A cooperative population learning algorithm for vehicle routing problem with time windows

Population-based hybrid metaheuristics, often inspired by biological or social phenomena, belong to a widely used groups of methods suitable for solving complex hard optimization problems. Their effectiveness has been confirmed for providing good quality solutions to many real-life instances of different problems. Recently, an incorporation of the cooperative problem solving paradigm into metaheuristics has become an interesting extension of the population-based hybrid metaheuristics. Cooperation is meant as a problem-solving strategy, consisting of a search performed by different search agents running in parallel. During the search, the agents cooperate by exchanging information about states, solutions or other search space characteristics. This paper proposes an Agent-Based Cooperative Population Learning Algorithm for the Vehicle Routing Problem with Time Windows. In the proposed approach the process of search for the best solution is divided into stages, and different search procedures are used at each stage. These procedures use a set of various heuristics (represented by software agents) which run under the cooperation scheme defined separately for each stage. Computational experiment, which has been carried out, confirmed the effectiveness of the proposed approach.