FIELD-NETWORK MODEL FOR MINE FIRE SMOKE MOVEMENT

ＦＩＥＬＤＮＥＴＷＯＲＫＭＯＤＥＬＦＯＲＭＩＮＥＦＩＲＥＳＭＯＫＥＭＯＶＥＭＥＮＴ①ＪｉａｎｇＪｕｎｃｈｅｎｇ，ＷａｎｇＸｉｎｇｓｈｅｎＤｅｐａｒｔｍｅｎｔｏｆＭｉｎｉｎｇＥｎｇｉｎｅｒｉｎｇ，ＣｈｉｎａＵｎｉｖｅｒｓｉｔｙｏｆＭｉｎｉｎｇａｎｄＴ...     

复杂轴系扭振动力学建模网络法

提出了一种复杂轴系扭振动力学建模的网络法,该方法将轴系各支路特性及节点和终端边界条件用规范的矩阵形式表示,通过简单的矩阵运算将网络各支路与边界条件耦合起来,并可求得剩余扭矩函数。针对剩余扭矩方程求解中出现的“增根”问题,提出了一种简便的技术途径。算例证明了该方法的实用性和准确性。     

大红山铁矿数字化矿山建设的探讨

随着信息技术的快速发展,用信息技术武装矿山企业,是矿业企业的大势所趋,同时,也是提高矿业企业科学管理的有力手段。     

690合金中晶界网络分布的控制及其对晶间腐蚀性能的影响

利用形变及热处理工艺提高了690合金的低Σ重位点阵（Coincidence Site Lattice, CSL）晶界比例,通过电子背散射衍射（EBSD）技术表征了由不同类型晶界构成的网络特征,结果表明通过晶界工程处理,能够形成以大尺寸“互有Σ3n取向关系晶粒的团簇”显微组织为特征的晶界网络分布,这种显微组织是再结晶过程中多重孪晶充分发展的结果。通过晶间腐蚀浸泡实验表明通过晶界工程处理的样品抗晶间腐蚀性能较未经过晶界工程处理的样品明显提高。腐蚀后样品的显微形貌表明大尺寸“互有Σ3 n 取向关系晶粒的团簇”能够阻止晶间腐蚀向样品内部扩展,并且能够保护下层的显微组织。     

热处理过程中Mg-2Nd-4Zn-1Zr合金微观组织的结构特征(英文)

对比研究了Mg-2Nd-4Zn-1Zr合金在铸态和热处理态下显微组织结构的演变。借助TEM、SEM和XRD等研究发现,合金中的第二相有几种不同的种类特征。在铸态下,合金中存在的共晶化合物是由Mg、Nd和Zn构成的两种不同类型的三元相:T相和W相。然而,经过热处理,随着时间和温度的变化,T相几乎完全溶解进入α-Mg基体中,但W相仍然大量剩余在基体中。此外,随着热处理时间的延长,沿晶界分布的第二相的形貌由连续分布的网状逐渐转变为球状。在热处理过程中,合金中W相是一种高热力学稳定相,其数量随着热处理的进行而逐渐减少,并最终粗化、稳定地存在于基体中。     

CO_2腐蚀速率半经验预测模型研究

以Nesic机制为基础,建立了新的腐蚀速率预测模型,着重考虑了离子在流体边界层和腐蚀产物膜中的传质速率,提出了腐蚀产物膜因子的实验确定方法.进一步,通过高温高压CO2腐蚀的实验,给出了从具体实验结果来分析和确定腐蚀产物膜影响因子的技术.分析了矿化度对腐蚀速率的影响规律,结果表明模型与实际情况符合良好.     

Sintering and microstructure evolution in columnar thermal barrier coatings

Sintering of thermal barrier coatings changes their key properties, such as thermal conductivity and thermal shock resistance, thus adversely impacting their reliability. We present a novel modeling approach to study the evolution of coating structure during sintering. We model the sintering of individual columns using a thermodynamic principle, and incorporate the center-to-center approach rates for the columns calculated using this principle in a larger scale discrete dynamics model for the evolution of a large number of columns. Surface energies, grain boundary energies and strain energies associated with the deformation of the columns are all included in this framework, while sintering is assumed to occur by the concerted action of surface and grain boundary diffusion. Two sets of initial conditions corresponding to different extents of pre-sintering among neighboring columns are considered. When the extent of pre-sintering is small, we observe that small clusters containing 5–20 columns are formed. In contrast, where a larger amount of pre-sintering exists, we observe, especially at large column densities, that clusters containing 50–100 columns separated by large inter-cluster pores/channels that appear to organize themselves into a network are formed. These observations are in good agreement with recently published experimental observations. We also explain how these results can explain the development of a “mud-crack”-like pattern.     

