Modelling Geographical and Temporal Factors For Decision Support in Evolving Sensor Placement

城市化水平的提高和气候变化使城市的地表水管理成为一个主要关注的领域。在英国，有1400万条路边沟渠，这些沟渠支撑着24.5万英里的道路。当沟渠堵塞时，就会导致洪水泛滥。目前，地面排水基础设施的维修者对这个网络的运行方式以及如何维护它以保护市民免受洪水威胁的了解非常有限。这项计划将发展所需的统计基础设施，以管理我们日益增长的道路网络感应化。特别是，InTouch在英国各地的沟渠中部署了传感器，以监控英国沟渠网络的状况。由于实际考虑，这些传感器的数量有限，因此只能监视给定主机环境的一小部分。该项目将开发统计方法，以表明何时以及如何重新部署传感器将有利于知识获取。考虑到空间和时间依赖性的最佳时间和地理位置调整将增加单位时间的信息增益，相对于简单的静态网络实现的信息增益，并提供更准确的问题可预测性，以推动清洁制度和减少洪水。目前还没有可应用于这一问题的统计方法，因此本项目试图发展这些方法。

Increasing levels of urbanization and climate change have placed managing surface water in cities asan area of major concern. In the UK, there are 14M roadside gullies and these support 245,000 milesof road. When gullies become blocked this leads to flooding. At present the maintainers of thesurface water drainage infrastructure have a very limited understanding of the way in which thisnetwork operates and how it should be maintained to protect citizens from the risks of flooding.This project will develop the statistical infrastructure required to manage the increasingsensorfication of our road network. In particular, InTouch have deployed sensors in gullies acrossthe UK to monitor the state of the UK gully network. These sensors are limited in number due to costand practical considerations thus only a small portion of a given host environment may bemonitored. This project will develop statistical methods to indicate when and how redeployment ofsensors would benefit knowledge acquisition. An optimal time and geographical relocation, takinginto account spatial and temporal dependency, will increase the information gain per unit timerelative to that achieved by a simple static network and provide more accurate predictability ofproblems to drive cleaning regimes and reduce flooding. There are currently no statistical methodsthat can be applied to this problem and thus this project seeks to develop these.