An Attention Ant Colony Algorithm for Solving SDVRP

The Split Delivery Vehicle Routing Problem (SDVRP) is a hot topic in the field of logistics research. Recently, popular solution methods include heuristic algorithms and deep reinforcement learning algorithms. Heuristic algorithms can achieve high solution quality but require longer solution times, while deep reinforcement learning algorithms offer faster solution times but may compromise solution quality. To better balance solution quality and computation time, an Attention Ant Colony Optimization (AACO) algorithm is proposed, which combines heuristic algorithm with deep reinforcement learning algorithm. By utilizing the probability values output by the attention model as heuristic information for the ant colony algorithm, the guidance capability of heuristic information on ants is enhanced, thereby improving solution speed. The effectiveness of AACO is verified through testing on different datasets.