Research on Spark-Based Cultural Ant Colony Algorithm for Capacity Vehicle Routing Problem

In addressing the Capacitated Vehicle Routing Problem (CVRP), traditional Ant Colony Optimization (ACO) algorithms often encounter challenges of slow convergence and susceptibility to local optima. To overcome these limitations, a novel approach integrating the Cultural Algorithm (CA) with ACO was developed to parallelize the algorithm using the Spark big data computing framework to enhance scalability for largescale problems. In the algorithm, the population space and belief space were designed with the population evolution in the population space adopting the ACO system and an adaptive mechanism. The algorithm enables the dynamic adjustments of transition probabilities and pheromone intensities in iterations. The belief space consisted of the best individual. The 2-opt operator was used for optimization updates to guide the evolution of the population space. Subsequently, the Spark big data platform was employed, encapsulating ants into Resilient Distributed Datasets (RDDs) for distributed computation. Experimental results validated the superiority of the proposed algorithm in terms of solution accuracy and execution speed compared to other enhanced ACO algorithms, demonstrating promising prospects for practical applications.