全油门加速工况车内声品质研究

以23辆乘用车的3挡全油门加速工况的车内噪声为研究对象,对车内声品质采用等级评分法进行主观评价试验,分析计算各噪声样本的心理声学参数和非心理声学的客观参数,并应用多元线性回归理论建立声品质预测模型。研究表明,响度、线性度和粗糙度是影响听众对全油门加速噪声主观感受的最重要的因素,模型预测结果与主观评价试验结果相关系数R~2为0.853,预测值与主观评价实测值吻合度较高,所建立的声品质评价模型在统计学上是有意义的。     

三维空间路线连续性衰退对公路安全性影响分析

为深入分析公路三维空间线形连续性衰退对事故频数的影响,采集广东省内3条高等级公路243 km道路设计资料,共1 768条有效事故数据,利用Tobit回归模型分析三维空间路线连续性衰退对公路安全性的影响.应用微分几何空间的基本理论,提出公路线形连续性衰退表征参数,利用Tobit回归模型对数据样本中单元内总曲率差(TCD)...     

Travel attitudes and mode use among Asian international students at an Australian university

International students and immigrants in general are fuelling the growth in Australia’s population and the pressure that is putting on its urban transport systems. Yet, we know very little about their travel habits and limited previous research has been previously undertaken to address this gap. To address this research need, this study draws on a survey that was designed and distributed online and on-campus to students at Australia’s largest university (Monash University). Using multivariate statistical tools including factor analysis and multinomial logistic regression, significant differences were found between native-born Australian and Asian international students with regards to their mode perceptions, present travel habits and socio-demographics. For Asian students, attitudes towards non-car mode decrease and car-dependency increases the longer they reside in Australia. The findings of this research further highlight the need to encourage newer immigrants to use more sustainable travel options and highlight the relevance for local government and policy planners to support further research in this area, given that these international students have the potential to be key agents for travel behaviour change.     

Exploring vessel-price dynamics: the case of the dry bulk market

We investigate second-hand vessel price heterogeneity. Based on a sample of 5,591 purchase and sale transactions in the dry bulk sector over the period 1998–2016, we deploy a nonparametric regression technique to assess the determinants of vessel price and vessel price variability. Next, we use quantile regression to estimate the effect of regressors (sources of heterogeneity) at different parts of the vessel price distribution. We find evidence that main sources of vessel price heterogeneity are the age of the ship (the older the vessel the higher its maintenance cost), the 3-month LIBOR (reflecting the cost of financing) and the annual charter rate (revenues from operation). Their influence is stronger at all higher quantiles (periods of expansion) and the median of the vessel price distribution.     

Aural comfort prediction method for high-speed trains under complex tunnel environments

Aural comfort is negatively affected during a train’s passage through various tunnel environments. The objective of this study was to propose a prediction model for determining optimal operation parameter combinations to improve train occupants’ aural comfort. High-speed train model tests, combined with a mathematical transfer model, were used to obtain the interior pressure transients under varying speeds, tunnel lengths and seal indexes. Then, a middle ear finite element model was used to simulate the dynamic responses under the pressure transients, and three indicators were employed to assess the severity of aural sensations. Meanwhile, the aural discomfort were classified into four groups according to the duration. Based on the simulation results, the ordinal regression analysis method was used to reveal the effects of the considered factors on aural comfort. The results indicate that aural discomfort sensations begin when a train runs in the middle of a tunnel but are mitigated when it approaches the tunnel exit. Furthermore, aural discomfort is positively correlated with the train speed and the distance from the driver cabin of the head car but negatively correlated with the seal index and tunnel length. As a conclusion, a mathematical prediction model was established that incorporates factors including the train speed, seal index, tunnel length and car position. It can not only forecast aural sensations under certain operation parameters and tunnel environments but also be used for determining the optimal operation parameters to ensure the best aural sensations for high-speed-train occupants.     

卡尔曼滤波算法在空气质量数据校准中的应用

针对自建点实时监控存在的零点漂移和量程漂移导致空气质量数据测量精度受到制约的问题,提出一种基于卡尔曼滤波算法对空气质量数据进行校准的模型。以PM2.5数据为例,结合其他空气质量数据及气象参数经逐步回归得到调整的数据后,使用改进的卡尔曼滤波算法对自建点测量数据进行校准。经同一时间区间内的自建点、国控点测量及校准数据的对比分析表明:通过改进的卡尔曼滤波校准空气质量数据能够很好地跟踪实际数据的变化趋势,平均误差比较小;使用模型将2019年5月、2018年12月数据的均方根误差分别从11.367、27.188降低至修正后的7.555、10.759,精度得到了显著提高。     

基于网联数据的电动车快速充电影响因素分析

基于北京市私家电动汽车网联数据，按照充电行为类型提取车辆行程，并对行程中影响快速充电行为的潜在因素进行细致分析；基于Logistic回归模型进行显著性影响因素识别，结果表明，电动汽车续航里程、出行距离、出行时间等因素显著影响电动汽车的快速充电行为；最后，基于显著影响因素建立模型，对私家电动汽车快速充电行为进行预测，预测结果表明，预测模型具有较好的预测效果和可靠度.本文研究成果将有助于优化私家电动汽车的充电行为，提高充电效率.     

内河船舶的浅水阻力计算

本文提供了一种估算内河船舶浅水阻力的方法。根据184条船模浅水阻力曲线及部分相应的深水阻力曲线,计算了经济航速、临界航速和不同航速下的兴波阻力系数。应用经修正的Prohaska方法计算了深、浅水形状因子。将上述这些量视为H/T、L/B、B/T和C_B的幂函数方程,应用回归分析确定其指数值。利用求得的回归公式即可进行深、浅水阻力计算,协助选择优良船型,进行主尺度分析,选配经济航速以提高营运经济性。本方法具有计算简便、适应范围广、计算精度高等优点。     

利用回归预测技术进行港口吞吐量预测的方法研究

以某港口近15年的货物吞吐量作为原始数据,采用回归分析技术,建立回归预测模型,得到预测结果,为港口的发展战略研究提供一定的技术支持。     

基于三维谱回归的有效三维人脸识别算法

提出一种结合预处理和三维谱回归(3DSR)方法进行三维人脸识别的算法框架,提取有效的判别特征来克服3D人脸图像中一些尚未解决的问题,如噪声、表情和姿态等的影响。首先通过预处理步骤,从输入的人脸图像提取面部区域进行三维数据的匹配,克服大姿态变化的影响并且有效地提高了整个3D人脸识别性能。为处理大的表情变化和数据噪声,引入谱回归的概念,改进的三维谱回归方法可以充分利用局部统计信息的鲁棒性和有效性,并避免通常方法中密集矩阵的特征分解问题,降低了计算复杂度。实验中使用包含大姿态和表情变化的CASIA三维人脸数据库。实验结果显示算法有效、鲁棒、通用性强。