A filter-and-fan approach with adaptive neighborhood switching for resource-constrained project scheduling

This paper presents an effective heuristic algorithm based on the framework of the filter-and-fan (F&F) procedure for solving the resource-constrained project scheduling problem (RCPSP). The proposed solution methodology, called the filter-and-fan approach with adaptive neighborhood switching (FFANS), operates on four different neighborhood structures and incorporates improved local search, F&F search with multiple neighborhoods and an adaptive neighborhood switching procedure. The improved local search, in which a new insert-based move strategy and new time compression measurement of schedules having the same makespan are embedded, is utilized to identify a local optimum and a basic move list. The F&F search, aimed to further improve the local optimum, applies multi-neighborhood filter and fan strategies to generate compound moves and a neighborhood-switch list in a tree search fashion. When the current neighborhood cannot further improve the local optimum, the adaptive neighborhood switching procedure picks the most potential neighborhood for the next run of the local search procedure. The entire solution procedure is autonomous and adaptive due to its variable search range depending on the project sizes and characteristics. Computational results and comparisons with some state-of-the-art algorithms indicate the effectiveness and competence of the proposed FFANS. We design a new competitive local search-oriented method for solving the resource-constrained project scheduling problem.The filter and fan candidate list strategies with multiple neighborhoods are proposed in the method.The method is able to adaptively search among four different neighborhood structures according to the current search state.Adaptivity and robustness are verified and analyzed.