Analysing Capacity Bottlenecks in Rail Infrastructure by Episode Mining

We introduce a methodology to identify and analyse capacity bottlenecks in railway networks. It is based on operational data that has been recorded in real operation. In a first step, network areas that exhibit frequent and significant train delays are determined. Next, the actual causes of such delays are investigated by analysing interdependences between train runs and by distinguishing between primary and secondary delays. This is achieved by employing episode mining techniques to enable the systematic identification of temporal patterns that occur frequently in the data about train runs.