高压蒸汽管道保温研究

计算了不同管径和保温厚度下蒸汽管道输送高压蒸汽的热损失和温度降。结果表明:保温厚度越大,蒸汽的热损失越小、温度降越小;保温材料导热系数与蒸汽的热损失呈正比;环境温度越高,蒸汽的热损失越小;环境风速对热损失的影响不大。同时研究了保温材料选择和保温结构对节能的影响。     

Loss of Curing Agent During Thin Film (Droplet) Curing of Thermoset Material

Recently, it has been reported by our group and others^1.2 that loss of curing agent is encountered during the curing of small droplets or thin films of amine cured epoxies. In our earlier study³ results were reported on loss of curing agent in small droplets used in conducting the rnicrobond, single fiber test for determination of interfacial shear strength (ISS). It was reported that use of a volatile curing agent (meta-phenylene diamine (m-PDA) with DGEBA resin) resulted in increasing amounts of curing agent being lost (as measured by T₈ of the cured droplets) with decreasing droplet size during the processing procedure. Droplets smaller than 150 micrometers were seen to lose up to 40% of the curing agent leading to alteration of the mechanical properties of the droplet and, therefore, causing measured values of ISS to be exceedingly low. Use of a less volatile curing agent (Jeffamine 700, a polyether diamine, Texaco Specialty Chemicals) in combination with DGEBA resin produced results which indicated that loss of curing agent was not occuring. This study was undertaken to show the relationships between film (or droplet) size and the amount of curing agent lost (during the processing) for three different aminecured epoxy systems.     

新型8极9槽永磁同步电机静态场的分析

针对采用分布式绕组的缺点,提出一种采用集中式绕组的新型8极9槽永磁同步电机,并在此基础上改变磁钢的参数,对电机的结构进行优化。     

基于部分加权损失函数的RefineDet

针对目标检测网络单阶改进目标检测器（RefineDet）对类间不平衡数据集中小样本类别检测性能差的问题,提出一种部分加权损失函数SWLoss。首先,以每个训练批量中不同类别样本数量的倒数作为启发式的类间样本平衡因子,对分类损失中的不同类别进行加权,从而提高对小样本类别学习的关注程度;然后引入多任务平衡因子对分类损失和回归损失进行加权,缩小两个任务学习速率的差异;最后,在目标类别样本数量存在大幅差异的Pascal VOC 2007数据集和点阵字符数据集上进行实验。结果表明,与原始RefineDet相比,基于SWLoss的RefineDet明显提高了小样本类别的检测精度,它在两个数据集上的平均精度均值（mAP）分别提高了1.01、9.86个百分点;与基于损失平衡函数和加权成对损失的RefineDet相比,基于SWLoss的RefineDet在两个数据集上的mAP分别提高了0.68、4.73和0.49、1.48个百分点。     

改进YOLOV3算法的视频目标检测

由于监控中的行人检测存在背景复杂,目标尺度和姿态多样性及人与周围物体互相遮挡的问题,造成YOLOV3对部分目标检测不准确,会产生误检、漏检或重复检测的情况。因此,在YOLOV3的网络基础上,利用残差结构思想,将浅层特征和深层特征进行上采样连接融合得到104×104尺度检测层,并将K-means算法聚类得到的边界框尺寸应用到各尺度网络层,增加网络对多尺度、多姿态目标的敏感度,提高检测效果。同时,利用预测框对周围其他目标的斥力损失更新YOLOV3损失函数,使预测框向正确的目标靠近,远离错误的目标,降低模型的误检率,以改善目标间互相遮挡而影响的检测效果。实验结果证明,在MOT16数据集上,相比YOLOV3算法,提出的网络模型具有更好的检测效果,证明了方法的有效性。     

