Vehicle Routing Variants with Multiple Attributes - Metaheuristics and Timing Subproblems
Sprache des Titels:
Real-life vehicle routing applications bring forth a wide variety of problem attributes to represent customer, vehicle, driver and network specifics and needs. The resulting multi-attribute vehicle routing problems (MAVRP) have been the focus of extensive research. Yet, current methods remain limited in the number of attributes they can efficiently address.
In a first part of this talk, a new general-purpose Hybrid Genetic Search meta-heuristic for MAVRPs is introduced, which relies on efficient unified local search, genetic operators and advanced diversity management methods, and relegates problem specificities into small adaptive components. State-of-the-art results are reported on 25 main VRP variants with a single algorithm implementation and parameter setting.
In the second part, we survey and analyze the "timing? subproblems and algorithms for determining service dates to customers for any fixed delivery sequence, in presence of various time constraints and objectives. Such methods are critical to address routing settings with time attributes. In relation to frequent application cases within neighborhood searches, efficient solving of series of related timing problems is also investigated.