基于数据质量评价的目标估计

针对目标估计过程需要大量人工参与、自动化程度低的问题，提出了基于数据质量评价的目标估计方法。利用目标数据质量评价方法，对不同传感器得到的目标数据质量进行科学、有效的测度和评价，并根据质量得分动态调整各数据源在目标估计过程中所占的权重，从而减少人工干预，提高目标估计效能。仿真试验结果证明了该方法的有效性。     

基于单光子计数激光雷达的Bayesian检测研究

在单光子计数激光雷达检测领域,目前的检测方法在低信噪比情况下虚警概率会增加,同时也无法适应噪声变化的问题。针对这些问题,提出了一种基于Bayesian的检测方法,该方法首先通过雷达方程估计回波信号光子数的范围,将其作为先验信息,而后结合二项分布建立了累计概率模型,基于Bayesian判决准则计算得到检测阈值,此阈值能够在检测概率与虚警概率中间择其平衡。这种方法不仅克服了低信噪比检测困难的情况,还减少了先验信息的获取难度。实验结果表明,对比固定阈值其虚警概率降低了10倍。对比“恒虚警”其检测概率提高了约20。验证了方法具有良好的检测效果,具备一定的可操作性。     

基于误差估值累加开环校正的诱导式欺骗检测方法

诱导式卫星欺骗干扰可诱导航空器逐渐偏离预定航迹，难以被发现，因此及时有效地检测干扰是飞行安全的保障。在现有紧组合导航体制基础上，设计了一种基于误差估值累加开环校正的紧组合导航结构，并证明了其性能与传统闭环校正紧组合导航性能等效。在此结构中，将紧组合导航系统与自适应序贯概率比检测方法结合，提出了一种基于误差估值累加开环校正的诱导式欺骗检测方法，融合紧组合导航信息与其他不受欺骗影响的导航信息，构建欺骗检测统计量进行诱导式欺骗检测。仿真结果表明，开环校正结构可避免随时间累加的惯性导航系统误差所导致的组合导航滤波器发散问题，同时欺骗检测方法可进一步提高算法对“最坏”情形下微小诱导式欺骗的检测效果。     

机器翻译译文质量估计综述

机器翻译译文质量估计(Quality Estimation,QE)是指在不需要人工参考译文的条件下,估计机器翻译系统产生的译文的质量,对机器翻译研究和应用具有很重要的价值。机器翻译译文质量估计经过最近几年的发展,取得了丰富的研究成果。该文首先介绍了机器翻译译文质量估计的背景与意义;然后详细介绍了句子级QE、单词级QE、文档级QE的具体任务目标、评价指标等内容,进一步概括了QE方法发展的三个阶段: 基于特征工程和机器学习的QE方法阶段,基于深度学习的QE方法阶段,融入预训练模型的QE方法阶段,并介绍了每一阶段中的代表性研究工作;最后分析了目前的研究现状及不足,并对未来QE方法的研究及发展方向进行了展望。     

Decomposition and replacement: Spatial knowledge distillation for monocular depth estimation

Knowledge distillation has become a key technique for making smart and light-weight networks through model compression and transfer learning. Unlike previous methods that applied knowledge distillation to the classification task, we propose to exploit the decomposition-and-replacement based distillation scheme for depth estimation from a single RGB color image. To do this, Laplacian pyramid-based knowledge distillation is firstly presented in this paper. The key idea of the proposed method is to transfer the rich knowledge of the scene depth, which is well encoded through the teacher network, to the student network in a structured way by decomposing it into the global context and local details. This is fairly desirable for the student network to restore the depth layout more accurately with limited resources. Moreover, we also propose a new guidance concept for knowledge distillation, so-called ReplaceBlock, which replaces blocks randomly selected in the decoded feature of the student network with those of the teacher network. Our ReplaceBlock gives a smoothing effect in learning the feature distribution of the teacher network by considering the spatial contiguity in the feature space. This process is also helpful to clearly restore the depth layout without the significant computational cost. Based on various experimental results on benchmark datasets, the effectiveness of our distillation scheme for monocular depth estimation is demonstrated in details. The code and model are publicly available at : https://github.com/tjqansthd/Lap_Rep_KD_Depth.     

