摘要：School bus service is divided into many kinds of operation modes, such as single school, single load multiple schools and mixed load multiple schools and so on. When it comes to plan bus routes satisfying various constraints and different objectives, there are many applications of school bus routing problem (SBRP). It is generally recognized that middle and large sacle SBRP applications are resolved by heuristic algorithms, because SBRP is a NP-hard problem. This paper proposes a local search-based metaheuristic algorithm framework for SBRP on the basis of analysis of SBRP problem models. The framework, which is implemented by C#, is made up of basic data structures, operation functions, neighborhood operators, initial solution construction algorithm components and heuristic strategies, etc. The neighborhoodcentered metaheuristic algorithms, such as simulated annealing (SA), iterated local search (ILS), variable neighborhood search (VNS), can be designed based on the framework to solve a certain problem type of SBRP. Using this framework, we can solve three operation modes SBRP with homogeneous or heterogeneous bus fleet. Moreover, the proposed metaheuristic framework also provides the development of metaheuristic for the capacitated vehicle routing problem(CVRP) by reducing or modifying the constraints of SBRP. The experimental results of a set of instances show that the metaheuristic algorithm built based on the proposed framework, can be quickly realized and applied to different SBRP applications. Meanwhile, the designed metaheuristic framework is also very effective and extensible.
关键词：algorithm framework;school bus routing problem;local search metaheuristic;neighborhood-centered;general-purpose solver
