宁波市公共自行车服务满意度评价模型构建研究

以宁波市公共自行车为例，将因子分析法与二项logistic法相结合，对采集的1 688份居民满意度有效问卷展开分析。利用因子分析法筛选出6个因子，利用二项logistic法剔除并筛选出5个有效因子构建公共自行车服务满意度评价模型，合理评估当前宁波市公共自行车服务满意度。同时，利用随机抽取并不参与建模的5%问卷样本对评价模型进行校验，以确保模型的准确性与合理性。该研究有助于合理评价当前宁波市居民对公共自行车的服务满意度，同时适用于对未来一段时间内公共自行车服务满意度的预测。     

序列模式挖掘的两种典型算法及比较

序列模式挖掘是数据挖掘的重要分支,GSP算法与PSP算法是序列模式挖掘中的两种典型算法。本文介绍了这两种算法并对其进行了分析与比较。     

改善20CrMnTiH淬透性的研究

莱钢特钢厂20CrMnTiH钢材成分波动较大,淬透性带HRC8～10,范围较大,制约了齿轮钢质量的进一步提高.通过研究,发现铸坯中心偏析尤其是碳偏析为主要问题.通过工艺优化采取各项措施,尤其增加电磁搅拌后,中心碳偏析得到明显改善;C控制在0.19%～0.20%的炉次达到了95.6%;淬透性带控制在HRC6内的达到了92.1%;在HRC4内的达到了86.2%;提高了产品质量,满足了用户需求.     

数据驱动下交通异常行为的双层识别模型

为实现对车辆异常行为的准确识别,提高车辆行驶的安全性,提出了一种基于支持向量机(support vector machine, SVM)、监督学习与长短期记忆(long short term memory, LSTM)深度学习的交通异常驾驶行为双层识别模型。首先,对车辆轨迹数据筛除和滤波,构建异常行为数据集;其次,从异常行为轨迹特征中提取出特定异常行为的特征标签,并人为标定在训练集中;再次,构建SVM模型对训练集进行粗识别,基于SVM的二分法原理,从测试集中筛选出异常行为;最后,通过LSTM时间序列模型构建具体种类的异常行为模型,并通过深度学习的方法,从异常行为数据中细分为蛇形驾驶、急速变向、侧滑、大半径转弯、快速U型转弯、急刹车等具体的异常驾驶行为。本次实验选用下一代仿真(next generation simulation, NGSIM)数据中US-101高速公路和peachtree城市道路的数据集的轨迹数据验证SVM和LSTM双层识别模型的性能,包括均方根误差、识别准确率等。结果表明,构建的双层识别模型在第一层有98%的识别准确率,第二层有超过80%的识别准确率,可以较为准确地识...     

SrGa2S4:Ce blue TFEL in new structure

SrGa₂S₄: Ce blue TFEL device with a novel structure was prepared by electron beam evaporation. The dielectric constant of insulator layer in this structure is far less than that in the conventional structure, so the hot-electrons can be accelerated in SiO₂ layers. The stable TFEL with good chromaticity was achieved and the basic characteristics were investigated.     

氧同位素：追踪陆架海沿岸水团变化的良好标志

对于陆架海而言除了冰体积效应,温度效应外,还有由于大陆淡水注入引起的,使δ＾１８Ｏ值负向偏移的效应,而且在低海平面时期,当沿岸水团盛行时,这种效应起主导作用。笔者将这种效应从广州的盐度效应中分离出来,称为陆架海的负值效应。     

A new Euprox from the Late Miocene of Yuanmou,Yunnan Province,China,with interpretation of its paleoenvironment

The Yuanmou Basin of Yunnan Province is a hotspot for the research of early homind evolution and its environmental background.During the implementation of the Program“Research of the Origin and Evolution of Early Man and Its Environmental Baqckground” from 1998 to 2001,a huge quantity of hominoid and mammal fossils were collected.The present report systematically describes an Euprox robustus sp.nov.identified during the recent laboratory work on the collected material.Euprox is a group of earliest cervids with true antlers.The new species is the third one of the genus discovered in China.Judged by its morphologic characteristics,the new species feed on juicy and tender leaves of dicotyledon.It implies that the vegetation of its epoch in the Yuanmou Basin is a kind of southern subtropical evergreen forest and the climate is humid and temperate with evident seasonality.The latter is mostly influenced by the monsoon and secondly by the latitude.     

A new species of Gobiconodon （Triconodonta, Mammalia） and its implication for the age of Jehol Biota

A new species of Gobiconodon is found from the Yixian Formation of western Liaonlng, China. The new taxa,G zofiae sp. nov., has a confluent opening for branches Ⅱand Ⅲ of the trigeminal nerve on the anterior lamina of the petrosal. G zofiae sp. nov. is similar to Repenomamus in having an ossified Meckel‘s cartilage connecting the lower jaws and ear region. The new species, with enlarged I^1/I1, posteriorly located infraorbital foramen and four mental foramina,distinctly differs from the other species of Gobiconodon. The new material indicates that Gobiconodon has four, not five,upper molariforms. The presence of Gobiconodon in Jchol Biota makes it possible to correlate Jehol Biota with faunas in eastern Asia and North America, and suggests the age of the Yixian Formation to be Early Cretaceous.     

数字图像中照度不均匀校正技术研究

为了消除数字图像中的照度不均匀性，对现有数字图像的照度不均匀校正技术进行了归类和讨论．分析了这些方法在计算误差上的内在原因，并在此基础上提出了基于小波变换的数字图像照度不均匀校正技术．该技术兼顾了数字图像的频域和空域，使得采用本方法校正后的图像具有品质优，计算误差小和时间消耗较少等特点．     

地铁及轻轨车场计算机辅助设计方法的研究

利用Ｗｉｎｄｏｗｓ３１环境下的ＡｕｔｏＣＡＤＲ１２高级计算机辅助设计及绘图软件包，用Ｃ语言和ＡｕｔｏＬＩＳＰ语言开发地铁及轻轨车场计算机辅助设计软件，以加快设计进度、提高设计精度．实践证明，该软件在地铁及轻轨车场设计和应用过程中，取得了显著的经济效果