Introducing algorithm portfolios to a class of vehicle routing and scheduling problem

The paper presents a comprehensive foundation and implementation of Algorithm Portfolios to solve Theater Distribution Vehicle Routing and Scheduling Problems (TDVRSP). In order to evaluate the performance of proposed approach, it has been applied to varying dimensions of theater distribution problem. In particular, eight random search metaheuristics embedded in four processors, packed to form different portfolios. Four basic algorithms- Genetic Algorithm (GA), Simulated Annealing (SA), Tabu Search (TS) and Artificial Immune System (AIS), as well as their group theoretic counterparts have been utilized. The proposed approach also takes care of platform dependence and helps evolving a robust solution pack. The portfolio concept is shown to be computationally advantageous and qualitatively competitive over the benchmark set of problems. The paper does not only provide modeling to TDVRSP, but also aids in developing a generic solution framework for other problems of its kind.