Optimizing Vehicle Routing in Logistics through Hybrid Genetic Algorithms: A Multi-Population Approach

Abstract

As the logistics industry advances towards globalization and informatization, vehicle path planning becomes increasingly crucial. Transportation, constituting about 50% of logistics costs, is essential for cost reduction. Efficient vehicle route planning impacts transportation time and costs in scenarios like bus routes, logistics distribution, and garbage collection.Genetic algorithms (GAs), known for their parallel and adaptive search capabilities, often fail to consistently reach optimal solutions. To address this, a multi-population genetic algorithm is proposed, consisting of a main population and auxiliary populations. A local optimization algorithm was designed to enhance convergence speed. To improve crossover effectiveness, the algorithm increases high-quality crossovers, enhancing the chances of better solutions. An acceptance rate ensures population quality by accepting new solutions only if they surpass their parent solutions, otherwise rejecting them with a certain probability.Experimental results show that the improved algorithm offers superior computational efficiency and performance in vehicle path planning.