New methodologies for assessing the behaviour of vehicles under real driving emissions testing regimes

个人交通工具是当地污染的主要来源，减少这种污染至关重要。长期以来，旨在限制实验室实验中车辆排放的立法未能解决车辆在道路上运行时排放的问题。这导致了最近引入的真实驾驶排放，其中包括使用移动排放测量设备在公共道路上进行的测试。首先，这项立法同时扩大了推进系统需要符合的运行条件范围（高度、温度、燃料质量……）。其次，立法取消了对测试条件的详细规定，这意味着道路测试将是随机的，不可重复的。这两个方面都对目前汽车行业的开发和签署过程造成了重大干扰，特别是意味着车辆是否完全符合新法规并因此达到要求的排放水平仍然存在很大的不确定性。该项目旨在创建一个新的流程，利用虚拟方法来增加车辆开发的稳健性，并在实际驾驶排放方面签署协议。虚拟方法将在很大程度上依赖于建模和大规模模拟，以确保涵盖广泛的操作范围。这种虚拟方法的性能将以过程中使用的数学模型的准确性为基础。汽车尾气排放的形成是一个复杂的过程，既涉及燃烧过程，也涉及处理后的过程，这意味着目前的技术水平严重依赖于经验方法。预计这里还需要经验方法，首先描述排放模型的特征，然后验证虚拟方法的结果。因此，本博士需要解决的研究挑战是：1。以数学和结构化的方式表达道路驾驶的广泛性和看似随机性，并以此来创建涵盖实际驾驶排放法规所涵盖的竞争操作范围的合成驾驶循环。这将涉及真实驾驶情况的数据收集和文献综述，这些数据的统计分析以及使用马尔可夫链等技术生成合成循环的新算法的创建。这还需要通过工程分析进行消毒，以确保纯数学生成的循环在物理上是一致的。本研究的结果将有助于更好地理解什么是真正的驾驶，并对不同的驾驶条件进行量化和分类。车辆排放模型的特征，为进行合成循环的排放模拟提供了实验努力和预测准确性的适当平衡。这将需要调查不同的建模方法（物理的、半物理的和经验的），以确定模型最合适的数学格式。这将由实验工作来补充，以表征和验证模型。本博士的实验工作本身并不简单，需要在难以测量的条件下测量大量的车辆特性（排放、道路拓扑、车辆状态……）。因此，这项工作的一个重要部分涉及到这些数据的捕获和分析其稳健性。这将涉及对测量原理的物理分析，并提出捕获或处理原始数据到所需数量的新方法。结果将是产生新的方法，以提高开发和认证新车的稳健性和速度。最终会使更便宜、更清洁的车辆早日投入市场，从而为当地的空气质量带来相关的好处

Personal transportation is a major source of local pollution and it is vital that this is reduced. Over time, legislation aimed at imposing limits on vehicle emissions in laboratory experiments has failed to address the problem of emissions from vehicle when operating on the road. This has led to the recent introduction of Real Driving Emissions that include tests conducted on public roads using mobile emissions measurement devices. Firstly, this legislation simultaneously broadens the range of operating conditions over which the propulsion system needs to be compliant (altitude, temperature, fuel qualities...). Secondly, the legislation removes the detailed stipulation of the testing conditions, meaning the on-road tests will be stochastic and not repeatable. Both these aspects create significant disruption to the development and sign-off processes currently in place in the automotive industry and in particular mean that there remains significant uncertainty as to whether vehicles will fully comply with the new legislation and therefore deliver on the required levels of emissions. This project aims to create a new process making use of virtual methods to increase the robustness of vehicle development and sign-off with regards to Real Driving Emissions. The virtual approach will rely heavily on modelling and mass simulation in order to ensure the broad operating ranges are covered. The performance of such a virtual approach will be underpinned by the accuracy of the mathematical models used within the process. Vehicle emissions formation is a complex process involving both combustion but also after treatment, meaning that current state of the art relies heavily on empirical approaches. It is expected that empirical approaches will also be needed here firstly to characterise the emissions models and subsequently to validate the findings form the virtual approach. The research challenges to be addressed by this PhD are therefore:1. Expressing in a mathematical and structured way the broadness and seemingly randomness of on-road driving and using this to create synthetic driving cycles that cover the compete operating range covered by Real Driving Emissions legislation. This will involve data collection and literature review of real driving situations, statistical analysis of this data and the creation of new algorithms to generate synthetic cycles using techniques such as Markov chains. This will also need to be sanitised through an engineering analysis to ensure the cycles generated purely from mathematical are physically coherent. The output from this research will be a better understanding of what real driving is and a quantification and classification of different driving conditions.2. The characterisation of vehicle emissions models that present an appropriate balance of experimental effort and predictive accuracy for undertaking the simulations of emissions over the synthetic cycles. This will require investigation into different modelling approaches (physical, semi physical and empirical) to determine the most appropriate mathematical format for the models. This will be complemented by experimental work to characterise and validate the models.3. The experimental work in this PhD itself is not trivial, requiring the measurement of numerous vehicle characteristic in difficult conditions for measurement (emissions, road topology, vehicle states...). Therefore, a significant part of this work relates to the capture of this data and the analysis of its robustness. This will involve a physical analysis of the measurement principles and the proposal of new methods for capturing or processing raw data into the required quantities.Outcomes will be the generation of new methodologies to improve the robustness and speed for developing and certifying new vehicles. Will ultimately lead to less expensive, cleaner vehicles being brought to market sooner which will bring associated benefits in local air quality as these veh