부산대학교 도서관

The Multi-Depot Vehicle Routing Problem (MD-VRP) is a known routing problem whose goal is to minimize the total cost while visiting a set of routes from their respective depots. We consider a particular application of the MDVRP, called the Electricity Technician Dispatch Problem (ETDP), in which we look for optimal tours for technicians to provide services to their customers. To overcome the inherent exponential time cost of this NP-hard problem, we propose a hybrid method structured in two stages. In the first phase, we rely on a clustering algorithm to identify the appropriate set of customers for each technician based on availability, demand, and proximity of the customer's location to each technician's depot. We then consider several approximate and exact methods to get the optimal route for each set of customers. To evaluate the performance of our hybrid solving method, we conducted experiments on a set of problem instances. The results show the effectiveness of our method when Stochastic Local Search (SLS) is used as a search technique.