Fuzzy Resource Constrained Project Scheduling Optimization with Hybrid Multiobjective Genetic Algorithm

Abstract

Fuzzy resource constrained project scheduling problem (FRCPSP) is an extended problem of RCPSP considering uncertainty. It is a very important research issue, as a NP-hard combinatorial optimization problem and actual application of project scheduling. This paper proposes a hybrid genetic algorithm that combines a non-random initialization, a neighborhood search-based mutation, and two local search strategies. Fuzzy RCPSP uses fuzzy set method to describe uncertainty. It assumes that the activities with random duration changed in an interval, which is composed of optimistic time, pessimistic time and possible time. This paper innovatively converts the interval into 3 optimization objectives, reformulates FRCPSP into a multiobjective optimization model, and designs a hybrid multiobjective genetic algorithm based on NSGA-II for solving this FRCPSP. Finally, benchmarks of RCPSP and extended datasets with fuzzy processing time are adopted to test our approach. Computational results show that our approach performs better than the existing state-of-the-art methods.