基于5G时长驻留的分析及优化提升方法

随着中国移动5G网络的逐步发展，目前已实现城区的5G网络连片覆盖，全网5G用户大规模增长。本文通过对经分数据用户产生的话单，结合用户使用的网络类型，利用成熟的经分平台，实现直接针对用户使用的网络、流量、时长等方面的分析，直观的体现了用户在5G网络的驻留时长占比以及5G网络的流量占比，为网络优化工作提供了重要分析依据。     

Effect of nitrogen on deflagration characteristics of hydrogen / methane mixture

In order to study the influence of nitrogen on the deflagration characteristics of premixed hydrogen/methane, the explosion parameters of premixed hydrogen/methane within various volume ratios and different dilution ratios were studied by using a spherical flame method at room temperature and pressure. The results are as follows: The addition of nitrogen makes the upper limit of explosion of hydrogen/methane premixed gas drop, and the lower limit rises. For explosion hazard (F-number), hydrogen/methane premixed fuel with a hydrogen addition ratio of 10% has the lowest risk, and nitrogen has a greater impact on the dangerous degree of hydrogen and methane premixed gas whose hydrogen addition ratio does not exceed 30%. In terms of flame structure, the spherical flame was affected by buoyancy instability as the percentage of nitrogen dilution increased, but the buoyancy instability gradually decreased as the percentage of hydrogen addition increased. The addition of diluent gas reduces the spreading speed of the stretching flame and reduces the stretching rate in the initial stage of flame development. The laminar flame propagation velocity calculated by the experiment in this paper is consistent with the laminar flow velocity of the hydrogen/methane premixed gas calculated by GRI Mech 3.0. Considering the explosion parameters such as flammability limit, laminar combustion rate and deflagration index, when hydrogen is added to 70%, it is the turning point of hydrogen/methane premixed fuel.     

Strength and deformation properties of frozen sand under a true triaxial stress condition

In recent years, the mechanical properties of frozen soils under complex stress states have attracted significant attention; however, limited by the test apparatus, true triaxial tests on frozen soils have rarely been conducted. To study the strength and deformation properties of frozen sand under a true triaxial stress state, a novel frozen soil testing system, i.e., a true triaxial apparatus, was developed. The apparatus is mainly composed of a temperature control system, a servo host system, a hydraulic servo loading system, and a digital control system. Several true triaxial tests were conducted at a constant minor principal stress (σ₃) and constant intermediate principal stress ratio (b) to study the effect of intermediate principal stress (σ₂) on the mechanical properties of frozen sand. The test results showed that the stress–strain curve can be mainly divided into three stages, with evidence of strain hardening characteristics. The strength, elastic modulus, and friction angle increased with the increase in b from 0 to 0.6, but decreased when increasing b from 0.6 to 1, whereas the cohesion varied little with the variation in b. The deformation in the direction of σ₂ changed from dilative to compressive and that in the direction of σ₃ remained dilative throughout.     

Post-erosion mechanical responses of internally unstable gap-graded soil under drained torsional simple shear and triaxial compression

Internal erosion is a major threat to hydraulic earth structures, such as river levees and dams. This paper focuses on suffusion and suffosion phenomena which are caused by the movement of fine particles in the granular skeleton due to seepage flow. The present study investigates the impact of internal erosion on the dynamic response under cyclic torsional shear and monotonic responses under triaxial compression and torsional simple shear. A series of experiments, using a gap-graded silica mixture with a fines content of 20%, is performed under loose, medium, and dense conditions using a novel erosion hollow cylindrical torsional shear apparatus. The erosion test results indicate that the critical hydraulic gradient and the rate of erosion are density-dependent, where a transition from suffosion to suffusion is observed as the seepage continues. Regardless of the sample density, variations in the radial strain and particle size distribution, along the specimen height after erosion, are no longer uniform. Furthermore, the dynamic shearing results show that the small-strain shear modulus increases, but the initial damping ratio decreases after internal erosion, probably due to the removal of free fines. In addition, the elastic threshold strain and reference shear strain values are found to be higher for the eroded and non-eroded specimens, respectively. Finally, based on drained monotonic loading, the post-erosion peak stress ratio increases remarkably under triaxial compression, while that under torsional simple shear depends on the relative density where the direction of loading is normal to the direction of seepage. These observations indicate that the horizontal bedding plane becomes weaker, while the vertical one becomes stronger after downward erosion.     

Effect of the particle size ratio on macro- and mesoscopic shear characteristics of the geogrid-reinforced rubber and sand mixture interface

As a new type of material for civil engineering projects, the rubber and sand mixture is widely used in roadbed fillers, offering environmental benefits over traditional tyre disposal methods. This study uses a large-scale direct shear apparatus to examine the interface shear properties of the geogrid-reinforced rubber and sand mixture, considering different particle size ratios (r), rubber contents, and normal stresses. Based on indoor tests, direct shear models of the mixture with different values of r are established in PFC^3D, revealing the meso-mechanical mechanism of the mixture in the direct shear process. The results show that when r is greater than 1, incorporating a certain amount of rubber particles can increase the shear strength of the mixture. The r values of 15.78, 7.63, and 3.98 correspond to an optimal rubber content of 30%, 10%, and 20%, respectively. When r is less than 1, mixing rubber particles can only reduce the shear strength of the mixture. When the rubber content is low, the smaller the value of r, the greater is the thickness of the shear band. Furthermore, the normal and tangential contact forces are greater. The fabric anisotropy evolution law of the mixture is consistent with the change in the contact force distribution.