Hybrid modified ant system with sweep algorithm and path relinking for the capacitated vehicle routing problem

Vehicle routing problem is a widely researched combinatorial optimization problem. We developed a hybrid of three strategies-a modified ant system, a sweep algorithm, and a path relinking-for solving a capacitated vehicle routing optimization problem, a vehicle routing problem with a capacity constraint. A sweep algorithm was used to generate initial solutions and assign customers to vehicles, followed by a modified ant system to generate new generations of good solutions. Path relinking was used for building a better solution (candidate) from a pair of guiding and initial solutions. Finally, a local search method-swap, insert, reverse and greedy search operations-was used to prevent solutions from getting trapped in a local minimum. Performance of the proposed algorithm was evaluated on three datasets from CVRPLIB. Our proposed method was at least competitive to state-of-the-art algorithms in terms of the total route lengths. It even surpassed the best-known solution in the P-n55-k8 instance. Our findings can lower the transportation cost by reducing the travelling distance and efficiently utilizing the vehicle capacity.

Automated summary

The vehicle routing problem and its optimization techniques were studied, and a hybrid algorithm combining multiple strategies was proposed to effectively reduce transportation costs.