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。     

Trace elements in particles of motor vehicle exhaust in Shanghai

A nuclear microprobe with high spatial resolution and high analytical sensitivity was applied to analyze trace elements,especially lead,in vehicle exhaust of Shanghai city.The result shows that the chemical composition and its corresponding x-ray relative intensity are different among different vehicle exhausts.There are many kinds of metal elements in particles of vehicle exhaust.most are harmful to people,such as Ti,Cr,Mn,Pb,etc.We found that the lead concentration was 6820μg/g and the bromine concentration was 5300μg/g in the exhaust from Santana using leaded gasoline(SULG).which is higher than any other kinds of vehicle exhausts.We have also detected the minimum lead in the particles of unleaded gasoline and its content varies from one to another.Its mean concentration was 450μg/g and the highest reached 6210μg/g.The unleaded gasoline‘s Pb existed in the whole particle while the leaded gasoline‘s enriched in the surface of the particle and was more harmful to the human beings.     

我国清洁代用燃料汽车发展状况

随着汽车数量的迅速增加，汽车排放的有害废气也急剧增加，人们的健康受到威胁，解决这一问题的最好办法是发展清洁代用燃料汽车。介绍了天然气汽车和液化石油气汽车的主要优点，分析了目前我国清洁代用燃料汽车的发展现状，阐述了近期适宜发展我国清洁代用燃料汽车的机遇和面临的困难。最后就提高人们的环保意识、修改和完善汽车尾气污染控制法及排放标准、清洁代用燃料汽车的有关标准和安全措施。对发展这种汽车应给予优惠政策以及零部件生产和车辆改装企业的管理等方面，提出了建议。     

2010-2020年深圳市机动车发展和油气需求量预测

介绍了深圳市机动车发展现状,分别用趋势外推法、城市比较分析法和分类法预测了2010-2020年深圳市的机动车保有量,最后预测了机动车的油气需求量。     

液体火箭发动机研制情况的一些回顾

４０年来，我国液体火箭发动机的研制工作取得了很大成绩。文中对发动机的一些研制情况作了回顾，对ＹＦ－７３和ＹＦ－７５氢氧发动机的设计特点、飞行结果作了简单叙述。对今后工作也提出了看法。     

一步法合成松香改性酚醛树脂

探讨了一步法合成松香改性酚醛树脂的反应机理，确定了最佳原料组成及合成工艺，由此制备出高溶解粘度和矿油容纳度的松香改性酚醛树脂，是一种性能优异的高级胶印油墨连接料     

OSEK/VDX——汽车电子系统的开放式平台

介绍了当前在国际汽车工业界日益占据主导地位的汽车电子系统开放式平台--OSEK/VDX体系.详细地论述了这种体系的由来、运行机制和最新的发展动向,以及它对汽车电子技术发展的深远影响.     

基于小波分析的车型识别

本文论述了用小波分析提取车辆图像边缘信号的方法，得到了连续、光滑的边缘图像，并将其应用于公路车辆车型的检测与识别系统中，实验表明识别过程准确、速度快，具有较好的应用前景。     

DSP技术在液压混合动力汽车能源控制方面的应用

文章分析了液压混合动力汽车的发展趋势,简述数字信号处理器(DSP)作为控制器的优势和DSP作为控制器对并联式混合动力液压车的能源控制的工作原理,给出了能源控制系统DSP信号流程图和主程序流程图。测试结果表明,DSP能源控制系统能够比较精确的对液压混合动力车的能量消耗与分配进行控制,具有很好的应用前景。     

Design of nonlinear controllers for active vehicle suspension with state constraints

Design of active vehicle suspension has tradeoffs between three main performance metrics (passenger comfort, suspension deflection and road holding ability). It has been known that each performance can be achieved by H _∞ controller and they can be gathered by LPV (Linear Parameter Varying) method. However, because the suspension deflection limit was not explicitly considered, this limit may be exceeded. In this paper, the authors propose a “reference shaping“ based method in order to improve the control performance. In this approach, a “virtual reference” signal is imposed to the system such that the suspension deflection is kept small. The effectiveness of the approach is examined by numerical simulations. This work was presented in part at the 13th International Symposium on Artificial Life and Robotics, Oita, Japan, January 31–February 2, 2008