基于互相关检测的滚动轴承实时故障诊断方法

提出了一种易于用模拟电路实现的基于互相关检测的滚动轴承实时故障诊断方法,首先,用两个加速度传感器在不同测点采集轴承振动信号,将其分别送入相应通道的高Q带通滤波器来选择最优共振带;然后,将两路带通滤波器的输出信号进行互相关检测,将互相关检测得到的信号经低通滤波器,保留低频故障信号;最后,将低通滤波器输出的时域信号通过频谱分析仪显示滚动轴承故障特征频率的谱线以实现滚动轴承的实时故障诊断。用模拟电路的形式将该方法进行搭建,并在QPZZ-II实测平台完成滚动轴承的实时故障诊断。结果表明:该方法克服了单一信号源的局限性,能利用互相关函数削弱共振带内部噪声,使诊断结果具有更高的频谱辨识率,而且能够用结构简单、易于维护的模拟电路实现,对轴承实时故障诊断方法的应用与普及具有一定的参考价值。     

Distribution and Bioaccumulation of Chlorinated Diphenyl Ethers in a Contaminated Embayment of Lake Ontario

Polychlorinated diphenyl ethers (PCDEs) are halogenated aromatic hydrocarbons that have structural similarities to polychlorinated biphenyls, dibenzo-p-dioxins, and dibenzofurans. PCDEs have been detected in piscivorus fish, fish eating birds and humans, but little is known about the transfer of these compounds through aquatic ecosystems. At sampling stations in the contaminated area of Whitby Harbour on the north shore of Lake Ontario, mean concentrations of total PCDE congeners in sediment were between 622 and 1,929 ng/g dry weight. Analysis of semipermeable membrane devices (SPMDs) deployed at several stations around the harbor indicated that PCDEs were distributed homogeneously in the water. Mean total PCDE concentrations in biota were as high as 672 ng/g wet weight (105 μg/g lipid) in white sucker (Catostomus commersoni) and 791 ng/g wet weight (55 μg/g lipid) in brown bullheads (Ameiurus nebulosus). The patterns of 45 PCDE congener analytes differed among the SPMDs, sediment, and biota, but between invertebrates and fish the congener patterns were relatively homogeneous. Bioamagnification factors (BMFs) calculated for total PCDEs in a simple benthic food chain (oligochaetes to white suckers) and a simple pelagic food chain (plankton to pumpkinseed) indicated that there was biomagnification, with BMFs < 50 on a lipid-normalized basis and BMFs < 5 on a wet weight basis. For individual PCDE congeners, there was a slight trend to increasing BMFs with increasing chlorine substitution.     

Kinetics of diffusion induced grain boundary migration of [100] twist boundaries in the Cu(Zn) system

The kinetics of diffusion induced grain boundary migration (DIGM) in the Cu(Zn) system was experimentally studied for [100] twist boundaries using Cu bicrystals annealed at 693 K for various times between 5.4×10⁴ and 1.73×10⁵ s. The experiment was carried out for bicrystal specimens with misorientation angles of 15, 20, 23 (Σ13), 25, 28 (Σ17), 32, 37 (Σ5), 40, 44 (Σ29) and 45°. During DIGM, the grain boundary migrates rather unidirectionally towards one of the crystal grains for most of the specimens and becomes wavy with increasing annealing time. The migration distance at 5.4×10⁴ s is smaller for coincidence site lattice (CSL) boundaries with low energy than for random boundaries with high energy. The migration rate of the moving boundary was observed to be almost constant independent of the annealing time between 5.4×10⁴ and 1.73×10⁵ s. The steady state migration rate is larger for low-energy CSL boundaries than for high-energy random boundaries. The experimental results were quantitatively analyzed using the energy balance model proposed by Kajihara and Gust. In this analysis, the effective driving force for DIGM was calculated as a function of the migration rate. The calculation indicates that 75–93% of the chemical driving force is consumed by the volume diffusion of Zn in the untransformed Cu matrix ahead of the moving boundary under the given experimental conditions. Considering the energy consumption, the mobility M of the moving boundary was evaluated for the steady state stage. The evaluation yields M=6.6×10⁻¹⁷ m⁴/J s as the mobility of the random boundary for DIGM in the Cu(Zn) system at 693 K.     

Numerical analysis of fluid flow of rotating squared crystal in the pulling method and its effect on growth shape

A numerical analysis of the effect of a rotating squared crystal on the environmental solution flow in the pulling method was performed. These calculation results show a drastic change in the velocity boundary layer thickness around the squared edge at a certain condition. This velocity boundary layer thickness has generally a strong relation with a solute boundary layer thickness and that corresponds to the difference of the crystal growth rate. We calculate the crystal surface spiral ledge growth rate assuming typical BCF model and with the different solute boundary layer thickness at different positions. This solute boundary layer thickness difference brought different ledge growth rates. We calculate ledge growth rate in the case of the small solute boundary layer thickness near the spiral center and large thickness far from the spiral center. At the region of the two solute boundary layer thickness exists, the higher growth rate ledge that located near the spiral center about to reach the far located and low growth rate ledge. So as a result, the ledge spacing becomes smaller at the boundary of the two different thickness boundary layers.     

