Brain Storm Optimization Algorithm based on adaptive inertial selection strategy for the RCPSP

Abstract

RCPSP(Resource-constrained Project Scheduling Problem) that has been proved to be a NP-hard problem is a kind of combination optimization problem. With the problem scale becoming larger and larger, the problem will become more difficult to solve. A discussion mechanism based on brain storm optimization (DMBSO) for the RCPSP was presented. According to the characteristics of the RCPSP and the problem of poor accuracy and slow convergence in later period of algorithm. We proposed an adaptive inertial selection strategy based on DMBSO (AD-DMBSO) to solve the RCPSP. The algorithm can control the probability of inter-group and intra-group discussion by adjusting the probability of population grouping strategy, which can strengthen the global search in the early stage, and enhance the local search in the later stage. We test AD-DMBSO algorithm on four sets of problem from PSPLIB. The experimental result showed that the proposed algorithm is superior to the DMBSO and particle swarm optimization (PSO) algorithm in solving the local optimum problem and converging to better optimal values. The excellent performance of the AD-DMBSO algorithm shows its great potential in solving the RCPSP.