Autonomously Controlled Adaptation of Formal Decision Models – Comparison of Generic Approaches

Abstract

We consider a deployment problem from transportation logistics that requires a repeated update of an existing transportation plan. The system load varies significantly and unpredictable over time. In order to consider modifications of the decision premises we propose to adjust reactively the used decision model for a transport planning problem. Since the transportation plan update instances are represented in terms of an optimization problem we try to adapt the objective function as well as the constraint set in order to inject up-to-date problem knowledge into the formal problem representation. Furthermore, we provide a comprehensive report and comparison about the numerical results observed in simulation experiments and describe the different impacts of adapting the objective function and the constraint set of the model.