AN ACCELERATION STRATEGY FOR RANDOMIZE-THEN-OPTIMIZE SAMPLING VIA DEEP NEURAL NETWORKS

Randomize-then-optimize (RTO) is widely used for sampling from posterior distribu-tions in Bayesian inverse problems.However,RTO can be computationally intensive for complexity problems due to repetitive evaluations of the expensive forward model and its gradient.In this work,we present a novel goal-oriented deep neural networks (DNN) sur-rogate approach to substantially reduce the computation burden of RTO.In particular,we propose to drawn the training points for the DNN-surrogate from a local approximated posterior distribution-yielding a flexible and efficient sampling algorithm that converges to the direct RTO approach.We present a Bayesian inverse problem governed by elliptic PDEs to demonstrate the computational accuracy and efficiency of our DNN-RTO ap-proach,which shows that DNN-RTO can significantly outperform the traditional RTO.     

p-Adic函数空间上Hardy型算子的多线性交换子估计

在p-adic域上研究分数次Hardy型算子与CMO(Qnp)函数生成的多线性交换子,建立了交换子在Lebesgue空间和Herz空间上的有界性.对Hardy算子的多线性交换子也得到了相应的结果.     

Modified Lagrangian vortex method with improved boundary conditions for water waves past a thin bottom‐standing barrier

Herein, the modified Lagrangian vortex method (LVM), a hybrid analytical‐numerical algorithm per se, is devised to simulate the process of vortex formation and shedding from the sharp edge of a zero‐thickness vertical plate under linear water‐wave attack. Application of the Helmholtz decomposition facilitates a convenient switch between the inviscid‐ and viscous‐flow models, thereby enabling easy incorporation of vorticity effects into the potential‐flow calculations for the viscous‐dominated region. In evaluating the potential‐flow component, making good use of the quickly convergent technique with singular basis functions, correctly capturing the singular behavior in velocity fields near the tip of the plate, leads to a considerable reduction of computational burdens and to 12‐decimal‐place accuracy. The viscous correction is carried out via the meshless LVM with improved boundary conditions. Comparisons with previously published results show good agreement. Simulations of vortex generation and evolution illuminate the ability of the present method, and provide a supplement to pertinent experimental works. The hybrid scheme proposed herein allows flexibility for the former LVM and convenience in the code development. Such a compromise fits particularly well for the high‐resolution modeling of sharp‐edged vortex shedding without heavy numerical developments. Copyright © 2014 John Wiley & Sons, Ltd.     

一种双星敏感器联合确定船体姿态方法

单星敏感器横滚测量精度偏低会影响航向测量精度,采用双星敏感器联合的方法可以有效提高船姿精度,其中关键是蒙气差修正和联合姿态确定方法。目前蒙气差修正一般采用基于恒星观测矢量的修正方法,联合姿态确定方式一般采用基于双视轴指向联合的方法。以上在蒙气差中需要引入初始船体姿态,而且修正迭代解算过程复杂,再者单星敏感器需要识别4颗以上恒星才能获得视轴指向,如果其中一个无法获得视轴指向,双星敏感器无法解算船体姿态。针对上述情况,提出了一种新的双星敏感器联合确定船体姿态的方法,包括基于恒星参考矢量的蒙气差修正方法和基于恒星观测矢量的联合姿态确定方法。基于该方法蒙气差修正不再需要初始船姿,而且在理论上只需双星敏感器各自至少有1颗可识别恒星,即可完成船体姿态角的解算。试验结果表明该方法降低了双星敏感器获取船体姿态角所需的可识别恒星数量限制,提高了双星敏感器获取船体姿态的能力。     

光伏支架柱平面内有侧移失稳计算长度系数研究

光伏支架是一种典型的不等高单层框架,有侧移失稳的光伏支架柱计算长度系数需考虑多方面因素的影响,无法直接套用现有规范公式．综合考虑了光伏支架短柱和长柱的刚度比γ、短柱和长柱的高度比β以及梁柱线刚度比K1的影响,进行了理论推导,得出了不等高单层框架柱有侧移失稳计算长度系数的计算公式,并制作表格以便工程人员查阅使用．然后以某典型光伏支架实际工程为算例,建立了有限元数值分析模型,并分别进行了线性屈曲分析和考虑几何非线性以及材料非线性的极限承载力分析．分析结果表明,提出的理论公式和计算表格能够精确地计算光伏支架有侧移失稳的极限承载力．     

垫层的E/d对水电站蜗壳结构的影响分析

利用ANSYS大型有限元软件,建立垫层蜗壳结构的三维有限元模型,研究了结构自重和内水压力作用下垫层弹性模量与厚度比值E/d对水电站蜗壳结构的影响,给出了蜗壳外围混凝土、蜗壳以及座环的特性随垫层E/d变化的规律。结果表明:在有垫层区,蜗壳外围混凝土承载比、与蜗壳接触处混凝土切向应力随垫层E/d增大而增大,蜗壳应力以及最大相对变形量随垫层E/d的增大而减小,当垫层E/d确定时蜗壳结构外围混凝土承载比、与蜗壳接触处混凝土切向应力、蜗壳和座环应力以及蜗壳变形量一定,不会随E或d的变化而变化;在无垫层区,蜗壳外围混凝土承载比、与蜗壳接触处混凝土切向应力、蜗壳应力变形与垫层E/d无关。     

纸页受力微损伤演化过程的声发射及多元统计特征

纸页受载后的损伤断裂行为涉及其使用性能,对其损伤断裂过程的研究不仅有助于揭示纸页破坏行为的物理本质,而且有助于对纸页质量及其生产、加工工艺作出更加科学合理的评定。本文基于声发射(Acoustic Emission,AE)实验研究纸页内部微损伤演化过程,给出了评估纸页宏观损伤断裂性能的定量描述方法。首先,根据AE信号与微损伤事件之间存在本质上的关联性,通过采集和观测纸页试样受载过程中的AE信号来研究纸页损伤断裂的基本特征;其次,针对所采集AE信号的数据量大、蕴含信息丰富导致数据分析和处理困难的特点,本文采用多元随机变量的分析方法。利用概率熵和Andrews曲线法等统计分析手段对AE数据展开研究分析,并最终利用一条损伤状态轨迹曲线实现了对纸页微损伤演化过程的完整宏观描述。结果表明,纸页微损伤演化过程呈现明显阶段性变化特征;Andrews曲线的聚类结果有效地区分了不同损伤阶段的相似性和差异性,同时,不同损伤阶段中纸样的微观形貌观测提供了实验性的证据。本文以一种典型的牛皮箱板纸为试件,验证了上述方法的有效性。