基于深度学习的大规模图像高效识别算法研究

面对规模较大的图像识别任务时,基于卷积神经网络的深度学习方法存在训练时间过长的问题,导致识别效率不高。因此,提出一种基于局部特征深度信念网络的大规模图像高效识别算法。首先,该方法从原始图像中提取多个局部特征,并根据分配给图像的标签将每个局部特征分类。然后利用分类后的图像局部特征训练深度信念网络,获得网络的相关参数。最后利用深度信念网络进行图像识别。在CAS PEAL R1 大规模图像数据集上进行了图像识别实验,结果显示:提出的算法优于其他深度学习方法,具有较好的准确性和高效性。     

Grain shrinkage driven by surface and grain boundary energy in Ba5Nb4O15 powder

The shrinkage and disappearance of small Ba₅Nb₄O₁₅ grains in a large grain matrix at 1133 K observed by transmission electron microscopy (TEM), included grain volume and boundary shrinkage. Rate equations for these processes were formulated based on the concept that total excess free energy directly stimulates material transport in volume, surface, or grain boundary diffusion. Based on these equations, grain vanishing was simulated and it was found that volume diffusion combined with boundary diffusion occurred, and high grain boundary energy and low grain boundary diffusivity made grains vanish while maintaining a truncated spherical shape.     

LCD模糊熵和SOM神经网络在液压泵故障诊断中的应用

针对液压泵故障诊断问题,提出了一种基于局部特征尺度分解(Local Characteristic-scale Decomposition,LCD)、模糊熵和SOM神经网络三者相结合的故障诊断方法。对液压泵振动信号进行LCD分解,得到若干个内禀尺度分量(Intrinsic Scale Component,ISC);将ISC分量分别与原信号进行相关分析,筛选出包含主要故障信息的前几个ISC分量,计算其模糊熵并组成特征矩阵;将特征矩阵输入SOM神经网络进行分类识别。液压泵故障诊断实例表明,该方法能够准确识别液压泵典型故障,具有一定优势。通过与BP神经网络分类结果相对比,显示了SOM神经网络在特征分类方面的优越性。     

铸件三维充型过程耦合数值模拟

采用改进后的ＳＯＬＡ－ＶＯＦ方法处理自由表面,有效地克服了传统的ＳＯＬＡ－ＶＯＦ法中的“施体－受体”法在处理三维自由表面时精度差和时间步和菠的缺点。在自由表面处理上采用了新的压力插值函数,使三维自由表面的处理变得简单明了。用有限差分法建立了铸件三维充型过程流动与传热的耦合计算数值模型,并对大型灌壁铅合金件进行了数值计算。后处理中采用插值技术,直观地显示了充填过程中自由表面的位置和形状。     

液压支架油缸内表面的腐蚀原因

采用化学成分分析、金相显微、能谱、X射线衍射和电化学等手段对某煤矿液压支架油缸内孔腐蚀失效件进行了检测分析。结果表明:该油缸内表面的腐蚀以点蚀为主,腐蚀产物中含有一定量的S元素;该矿配液水质和液压液浓度均不符合MT76-2002标准要求,油缸内表面的点蚀是由液压液使用不当造成的。     

热处理对含硼低合金高速钢组织和性能的影响

研究了淬火和回火温度对含硼低合金高速钢显微组织和性能的影响.结果表明,含硼低合金高速钢组织主要由马氏体和碳硼化合物构成,铸态下碳硼化合物主要以连续网状分布,而淬火后,碳硼化合物出现了断开的倾向,随着淬火温度提高,碳硼化合物的断开倾向越明显,而且部分碳硼化合物溶人基体,促进了基体的强化.回火温度的变化,对碳硼化合物的影响不明显,但随着固火温度的升高,马氏体中会析出二次碳硼化合物,使基体中合金固溶度下降.随着淬火温度上升,含硼低合金高速钢轧辊材料硬度提高.随着回火温度上升,硬度开始下降.     

Fracture roughness scaling and its correlation with grain boundary network structure

Roughness scaling laws for intergranular cracks deviate from self-affine (fractal-like) behavior at length scales related to the polycrystalline microstructure. We consider two versions of the same alloy material with many of the same microstructural length scales but differing in their processing history: one conventional and one grain boundary engineered. The engineered material, processed to contain a high fraction of “special” grain boundaries, fails more slowly and more isotropically. We present evidence that the difference is determined by processes related to clusters of twin-related grains, shown through analysis of scales of the fracture roughness measured with confocal microscopy and the special grain boundary network determined by electron backscatter diffraction. Above the cluster scale, the fracture roughness exponents in the two materials are nearly indistinguishable (confirming theoretical predictions); below this scale conventional cracks exhibit correlations indicating consistently weak paths for crack propagation, suggesting percolation of “random” boundaries.     

