有良好的应用前景。

摘要：Legal English is characterized by the great number of legal terms,which cause difficulties for learners or translators to fully understand these legal concepts and to find equivalents in their native *** comprehend these special terms accurately,we need to resort to semantic taxonomy and frame semantics which provide cognitive frameworks to illustrate the semantic complexities of legal *** is high time as well as the first time to apply L?bner’s theory of semantic taxonomy and Fillmore’s frame semantics to interpret English legal *** to L?bner,meanings of some linguistic unit can be analyzed into three levels:expression meaning,utterance meaning and communicative meaning,corresponding with three types of equivalency:formal,semantic and *** dictionaries,as well as encyclopedias,have been trying to expose these three levels of meaning in the *** far as legal dictionaries are concerned,the job should be more complicated since legal terms embody unusual expression meaning though the expressions may take on common looks and the utterance meaning in addition to the functional meaning involve more intriguing legal procedures or *** abundance of legal language contributes to this complexity in interpreting legal ***’s theory and methodology of frame semantics pave the road to solving this *** paper provides examples of how to use the special methodology to do the interpretation for the purpose of promoting legal literacy.

摘要：抗冲击结构在军民安全领域有着广泛且重要的应用,但其优化设计因涉及极端载荷下的强非线性而面临结构响应求解耗时与灵敏度分析困难的挑战.针对该问题,提出了一种自训练分类判断优化设计方法.该方法采用基于支持向量机的自训练分类判断代理模型和遗传算法,求解抗冲击结构尺寸优化列式.与基于回归代理模型的传统方法不同,自训练分类判断代理模型可以减少样本集构建计算成本.同时,提出了面向分类判断代理模型的遗传算法约束处理策略及个体适应度计算方法.基于该方法开展的波纹夹芯结构抗爆性能和聚脲/陶瓷复合板抗侵彻性能优化设计,验证了其有效性和高效性.为抗冲击结构设计提供了一种新的高效优化方法,以期保障极端冲击载荷下装备结构服役性能和人员安全.

摘要：Designing airfoils according to given pressure (or velocity) distribution is one kind of free boundary problems. Free boundary condition can be coupled with the flow governing equations by variable-domain variational calculus, which makes it possible to calculate simultaneously the flow field and the free boundary. An accurate deduction of the variable-domain variational principles is taken herein to design airfoils in compressible and incompressible flows. Furthermore, two grid types (H and O) are used in the calculation with better results for the O-type grid. It is shown that convergence is accelerated and good results can be obtained even if the initial guessed airfoil shape is a triangle, demonstrating the strong adaptability of this method.

摘要：近几年,基于Transformer的预训练模型展现了强大的模态表征能力,促使了多模态的下游任务(如图像描述生成任务)正朝着完全端到端范式的趋势所转变,并且能够使得模型获得更好的性能以及更快的推理速度.然而,该技术所提取的网格型视觉特征中缺乏区域型的视觉信息,从而导致模型对对象内容的描述不精确.因此,预训练模型在图像描述生成任务上的适用性在很大程度上仍有待探索.针对这一问题,提出一种基于视觉区域聚合与双向协作学习的端到端图像描述生成方法(visual region aggregation and dual-level collaboration,VRADC).为了学习到区域型的视觉信息,设计了一种视觉区域聚合模块,将有相似语义的网格特征聚合在一起形成紧凑的视觉区域表征.接着,双向协作模块利用交叉注意力机制从两种视觉特征中学习到更加有代表性的语义信息,进而指导模型生成更加细粒度的图像描述文本.基于MSCOCO和Flickr30k两个数据集的实验结果表明,所提的VRADC方法能够大幅度地提升图像描述生成的质量,实现了最先进的性能.

摘要：针对双屏蔽电机端部结构件复杂,端部漏抗难以计算的问题。采用三维有限元数值解法对双屏蔽电机端部漏抗进行计算,利用端部漏抗计算结果采用二维场路耦合有限元法对双屏蔽电机性能进行计算,并将计算结果与实测值进行对比。在此基础上,分析了解析法在计算端部漏抗时产生误差的原因以及解析法的计算误差对电机性能计算的影响。计算结果表明:利用数值法计算端部漏抗得到的电机性能与实测值更接近,与数值法相比,由解析法计算误差引起的起动电流倍数和起动转矩倍数相对误差可分别达到11.75%和20.48%。计算结果为双屏蔽电机设计及计算提供了更准确的参考依据。

摘要：针对车内语音交互在复杂噪声环境下识别率低以及难以在有限计算资源设备上部署问题,本文设计了轻量化的语音增强模型和语音识别模型并进行联合训练.语音增强模型引入多尺度通道时频注意力模块来提取多尺度时频特征和各个维度上的关键信息.在语音识别模型中提出了多头逐元素线性注意力,显著降低了注意力模块所需的计算复杂度.实验表明,在自制数据集上这一联合训练模型表现出良好的噪声鲁棒性.

摘要：基于系统能耗最小原理的极限分析上限定理,利用MATLAB强大的数据处理功能,对边坡稳定系数Ncr采用序列二次规划法(SQP)进行优化计算,并与枚举法相结合,从而解决SQP算法求解优化解容易陷入局部最优解的问题。同时开发了一款简单实用的边坡稳定性分析软件,实现了计算结果与临界滑裂破坏面图形的可视化输出。与经典算例对比分析,计算结果具有较好的一致性,表明了程序的可靠性。分析了内摩擦角、边坡倾斜角、3个参数对边坡稳定系数Ncr的影响。计算结果表明:边坡稳定性系数Ncr对、较为敏感,而对的敏感次之。其次,当、比较接近或在小范围内波动时,对稳定性系数Ncr的影响特别敏感;当、值较小时,滑动面通过坡趾下方,而当值比较大时,滑动面通过坡趾。

摘要：建立了基于欧拉方程组的二维振荡机翼非定常气动设计反命题方法的数学模型。通过一系列变换,将物理时空中的求解域转换成映象坐标系中的规范区域;并导出了映象坐标下的欧拉方程,结合非定常反命题的边界条件,便可用有限差分方法求解。本方法引入映象坐标系解决反命题边界形状的不确定性,并能利用欧拉方程现有的各种差分格式于反命题求解。