基于多尺度双线性卷积神经网络的多角度下车型精细识别

针对多角度下车辆出现一定的尺度变化和形变导致很难被准确识别的问题，提出基于多尺度双线性卷积神经网络（MS-B-CNN）的车型精细识别模型。首先，对双线性卷积神经网络（B-CNN）算法进行改进，提出MS-B-CNN算法对不同卷积层的特征进行了多尺度融合，以提高特征表达能力；此外，还采用基于中心损失函数与Softmax损失函数联合学习的策略，在Softmax损失函数基础上分别对训练集每个类别在特征空间维护一个类中心，在训练过程中新增加样本时，网络会约束样本的分类中心距离，以提高多角度情况下的车型识别的能力。实验结果显示，该车型识别模型在CompCars数据集上的正确率达到了93.63%，验证了模型在多角度情况下的准确性和鲁棒性。     

Fringing in tubular permanent-magnet machines: Part I. Magnetic field distribution, flux linkage, and thrust force

This paper extends an analysis developed for tubular permanent-magnet machines to account for the effect of fringing associated with the finite length of the ferromagnetic armature core. The magnetic field distribution, established by using an analytical technique formulated in the cylindrical coordinate system, provides an accurate means of evaluating the effect of the fringing flux on the thrust force and the back-electromotive force. Finite-element calculations confirm the analytically derived results. In Part II of the paper, the analysis is used to predict the cogging force that results from end effects and to facilitate minimization of the force.     

Design optimization of radially magnetized, iron-cored, tubular permanent-magnet machines and drive systems

In this paper we deduce, from analytical field solutions, the influence of leading design parameters on the performance of a radially magnetized, iron-cored, tubular permanent-magnet machine and its drive system. We derive analytical formulas for predicting the open-circuit electromotive force, the thrust force, the iron loss, and the winding resistance and inductances, as well as the converter losses. The force density, the machine and drive system efficiencies, and the power factor and converter volt-ampere (VA) rating are established as functions of a set of machine dimensional ratios, with due account of magnetic saturation and subject to a specified thermal constraint. We validate the utility and accuracy of the analytically derived formulas by finite-element calculations. Finally, we show that the design optimization of such a linear drive system must account for the losses and VA rating of the converter as well as the design parameters of the tubular machine.     

A low-power, linear, permanent-magnet generator/energy storagesystem

This paper describes the design, analysis and characterization of a linear permanent magnet generator and capacitive energy storage system for generating electrical power from a single stroke of a salient-pole armature. It is suitable for applications that require relatively low levels of electrical power, such as remote electronic locks. An electromagnetic analysis of the generator is described, and a design optimization methodology for the system is presented. Finally, the performance of a prototype is validated against measurements     

纳米Al2O3颗粒对植物绝缘油微观特性影响的分子模拟研究

采用分子模拟的方法研究了纳米Al2O3对植物绝缘油微观特性的影响,并通过试验加以验证.对纳米Al2O3与植物油分子的表面相互作用机理进行了分析,建立了未改性和改性后的植物油模型,研究分析其中的氢键、油分子的径向分布函数(RDF)和水分子的扩散系数等参数,同时对不同改性浓度纳米Al2O3的植物油进行了热老化实验.结果 表明:改性植物油模型中的氢键数量更多,RDF的峰值更大,水分子的扩散系数更小.热老化过程中改性油的介质损耗小于未改性油,表明纳米Al2O3改性植物绝缘油具有优良的稳定性,改性植物绝缘油的热稳定性和绝缘性能得到增强.     

覆膜滤料用聚四氟乙烯微孔膜的制备及其性能研究

针对市场上常用4种SSG范围的聚四氟乙烯(polytetrafluoroethylene,简称PTFE)树脂原料,采用双向拉伸法制备PTFE微孔膜,分析微孔膜性能的差异,并确定最适用于覆膜滤料用PTFE树脂原料.分析了不同分子量PTFE树脂原料的相对标准密度(SSG)、粒径、挤出压力和含水率等参数,采用不同分子量树脂原料来制备PTFE微孔膜,并表征PTFE微孔膜力学、厚度、结晶度、孔径、孔隙率和透气性能.研究表明:SSG的大小会影响PTFE微孔膜的力学性能(最大力、断裂伸长率)和孔结构(孔径、孔隙率),随SSG的增大呈现先上升后下降的趋势,在2.170～2.189范围内达到最优;结晶度与SSG的大小成正相关,而微孔膜透气性能主要与其孔结构有关,与孔径和孔隙率的变化趋势具有一致性.得出结论:PTFE树脂原料SSG分布在2.170～2.189范围时,制备的PTFE微孔膜最适用于覆膜滤料领域.     

BOILING HEAT TRANSFER ON SURFACES WITH ARTIFICIAL NUCLEATION SITES (Ⅰ) EFFECT OF CAVITY SIZE, CAVITY DENSITY ON BUBBLE DEPARTURE DIAMETER AND FREQUENCY

By means of high-speed photography and optical-fiber probe, the effect of nucleation cavity size and cavity density on the bubble departure diameter and frequency has been investigated. The enhanced heating surface employed has artificial nucleation sites arranged regularly.Based on the analysis of the various forces exerted on the vapor bubble, a mathmatical model has been developed for correlating bubble departure diameter and frequency with the surface geometry and the degree of superheat.Experimental data has confirmed the availability of the model presented.     

我厂解决黄心料的措施

我厂是日产2000t／d熟料新型干法生产线，自投产以来一直存在着黄心熟料问题，虽然黄心熟料对其物理性能影响不大，如水泥强度、安定性、凝结时间，标准稠度需水量等，但是它会使水泥颜色发黄，而且由于黄心熟料易磨性差，水泥磨产量降低，电耗升高，因此我厂近年来采取了一系列的改进措施，使得黄心熟料得以消除。     

用于确定双吸离心泵Suter曲线的三维内特性法研究

水泵Suter曲线是泵站水锤分析的必备条件之一。现有用于获取双吸离心泵Suter曲线的实测法具有难度大、成本高的特点,而间接插值估算法的精度又比较低,本文提出了一种通过水泵流场三维数值计算确定双吸离心泵Suter曲线的新方法,简称三维内特性法。该方法采用计算流体力学(CFD)技术,对双吸离心泵的水力模型进行三维造型、工况离散、流场计算及结果转化,以快速、准确地获取双吸离心泵Suter曲线,在正水泵工况、正水轮机工况、反水轮机工况、反水泵工况等工况下,计算出相应流量、转速、扬程和转矩关系。定量分析表明,相比于间接插值估算法,采用三维内特性法获得的Suter曲线中,Wh(x)的计算精度平均提高了39.2%,Wb(x)的计算精度平均提高了26.3%;在将采用该方法获得的双吸离心泵Suter曲线用于事故停泵工况水锤分析时,计算得到的系统最大压力降低了15.0%,管线发生汽化压力的范围缩减了81.4%,双吸离心泵的最大倒流量增大了11.9%,最大倒转速降低了6.1%,与实际测试结果更为接近。该方法为提高双吸离心泵站水锤分析精度提供了一种新技术途径。     

敏感性问题的一种调查方法

基于调查数量特征的敏感性问题的随机变量乘法模型提出了一种改进模型,该模型比随机变量乘法模型有着更高的精度．由此,将此改进模型推广到最优分配下的分层抽样情形,在总体适合分层时,能提高总体估计量的估计精度;另外,由于在不同层中可以使用不同的随机化装置,这在实际的调查操作中有着更强的可行性．