Functional form & transport infrastructure, supply and demand

在交通规划中，由于功能形式、交通基础设施和需求，出现了许多需要使用模型的问题，因为模型在解释和预测结果方面比我们的直觉做得更好。为了建立这些结果与它们的决定因素之间的联系，我们假设了关系，并测试了联系的存在、方向和大小。该项目涉及运输系统内这些统计联系技术的三个方面。理论上。我们关注的是统计关系不能独立于定义它们的数学形式而建立的方式。我们致力于形式和联系，因为它们必须共同研究和建立。工具开发。在应用工程的许多领域，样本的大小都在增加。这种信息漂移或复杂因素需要对为更有限的数据库设计的计算机代码进行调整。这种对工具的调整可以使大样本具有日常相关性。申请。这些方法，无论是理论上的，还是基于工具的，都适用于运输系统的组成部分。在这里，我们将重点关注需求、供应和事故性能组件，就像我们之前所做的那样。但我们将应用范围扩展到基础设施耐久性问题：(I)因不同重量的车轴(例如火车和卡车)对基础设施造成的损坏；(Ii)或暴露在侵蚀性化学环境中的混凝土退化。在后一种情况下，概率方法的使用相对较新，对建立多元统计因果关系及其形式提出了新的挑战。

FUNCTIONAL FORM AND TRANSPORT INFRASTRUCTURE AND DEMAND In transport planning, a number of questions arise that require the use of models, because models do better than our intuition at explaining and forecasting outcomes.To establish links between such outcomes and their determinants, relationships are postulated and the existence, direction and size of linkages is tested. This project deals with 3 aspects of these statistical linkage techniques within transport systems. THEORY. We focus on the way in which statistical relationships cannot be established independently from the mathematical form defining them. We work on form and linkage because they have to be studied and established jointly. TOOL DEVELOPMENT. In many areas of applied engineering, the size of samples is increasing. This informational drift, or sophistication factor, requires the adaptation of computer codes designed for more limited databases. Such adaptation of tools can make large samples of daily relevance. APPLICATION. These approaches, both theoretical and tool-based, are applicable to the components of transport system. Here we focus on demand, supply, and accident performance components, as we have done previously. But we extend applications to infrastructure durability matters arising: (i) from the damages caused by vehicles axles (e.g., trains and trucks) of varying weights to infrastructure; (ii) or from the degradation of concrete exposed to aggressive chemical environments. In the latter case, use of probabilistic methods is relatively recent and presents new challenges to the establishment of mutivariate statistical causality and of its form.