Optimization of large-scale real-time problems in urban contexts

这个研究项目是围绕城市环境中产生的现实问题，实时处理大量数据。随着新技术的使用，实时数据采集可以做出更好、更快、更明智的决策。在本研究中，我们利用实时信息对城市物流、交通回避和水资源分配做出更好的决策；这些问题对社会和工业都很重要。关于交通运输，将人和货物从一个地方运送到另一个地方是一项复杂的任务。无数的可用模式和不同的交通成本/时间使得不可能找到“一个解决方案”。最快的方式未必是最便宜的，未必是最环保的，未必是最方便用户的。这里，优化在管理这些相互矛盾的目标方面发挥着重要作用，目标是根据一组标准确定最佳解决方案。此外，诸如紧急交付之类的实时请求可能会以不希望的和不可预测的方式影响解决方案。许多输入参数随着时间的推移而变化，这被称为时间相关问题，这可能很容易预测，也可能不容易预测。对于处理实时数据的更主动的操作，做出实时决策非常重要。在动态的城市交通场景中，信息是不断变化的，能够及时做出反应以便在实践中解决问题是非常重要的。当卡车已经在执行取货和/或交付活动时，就会出现新的请求。这些新信息必须快速整合，以便提供及时的解决方案，这些解决方案不影响公司的底线，满足服务水平要求，并尊重所有操作约束。其中一些问题已经出现在文献中，但使用实时信息和考虑随机数据的频率要低得多。为了允许行业使用我们的解决方案，我们计划合并这些方面，不假设所有的未来信息都是已知的，而只是其中的一部分，而一些信息是随着时间的推移而实现的。饮用水是城市向市民提供的主要服务之一。一个巨大的管道网络紧密相连，水流受到不同的压力和力。与车辆在街道上行驶或货物在生产线上移动不同，水流受到高度非线性的流动条件的影响，会造成压力和能量损失。利用我们在运输方面的知识来优化饮用水分配，我们创建了能够提供有关质量问题（如氯残留量、水龄、流向等）的操作计划的算法。从操作的角度来看，挑战在于有效地使用泵来节省能源，并满足所有需求，保持压力在一定范围内。

This research program is centered around real-life problems arising in urban contexts dealing with huge amounts of data in real-time. With the use of new technologies, real-time data acquisition allows for better, faster and more informed decisions to be made. In this research, we use real-time information to make better decisions for city logistics, traffic avoidance, and water distribution; these problems are important for society and for industry. Regarding transportation, moving people and goods from one point to another is a complicated task. The myriad of modes available and different costs/times due to traffic makes it impossible to find a "one solution fits all". The fastest way may not be the cheapest, which may not be the most environmental-friendly, which may not be the most convenient for the user. Here, optimization plays an important role in managing these contradicting objectives with the goal of determining the best solution according to a set of criteria. Moreover, real-time requests such as an urgent delivery may affect the solution in undesirable and unpredictable ways. Many input parameters change over time, in what is known as time-dependent issues, which may or may not be easily predicted. For more proactive operations dealing with real-time data, it is important to make real-time decisions. In dynamic urban traffic scenarios, information constantly changes, and it is important to be able to react in time allowing to solve problems in practice. This is the case when a truck is already performing its pickup and/or delivery activities, and new requests arise. These new information must be quickly incorporated in order to provide timely solutions that do not affect the bottom line of the companies, that satisfy service level requirements, and that respect all operational constraints. Some of these problems are already emerging in the literature, but the use of real-time information and considering stochastic data is much less frequent. In order to allow industry to use our solutions, we plan to incorporate these aspects, by not assuming all future information is known, but only parts of it, while some information is realized over time. Drinking water is one of the main services that cities provide to its citizens. A huge network of pipes is strongly connected, and the flow of water is subject to different pressures and forces. Unlike vehicles traveling through streets or goods moving through a production line, water flowing is subject to highly non-linear flow conditions, causing pressure and energy loss. Using our knowledge from transportation to optimize drinking water distribution we create algorithms capable of providing operational plans respecting quality issues (such as chlorine residuals, water age, flow direction, etc).  From an operational perspective, the challenge is to efficiently use pumps to save energy and to satisfy all demand keeping pressure within bounds.