Search space reduction for electric vehicle-based tour schedules

Abstract

This paper develops speedup techniques for tour schedules based on electric vehicle (EV)-specific constraint processing. Rather than the tour length, the waiting time is more critical to EVs due to long charging time and immature charging infrastructure. First, grouping tour spots in the vicinity significantly cuts down the search space, which is very sensitive to the number of destinations to visit, as the subsequence within a group has little impact on the waiting time. Second, the per-spot precalculation of the minimum cost to the other destinations makes more branches be pruned. Finally, the combination of well-established techniques in the traveling salesman problem leads to a tighter constraint on intermediary node expansion. The reduced overhead provides an acceptable response time for EV drivers who want to make an energy-efficient tour plan having a tolerable waiting time bound.