Spatio-Temporal Models For Sustainability
Nico Piatkowski, Sangkyun Lee, Katharina J. Morik · 2012
Many applications that aim at enhancing sustainability rely on some sort of spatio-temporal model. The task can be monitoring or prediction in traffic networks, power grids, building energy management, river flow volume, and sea level – to mention just a few. The positive effect on the environment is achieved by a better control, better planning of processes or better disaster management. Spatio-temporal models predict some states of variables over time that are spatially ordered in some topology. They can be used in a variety of applications, ranging from energy-saving technological administration to monitoring for early alarms or to supporting better emergency plans. Graphical Models have been successfully used for predictions based on structured data. Here, we introduce a functional form for the model parameters that allows a spatio-temporal predictive analysis with continuous time. Though intended also to be used in engineering approaches, there, we are missing public data sets. Hence, the use of our model is demonstrated on two exemplary applications, namely the prediction of sea-levels for tsunami prediction and the prediction of traffic jam as a consequence of a flow of refugees in case of a nuclear accident. Categories and Subject Descriptors V.2 [Transportation]; H.3 [Mining massive throughput