铸态Mg-8Y-6Gd-1Nd-0.17Zn稀土镁合金高温压缩本构行为及加工图

研究了铸态Mg-8Y-6Gd-1Nd-0.17Zn镁合金在应变量为50%、温度350℃~450℃、应变速率0.0001s-1~0.1s-1的范围内热压缩过程中的本构行为、组织演变和热加工性能。通过选用双曲正弦本构方程来描述合金的流变行为以及变形参数间的关系。实验结果表明，温度和应变速率对Mg-8Y-6Gd-1Nd-0.17Zn镁合金的流变应力行为有重要影响，其流变应力随温度的降低和应变速率的增加而增大，并且在温度高于400℃压缩时，合金的真应力应变曲线具有典型的动态再结晶特性。在本实验条件下，该合金变形期间的活化能（Q）和应力指数（n）分别为359.258 KJ / mol 和5.24，实验值与计算值之间的平均误差（ARE）为3.37％。最后基于动态材料模型加工理论，结合热加工图和压缩过程中的组织演变，确定了该合金的最佳热加工参数为：加热温度400~450℃，应变速率为0.0001s-1~0.001s-1。     

不同干密度尾矿料压缩特性探究

为研究不同干密度尾矿料的压缩特性,采用筛分法将尾矿料分别配制成不同干密度的尾矿砂及尾矿土,进行渗透、压缩试验。试验表明:尾矿砂的沉降主要发生在施加荷载的早期阶段,且沉降量较大。压缩模量随着干密度的增大而增大,压缩系数随着干密度的增大而减小,在相同压力增量下,尾矿土的压缩性更强。     

采空区矸石压实过程对煤柱支承压力的影响

为了研究采空区压力及煤柱上支承压力随采空区矸石压实程度的变化规律,通过实验室试验、数值模拟等方法,探讨建立了采空区矸石压缩模量与矸石压实率之间的关系式,分析了采空区矸石压力分布随矸石压实率变化规律;基于弹性基础梁理论和弹塑性极限平衡理论建立了煤柱上支承压力与采空区矸石压力间的关系式,分析了煤柱支承压力峰值大小、峰值位置受矸石压实率影响规律。最后,将采空区矸石压缩模量与矸石压实率之间的关系用于FLAC3D数值模拟,与理论形成对比,证明二者结果基本一致。研究表明:矸石压实率越大,采空区压力回升越快,压力恢复至原岩应力的位置距煤壁越近,对应的煤柱上支承压力峰值位置越靠近煤壁,压力峰值越小。     

基于VMD与时间序列分析的滚动轴承故障特征提取方法

为了在非线性、非平稳的滚动轴承故障振动信号中有效提取出敏感的故障特征,提出了基于变分模态分解(VMD)与时间序列分析相结合的特征提取方法。首先通过VMD将原始信号分解为不同预设尺度的本征模态分量(IMF),对各个IMF分量建立时间序列预测模型,通过叠加重构得到最终的预测模型,比较评价指标确定最优参数的选取。最后,通过仿真信号与滚动轴承实际故障数据分析,并与经验模式分解(EMD)进行对比,结果表明该方法能够有效的提取到故障特征频率。     

基于VMD瞬时能量与GA- RBF的滚动轴承故障诊断

针对滚动轴承早期故障的有效识别,提出了一种基于VMD瞬时能量与GA优化的RBF神经网络的滚动轴承故障诊断方法,可以有效对滚动故障做出诊断。首先,VMD将滚动轴承振动信号进行分解成合适数目的本证模态函数;其次,计算本证模态函数分量的瞬时能量并组成特征向量;最后,将特征向量输入到GA优化的RBF神经网络实现轴承故障识别。通过滚动轴承故障诊断实验对该方法进行验证。结果表明,该方法识别滚动轴承故障的准确率为96.43%,较默认参数的RBF神经网络和EEMD瞬时能量与GA-RBF神经网络有明显的提高,证明了所提方法的可行性。     

随机森林算法在温度分布重建中的应用

为提高温度分布重建精度,提出了使用随机森林算法对温度测点进行优化布置的新方法。 将测点位置作为样本特征, 以不同的测点布置方式及其对应的重建误差作为样本数据集。 使用样本数据集构建随机森林模型,评估样本特征重要性,根据 特征重要性排序实现温度测点的优化布置。 设定仿真实验与燃烧实验验证优化布置算法的可行性与有效性。 分析实验数据, 所提出的优化布置算法相对于原有算法,重建精度提升了 20%以上。 研究结果表明,随机森林算法在温度分布重建中具有良好 的应用价值,并为解决工业实际问题提供了新思路。     

