Uma meta-heurística Adaptive Large Neighborhood Search com mecanismos de paralelismo, detecção de estagnação e perturbações para o problema de roteamento de veículos com frota heterogênea, periódico e Multi-Trips

Abstract

The planning of vehicle routes is a major issue involved in supply chains. In real environment we can find situations involving a very large number of clients or constraints witch indicate that exact methods should be avoided. In this context, this work presents an metaheuristic for solving some variants of the vehicle routing problem (VRP): Heterogeneous VRP, VRP Periodic and VRP with multi-trips. The metaheuristic chosen, called Adaptive Large Neighborhood Search (ALNS), combines the power of successful strategies in the literature as a large neighborhood search and adaptive mechanisms with new features such as parallelism, detection of stagnation and perturbations. Our ALNS was implemented in such a way that all variants of the VRP are solved without changes in the code. The results for several instances proposed in the literature are satisfactory, showing the good performance of the approach.