Particle Swarm Optimization Enhancement by Applying Global Ratio Based Communication Topology

Multi-mode Resource-Constrained Project Scheduling Problems (MRCPSPs) have been proven as NP-hard problems. Among the methods for obtaining solutions to NP-hard problems, the particle swarm optimization is an efficient one. MRCPSPs include two sub-problems, namely activity operating priority and activity operating mode, two particle swarm optimization are adopted for solving these two sub-problems. A conventional PSO algorithm is applied to find activity operating priority while a discrete one is used to obtain activity operating mode. Meanwhile, a group communication mechanism based on global ratio (GR) is designed to determine dynamically whether the global topology or local topology should be used to stabilize convergence. In this study, PSO with GR design is named GRPSO. Moreover, a dual S-decreasing curve is also proposed to dynamically control the inertia weight, constriction factor and GR they are associated with the search area to ensure that the global exploration in the early stage of searches and the local exploitation in the later stage of searches are able to search the optimal solution effectively while maintain the exploration capacity at a certain level during the middle stage of searches to boost efficiency in finding the optimal solution. The results of the experiments prove the method proposed in this work can effectively find optimal solutions to MRCPSP problems.