基于动态时间弯曲的多模板匹配车型分类

针对环形感应线圈采集数据的特点,讨论一种基于动态时间弯曲(DTW)算法的车型分类新方法。由于采样波形与车速、车辆自身结构以及其通过线圈时的相对位置有关,导致实际同类车辆波形存在伸长、压缩,甚至波形残缺不全或局部振荡等复杂情况。为此,本方法首先对样本数据进行降噪滤波,补齐不完整信号等预处理;熟后结合DTW方法和最优聚类分析原理选取同类车型的多个模板,综合考虑误纳与误拒情况,可降低单模板匹配存在的误拒率;最后利用DTW算法和多个分类区分度指标进行多模板匹配,得出分类结论。论文仿真部分以一类车型为例,分析了本文方法与其他方法比较存在的优势。     

Categorisation for air quality assessment in car cabin

The objective of this work was to propose a way of preparing information about in-cabin air quality for the car driver or passenger. It was assumed that information should be objective, relevant, current, complete and useful. The major elements of our concept were: (1) monitoring of selected air parameters inside the cabin, (2) determination of indicators that characterise indoor air, based on measurements, and (3) the method of information extraction. We proposed to realise the first task with a sensor system. The second and third goals were reached by categorisation. The basis for defining categories were ranges of values of measured parameters or their combinations. In a way, the categories were used to quantify indicators of air quality. Shannon entropy and mutual information were applied to find the best categorisation. The concept was investigated using experimental data from car cabin air monitoring in various driving conditions. We conclude that information about air quality in car cabins may be successfully conveyed using the following indicators: thermal conditions – determined based on temperature and relative humidity measurements; air exchange – determined based on CO₂ concentration; and air freshness – determined based on volatile organic compounds (VOCs) content in car cabin air. Each indicator has three categories: bad, intermediate and good.     

基于Markov链修正的铁路运量灰色组合预测模型研究

预测模型的准确与否直接决定着未来经济规划与决策的有效制定。将灰色GM(1,1)-Verhulst组合预测模型与马尔可夫链方法相结合,同时引入信息熵理论的知识,提出基于Markov链修正的熵权法灰色组合预测方法,并以甘肃省2004年～2015年铁路客运量作为原始数据序列进行模型拟合,而且还以此为基础对甘肃省未来几年内的客运量发展趋势进行预测。结论:(1)在已知实际客运量年份内,该灰色组合预测模型的预测精度比单一灰色预测模型更高、更加准确;(2)采用马尔可夫链方法获得该组合模型的偏差规律,并依照此规律对预测结果进行修正,即由一个单一的预测数值修正成为区间和概率组成的预测范围;(3)通过比较2016年～2017年的客运量实际值、组合预测模型的单一预测值和Markov链修正的预测区间值,发现Markov链修正的预测结果与客运量实际值的吻合性良好,进一步验证此预测方法的可信性。     

基于熵权理想点法的铁路线路方案优选研究

铁路线路走向方案优选通常是一项复杂的系统工程,需要尽可能拟定多个方案,多角度、多层次的比选,所以是一个多目标的综合决策问题。传统的方案优选方法主要考虑工程投资,同时对其他指标进行定性比较,进而对方案做出评价,如果在经济指标差别不是很大的情况下,由于缺乏全面决策信息,很难选出较优方案。引用多目标决策方法熵权理想点法,应用于新建高速铁路西安至延安线铜川至富县段方案优选,在综合考虑其影响线路方案的因素基础上,建立方案综合评价指标体系,并对其拟定的走向方案进行全面评价。评价结果表明该方法具有较高的可信度,较传统方法更有参考性,可为线路决策者在优选方案时提供一定的借鉴。     