典型平面角钢桁架静风力风洞试验研究

平面桁架是构成空间桁架的基础,广泛地用于房屋建筑、桥梁、起重机、钢塔等工程结构中。以4种典型单片角钢桁架为研究对象,通过高频动态天平测力风洞试验,研究了风速、风向、杆件布置形式和挡风系数等参数对平面角钢桁架静风力的影响。研究结果表明：平面桁架垂直迎风(0°风向)时,阻力系数值最大,约为1.8;升力系数极值为-1.0,出现在25°风向,与同风向下阻力系数值比较,升力系数的影响不能忽略。风速的变化、杆件布置形式和结构挡风系数对研究的四种典型平面桁架(挡风系数Ф∈［0.21, 0.52］)风力系数影响较小。平面桁架的静力折算高度约为总高度的一半,风向、杆件分布形式和挡风系数等参数对折算高度的影响可以忽略。此外,常用规范中并未考虑平面角钢桁架风力系数随风向的变化,为此,基于试验结果,构建了平面桁架角度风系数公式,并对参数进行了拟合,二者吻合较好。     

硅树脂添加量对氧化铝基陶瓷型芯性能的影响

以氧化铝粉末为基体,添加具有粘结性和一定陶瓷产率的硅树脂粉末,通过干压成型和无压烧结制备出氧化铝基陶瓷型芯,重点研究了硅树脂添加量对氧化铝基陶瓷型芯性能的影响。结果表明:硅树脂在裂解过程中会形成二氧化硅,二氧化硅与氧化铝基体发生反应形成新相莫来石。由于硅树脂在交联和裂解过程中会释放大量气体,导致烧结体失重,且气体的逸出会抑制由烧结引起的收缩,因此,随着硅树脂添加量增加,产生的气体量增加,烧结体的失重率增加,收缩率降低。硅树脂含量的增加使得烧结体的气孔率变大和体积密度减小,烧结体的室温抗弯强度逐渐减小。硅树脂的添加虽然降低了其室温强度,但是保证了陶瓷型芯的尺寸精度。     

水下湿法手工自蔓延焊接技术

基于自蔓延高温合成技术（SHS技术）和湿法电弧焊原理,研究了一种用于水下金属结构应急维修的新型焊接方法—水下湿法手工自蔓延焊接技术.通过焊条结构设计、焊药设计等,研制了水下燃烧型焊条,进行了焊接试验,对接头组织和性能进行了分析.结果表明,该技术可在无电、无气、无其它设备的条件下实现水下金属构件的湿法焊接,单面焊双面成形,接头抗拉强度达267 MPa,冲击吸收能量13.8 J;SEM和EDS分析表明,焊缝金属为以α-Cu固溶体为基体、有大量块状富铁第二相析出的高铁铜合金,熔合区的成分和组织与焊缝第二相基本相同,焊缝合金与熔融母材交互结晶,形成了梯度熔焊连接;拉伸断裂发生在焊缝或熔合区,断口有大量韧窝,属韧性断裂.     

适于隔震结构的两阶段实时子结构试验方法

为在试验中同时模拟隔震结构中隔震层和被隔震结构的非线性,提出一种适于隔震结构的两阶段实时子结构试验方法。第一阶段先将隔震支座单独进行物理试验,通过试验观测数据训练神经网络,使其可以准确拟合隔震支座动力特性;第二阶段基于训练好的神经网络建立隔震层的数值子结构模型,与被隔震结构进行实时子结构试验,完成对隔震结构的动力性能测试。数值仿真中该方法与整体模型仿真结果的峰值误差在3%以内,试验验证中该方法与整体结构振动台试验的峰值误差在6%以内。数值仿真和试验对比表明：提出的两阶段实时子结构试验具有良好的精度。该方法避免了在振动台试验中由于缩尺导致隔震支座受到欠质量的影响,充分保证其力学性能的真实性。同时有效解决了在利用实时子结构试验对各部分都为强非线性的隔震结构进行试验测试时,由于数值子结构建模不准确导致的试验误差问题。为隔震结构的抗震性能测试提供了新的试验方法。     

基于共轭梯度的极速学习机

极速学习机(ELM)由于具有较快的训练速度和较好的泛化能力而被广泛的应用到很多的领域,然而在计算数据样例个数较大的情况下,它的训练速度就会下降,甚至会出现程序报错,因此提出在ELM模型中用改进的共轭梯度算法代替广义逆的计算方法。实验结果表明,与求逆矩阵的ELM算法相比,在同等泛化精度的条件下,共轭梯度ELM有着更快的训练速度。通过研究发现:基于共轭梯度的极速学习机算法不需要计算一个大型矩阵的广义逆,而大部分广义逆的计算依赖于矩阵的奇异值分解(SVD),但这种奇异值分解对于阶数很高的矩阵具有很低的效率;因为已经证明共轭梯度算法可通过有限步迭代找到其解,所以基于共轭剃度的极速学习机有着较高的训练速度,而且也比较适用于处理大数据。