Identification and quantification of influencing variables for integrated and robust planning in local rail passenger traffic

The DB Regio AG carries out its annual and in-year planning in several planning steps, usually performed successively. These include, in particular, timetabling, vehicle scheduling, crew planning, and crew rostering. These four subtasks differ significantly in their influenceability, complexity, and objective criteria. Due to the individual and partially contradictory objective criteria, usually one cannot adequately take interactions between the tasks into account during the consecutive execution of the individual planning steps. At the same time, however, the subsequent step's solution space is defined mainly by the previous planning steps. That means that planners of later planning steps are strongly limited in their decision-making possibilities by the specifications of preceding planning steps. However, these resulting interdependences have hardly been investigated so far, both in research and practice.

Comparing the DESIRED planning data with the ACTUAL disposition's data yields indicators that characterize the schedule's actual quality that sometimes deviates strongly from the planning. Within the project's scope, data-driven essential characteristics of the TARGET planning are to be identified and quantified. These characteristic indicators have a high informative value about the effects on subsequent planning steps and the robustness of the operation. For this purpose, statistical methods and machine learning are utilized. The aim is to develop a decision support system based on the indicators identified. For instance, a prototypical software in the form of a traffic-light system makes it possible to assess the effects of an individual planning step concerning the overall plan's expected robustness and quality.