Multi-objective design space exploration of road trains with evolutionary algorithms

Abstract

This paper examines the road train concept as a new alternative in long-distance freight traffic. The design of such a system is a difficult task since many different and conflicting criteria arise depending on the application spectrum, the legal conditions and the preferences of the carrier. Furthermore the evaluation of each decision alternative relies on a time consuming and sophisticated simulation. Evolutionary algorithms (EAs) have shown to be a useful tool for multi-objective optimization in engineering design. Based on a unified model, we develop a problem-specific evolutionary algorithm which features strong elitism, an unlimited archive of non-dominated solutions and density dependent selection. This EA is able to create alternatives which dominate previous manually engineered solutions as well as those derived from exhaustive search.