CO2 emissions in relation to street-network configuration and city size

The street-network efficiency of tens of British cities in relation to transport fuel consumption and CO₂ emissions are analyzed. The results show a strong linear positive correlation between length entropy and average street length, and a negative correlation between entropy and street density. Also, the results suggest that in a large city the street network is used more efficiently than in a small city, as indicated by the sublinear relations between city size (population) and the number of streets, total length of streets, and the area covered by the street network. The sublinear relation means that these parameters grow more slowly than the city size. By contrast, because a larger fraction of the street network is used at close to full capacity during considerable part of the time in a large city than a small one, the fuel consumption and the CO₂ emissions show a linear relation with city size and superlinear relation with total street length. The superlinear relation means that the CO₂ emissions increase faster than the total street length, a measure of the network size. Thus, large cities may be less energy efficient and environmentally friendly than small cities. In every city the street network needs to interconnect all the buildings, which requires a certain minimum size. In a small city, however, the network is used to a low capacity most of the time so that its relations to fuel consumption and the CO₂ emissions are more favorable than those in a large city.     

基于财务视角的地区船舶工业竞争力评价研究

从财务角度构建地区船舶工业竞争力评价指标体系,运用熵权法评价模型对江苏、辽宁、上海和浙江四个省市的船舶工业财务数据进行计算分析,得出江苏、辽宁、上海和浙江四个地区船舶工业竞争力的综合排名。排名结果基本符合上述地区竞争力的实际状况。运用财务数据来评价与反映地区船舶工业竞争力是一种新的探索,文章的研究视角与评价方法适用于地区船舶工业竞争力的综合性、整体性研判,也可作为常用评价视角与方法的补充。     

基于Entropy-FAHP综合定权法的山区公路路线方案优选研究

针对山区公路选线评价主观分析多、定量少的问题,为综合量化评价比选方案状况,建立了三维度多指标的山区公路路线评价体系,并将Entropy法与FAHP法进行有机结合,构建了山区公路路线多方案综合评价方法,采用该方法对定西市某公路K线和A线2个方案进行评价,计算各方案综合权重下的得分,得到了贴近工程实际的优选方案。     

公路站场规划布局方案综合评价的改进TOPSIS法

文章在确定了合理的评价指标体系的基础上,将熵权法和TOPSIS法相结合对公路站场规划方案进行综合评价,为选择公路站场布局方案提供依据。并通过某市站场规划布局的实例应用,验证了该评价方法的实用性。     

在役桥梁检测的灰色熵权模糊可靠性诊断方法

文章通过分析现行桥梁检测诊断方法存在的问题,结合在役桥梁诊断的指标体系和等级标准,提出利用熵权法确定诊断指标权重的思路,引入灰色多层次模糊综合诊断方法,建立了灰色熵权法的在役桥梁诊断的多层次模型,并运用可靠度进行检验.通过实例分析,证明了该方法能够更客观地反映在役桥梁的技术状况,具有较好的可操作性和实用性,对更加科学合理地评价在役桥梁诊断的结果具有现实的参考价值.     

Entropy weighted average method for the determination of a single representative path flow solution for the static user equilibrium traffic assignment problem

The formulation of the static user equilibrium traffic assignment problem (UETAP) under some simplifying assumptions has a unique solution in terms of link flows but not in terms of path flows. Large variations are possible in the path flows obtained using different UETAP solution algorithms. Many transportation planning and management applications entail the need for path flows. This raises the issue of generating a meaningful path flow solution in practice. Past studies have sought to determine a single path flow solution using the maximum entropy concept. This study proposes an alternate approach to determine a single path flow solution that represents the entropy weighted average of the UETAP path flow solution space. It has the minimum expected Euclidean distance from all other path flow solution vectors of the UETAP. The mathematical model of the proposed entropy weighted average method is derived and its solution stability is proved. The model is easy to interpret and generalizes the proportionality condition of Bar-Gera and Boyce (1999). Results of numerical experiments using networks of different sizes suggest that the path flow solutions for the UETAP using the proposed method are about identical to those obtained using the maximum entropy approach. The entropy weighted average method requires low computational effort and is easier to implement, and can therefore serve as a potential alternative to the maximum entropy approach in practice.