磁流变液阻尼器阻尼通道内部塞流区边界位置的分析

磁流变液阻尼器阻尼通道内部磁流变液运动时塞流区边界位置与输出阻尼力之间存在着密切联系。为了得到塞流区边界位置的具体数值和变化规律,基于RD-8041-1型磁流变液阻尼器,对阻尼通道内部磁流变液的剪切应力和流动状态进行了分析,并得到了塞流区边界位置与活塞运动速度之间的函数关系。为了验证所得函数关系的正确性,在推导出磁流变液阻尼器输出阻尼力数学表达式的基础上,根据塞流区边界位置与活塞速度之间的函数关系得到了磁流变液阻尼器在输入电流为1 A时的输出阻尼力,并将计算得到的阻尼力与实际试验测试值进行了对比。对比结果表明:根据塞流区边界位置与活塞速度之间的函数关系计算得到的阻尼力与实测值保持了一致,从而验证了塞流区边界位置与活塞速度之间函数关系的正确性。     

基于小波分析和RBF神经网络的轴承故障诊断研究

为了提高轴承故障信号的诊断性能，采用小波分析和RBF神经网络相结合的方法对轴承振动信号进行故障分类。首先对轴承振动信号进行小波变化，采用软阈值去噪方法滤除振动信号噪声，然后对振动信号矩阵化处理，接着构建RBF神经网络，输入轴承振动信号特征向量，初始化权重和阈值，最后通过不断反向迭代得到稳定的RBF神经网络故障判别模型。实验证明：通过差异化设置隐藏层神经元数量，确定合适的RBF神经网络规模，经过小波去噪可以有效提高轴承故障判别准确率，相比于常见轴承故障分类算法，算法具有更高的故障判别准确率。     

基于轻量级卷积神经网络的人脸检测方法研究

针对现阶段卷积神经网络参数量较大，检测速度较慢，无法嵌入至移动端电子设备，且在复杂环境下检测精度较低的问题，设计了两层前后分离轻量级的卷积神经网络的人脸检测方法。第一层网络采用全卷积神经网络，用于快速提取人脸特征，并生成大量的人脸边界候选框。第二层网络采用深层全连接卷积神经网络，将第一层网络推断的人脸候选区域进行筛选，并输出人脸大小、坐标和置信度。实验表明，本文设计的人脸检测方法在人脸基准数据集FDDB上具备较高的检测精度和检测速度，轻量级的网络设计使得算法移植到前端电子设备成为了可能。     

Continuous recrystallization in austenitic stainless steel after large strain deformation

Static restoration mechanisms operating during annealing were studied in a 304 steel with strain-induced submicron grain structures. The initial microstructure with an average grain size of about 0.3 μm was developed by large strain deformation at 873 K. Early annealing leads to a full relaxation of high internal stresses associated with non-equilibrium strain-induced grain boundaries, while their boundary misorientations and the average grain size barely change. Further annealing results in a transient recrystallization followed by a normal grain growth. The average grain boundary misorientation increases up to around 45° in the former and becomes constant in the latter. This is associated with the change in the grain boundary misorientation distribution from a characteristic strain-induced one to a near random distribution corresponding to a fully recrystallized state. The annealing processes operating in the strain-induced fine grains take place homogeneously in the whole matrix and can be called continuous recrystallization.     

液力变矩器流场计算可视化及其OpenGL实现

本文利用VC完成了液力变矩器流场可视化数据接口的编制，并调用OpenGL实现矢量场显示及叶片NURBS曲面绘制，实现了ANSYS软件矢量场后处理的基本功能及ANSYS软件所不能满足的诸如特定剖面动画等的显示功能，增强流场计算结果显示的灵活性和真实性。     

Grain boundary faceting and abnormal grain growth in BaTiO3

When 0.1 mol% TiO₂–excess BaTiO₃ was sintered below the eutectic temperature in air, abnormally large grains formed in the fine matrix BaTiO₃ grains. The abnormal grains contained {111} twin lamellae, while the matrix grains did not. A TEM observation revealed that almost all the grain boundaries were faceted. On the other hand, however, when the air-sintered sample with faceted grain boundaries was annealed in H₂, the faceted boundaries became defaceted, and the growth of abnormal grains was suppressed while the growth of the matrix grains was enhanced, showing normal grain growth behavior. In addition, the abnormal grains that had been elongated along their twin lamella grew rather isotropically, irrespective of the presence of {111} twins. It appears therefore that {111} twins appear to enhance the growth of the abnormal grains along the twin lamellae only when the grain boundary is faceted. After re-annealing the H₂-annealed sample in air, however, the grain growth behavior and grain boundary structure were found to recover those observed in the air-sintered sample. From these observations, it is concluded that abnormal growth of BaTiO₃ grains observed is related to grain boundary faceting and that boundary faceting is a necessary condition for abnormal grain growth.