An agent-based memetic algorithm for solving three-level freight distribution problems

Abstract

Vehicle routing problems for the transport of people or goods are combinatorial problems belonging to the nondeterministic polynomial time (NP-Hard) category. This means that the solution is very limited if using mathematical programming or combinatorial optimization procedures. As alternative solutions, several metaheuristics have been developed, including memetic algorithms that explore possible solution spaces by implementing efficient procedures for vehicles routing in real applications. However, the combinatorial problem is still present and these solution techniques lose efficiency if the search space is enlarged or if shorter solution times are needed—both common demands in real-world applications. This chapter presents an evolutionary memetic type metaheuristic in a multi-agent system that combines global and local search strategies for solving a three-level freight distribution network. The proposed model uses coordination and collaboration strategies between several agents, to improve the performance of the freight transport process and use a specific gender alternative at the memetic algorithm that reinforces the metaheuristic evolutionary process. The results obtained by multi-agent system model application are compared with the Solomon insertion heuristic, generating better solutions that improve the distribution process in terms of less total travel distance and variability.