Proton irradiation damage mechanism of the InAlAs／InGaAs／InAlAs quantum well

InP-based high electron mobility transistors (HEMT) have shown a great potential for national defense and satellite radar in space radiation environment applications. This paper studies the proton radiation damage mechanism of its critical structure InAlAs／InGaAs／InAlAs quantum well. The proton projection range and vacancy defect information are obtained at different incident proton energies of 50keV, 75keV and 200keV by SRIM software. With the increase of proton energy, the proton injection depth is increasing and eventually protons pass through the material layers. Besides, the proton radiation induced vacancy defects numbers around hetero-junction increase first and decrease subsequently, and As vacancies are the main proton radiation induced defects. In addition, non-ionizing energy loss (NIEL) of In_0.52Al_0.48As and In_0.53Ga_0.47As material is computed under different incident proton energies by the analytical model. The change trend of NIEL is identical to the induced vacancy numbers, namely, NIEL first increases and then decreases as the incident proton energy increases. Finally the degrading effect of the radiation-induced As defect is detected in the two-dimensional electron gas in the quantum well, which confirms that the major proton radiation damage mechanism of the quantum well is the induced vacancy defects by NIEL.     

基于落锤激励的某新型减振扣件减振性能分析

采用落锤激励,对某新型减振扣件进行现场测试,从传递率、传递损失率、传递损失及Z振级等四个方面对其减振性能进行分析。结果表明,在4~200 Hz范围内,传递率基本都在2.5%以下;传递损失率基本都在97.5%以上;传递损失基本都在32 dB以上;道床垂向振动加速度经过计算后,其Z振级为86.15 dB;跟普通减振扣件相比,新型扣件的综合减振性能有了明显地提高。试验表明,落锤激励法操作方便,行之有效,是测试扣件减振性能的优良方法。     

LTE-A系统中基于资源利用率的时延优先调度算法

针对现有LTE系统中调度算法在吞吐量、公平性及丢包率性能指标上难以取得性能平衡的问题,提出一种改进算法,在考虑用户时延的前提下还考虑了用户对资源块(RB)的利用率,从实际获得速率的角度来更精确描述用户对资源的利用能力,更加合理地分配资源.仿真结果表明,该算法的丢包率性能有很大改进,同时在吞吐量、公平性方面也有着较好表现,实现了良好的综合性能,能够更好地满足用户的QoS需求.     

微小型涡喷发动机向心涡轮的流动仿真

为了详细研究微型向心涡轮的内部流动特性,对一台自研微型涡轮发动机的涡轮进行了气动设计和全三维粘性数值模拟仿真,验证了叶片型线设计方案的可行性,得到模拟计算结果（落压比为1.91,总压绝热效率0.73,出口气流马赫数为1.2,流量13.8g/s）。研究了微型涡轮设计工况下内部的气流状态,对微尺寸下涡轮内部的流动特性及各种流动损失产生的特点及原因进行分析总结。结果表明：导向叶流道内的流动损失主要为由气体粘性引起的叶型损失;动叶流道中流动损失主要为气体在动叶前缘产生冲击形成的滞止损失与边界层损失,动叶出口叶片尾缘处形成的尾迹涡流损失。     

矩形断面明渠中恒定均匀流的机械能损失

明渠流中机械能损失的确定是水力学乃至工程流体力学中的重要内容.以构建的明渠恒定流新的总流能量方程及矩形断面明渠三维层流运动的理论解为基础,通过理论分析与三维数值计算,就矩形断面明渠中恒定均匀流的机械能损失系数随雷诺数及宽深比的变化进行了深化研究,得到如下成果:1)在层流条件下,机械能损失系数一般均与雷诺数和宽深比有关,只有对宽深比大于6.5的宽浅明渠,其相应流动的机械能损失系数才仅与雷诺数有关,且在雷诺数相同的条件下,一般矩形断面明渠流的机械能损失系数要大于宽浅明渠流的机械能损失系数;2)在紊流条件下,机械能损失系数也一般均随雷诺数与宽深比而变,但对给定的宽深比,存在相应的特征雷诺数,当流动的实际雷诺数大于该特征雷诺数后,机械能损失系数则不再随宽深比而变化,且在宽深比一定的条件下,雷诺数越小,其机械能损失系数的修正值越大.