- Friday 18 May 2018
Lecture Theatre 2
- Dr Daniel Guimarans; Senior Researcher/Lecturer; Aviation Academy, Amsterdam University of Applied Science
Tea and Coffee available from 13:45.
Hybrid approaches combining different optimisation methodologies have proved useful to successfully tackle combinatorial problems, including many examples in logistics and transportation. Among these techniques, Large Neighbourhood Search (LNS) has shown to be particularly effective for solving some combinatorial problems, especially when paired with Constraint Programming (CP) models. CP is a powerful paradigm that permits logically formulating the system's constraints, and using tailored methods to speed up the search. This LNS and CP combination has been able to provide state-of-the-art results for many problems in the transportation literature. However, despite their effectiveness, such traditional approaches fail to capture the natural uncertainty many of these problems present in real-life applications. Simulation models, on the other hand, are able to reproduce the dynamic behaviour of the system, but generally lack a systematic way of exploring the solution space. Simheuristic approaches are designed to overcome these issues, using simulation not only to evaluate the behaviour of solutions, but also to guide the search towards more promising configurations in stochastic scenarios. In this talk, we will dive into some characteristics of these hybrid simheuristic methodologies, as well as reviewing a range of applications in different road and air transportation problems.