Um algoritmo genético aplicado no problema da roteirização periódica de veículos com caso prático

Abstract

The service level of a wholesale distributor can be measured by the frequency and regularity with which its sales staff serves customers. But as the market success also depends on the costs involved, the proper planning of systematic servings is critical. Taking advantage of the similarities between this situation and the Periodic Vehicle Routing Problem (PVRP), a technique for solving the later was proposed. For the PVRP, given a planning horizon of several days, visits to customers must be assigned to possible days according to predefined schedule combinations at the same time as routes are generated for each day, aiming to reduce the total cost of serving in the same planning horizon. The proposed technique was also adapted to be applied to the practical case of routing a sales team within a weekly planning horizon and it consists of a genetic algorithm for which was developed an original crossover operator. The technique was validated with instances from the literature for the PVRP and its solutions for the case study indicated significant annual savings.