Simulation of integrated novel PSA/EHP/C process for high-pressure hydrogen recovery from Coke Oven Gas

This paper introduces a novel Coke Oven Gas (COG) hydrogen purification/compression system based on the technologies of Pressure Swing Adsorption (PSA) and Electrochemical Hydrogen Purification and Compression (EHP/C). As the EHP/C tolerates O₂, N₂ and CH₄ impurities, PSA can be utilized solely for CO and CO₂ removal (other COG impurities were not considered in this work). A relaxation of PSA hydrogen purity could significantly enhance its recovery rate. In this study, the suitability of traditional hydrogen PSA as part of the hybrid PSA/EHP/C approach was investigated. Aspen Adsorption and Matlab were used to model the PSA and EHP/C systems, respectively. The effect of adsorption pressure, purge-to-feed-ratio (P/F-ratio) and adsorption time within cycle on PSA performance is reported. This study found that breakthrough of non-detrimental components is typically accompanied with poisonous CO. Hence, the CO removal with traditional H₂-PSA resulted into high purity product. In a two-bed PSA, 36.3% of hydrogen was recovered at 99.9988% purity and 0.18 ppm CO. Subsequently, as a result, the EHP/C purification capability was merely utilized, but polished this hydrogen to >99.999% purity. Simultaneously, hydrogen was isothermally compressed to 20 MPa, consuming a marginal 2.42 kWh/kg. Compared to mechanical compression, this is 31.6% more energy efficient. Recovering hydrogen from by-product COG was found to save 0.5 kg CO₂/kg H₂ compared to hydrogen produced from natural gas. Conventional hydrogen PSA, utilizing 70% Activated Carbon and 30% Molecular Sieve 5A, was found not to be effective to target the removal of CO specifically. To increase synergy between PSA and EHP/C, the PSA requires adequate design and operation using appropriate adsorbents and cycle steps to target elimination of CO. An increased EHP/C catalyst tolerance for CO also contributes to higher flexibility.     

基于动态行为学习的空中目标识别方法

针对空中移动目标识别问题,提出了一种基于动态行为学习的识别方法。首先,从数据源中提取目标要素,经过标注形成目标数据特征集;其次,基于应用场景进行特征组合,形成特征模型要素;然后,基于深度学习算法进行模型训练,得到评估最优的模型参数;最后,利用训练后的模型进行目标识别处理。通过数值仿真验证,该方法能够有效地识别无业务规则的目标身份。     

Comparative study of granular solid-structure interaction in a uni-axial compression system

Interaction between granular solids and confining structures is an elementary problem encountered in subsurface structural design and bulk solids storing and handling. A classic scenario is uni-axial compression of granular solids in a deformable cylindrical container. Despite being apparently simple in loading condition, the understanding of this scenario remains limited, mainly due to complex interactive deformation between the two components via frictional interfaces. This paper comparatively examines such a uni-axial compression particulate system by a laboratory experiment and two different numerical approaches, namely, continuum finite element method (FEM) and linked discrete-finite element method (linked DEM-FEM). In the continuum FEM approach, two intendedly chosen simple material models, linear elastic and porous elastic models, are attempted. The comparative study reveals that the majority of resultant characteristics show satisfactory agreement amongst the numerical predictions and the experimental measurements. The simple elastic continuum FEM models can hence be a useful alternative in modelling such problems with mild structural flexibility under a monotonic loading scenario. However, precise prediction of some characteristics, such as lateral pressure ratio, may demand more elaborated material model or parameter selection. The enhancements needed for each numerical approach in order to achieve an improved result are further discussed.     

Estimating posterior inference quality of the relational infinite latent feature model for overlapping community detection

Overlapping community detection has become a very hot research topic in recent decades, and a plethora of methods have been proposed. But, a common challenge in many existing overlapping community detection approaches is that the number of communities K must be predefinedmanually. We propose a flexible nonparametric Bayesian generative model for count-value networks, which can allow K to increase as more and more data are encountered instead of to be fixed in advance. The Indian buffet process was used to model the community assignment matrix Z, and an uncollapsed Gibbs sampler has been derived.However, as the community assignment matrix Z is a structured multi-variable parameter, how to summarize the posterior inference results and estimate the inference quality about Z, is still a considerable challenge in the literature. In this paper, a graph convolutional neural network based graph classifier was utilized to help to summarize the results and to estimate the inference quality about Z. We conduct extensive experiments on synthetic data and real data, and find that empirically, the traditional posterior summarization strategy is reliable.