Information Exchange And Processing Of Particle Swarm Algorithm

As a population-based intelligent optimization algorithm, particle swarm optimization(PSO) simulates the animal social behaviors according to the information sharing among particles. However, due to the limited group historical experience, the performance of standard particle swarm optimization (SPSO) is not always well. Therefore, inspired by the particle swarm optimization with passive congregation (PSOPC), this paper provides several variants by enhancing the information interaction and processing capabilities.As a novel variant, PSOPC improves the performance by adding a new stochastic information source particle. However, the stability is not very well due to this additional randomness. Therefore, to improve the PSOPC, this paper proposes nearest neighbor interaction particle swarm optimization (NNIPSO), taking the best current position among nearest neighbors as information interaction factor in neighborhood (including ring and small world model topology), which provides a frequently exchange with its nearest neighbors in order to get information faster and more accurately.The experimental results indicate that the above version is more efficiency than three other famous modifications, especially on high dimensional multimodal functions.Nearest neighbor information interaction results in the sharing of information in neighborhood, moreover, individuals may benefits the current findings and previous experience of other group members. Suggested by the above, this paper replaces the individual experience by the nearest neighbor interaction factor in NNIPSO, and proposes the neighborhood sharing particle swarm optimization (NSPSO) to improve the performance of NNIPSO. Simulation results verify the efficiency of NSPSO.PSO is a simulation of group foraging process, in which animals tend to get more food energy at less cost to maximize the foraging energy efficiency. Many modifications of PSO have been proposed so far, however, they didn't apply the above optimal foraging rule. Therefore, this paper defines foraging energy efficiency as the ratio between fitness difference and distance of two individuals and puts forward the optimal foraging optimization (OFPSO), in which each particle trends to its largest energy efficiency location among neighbors at the same iteration. The simulation result shows the strategy improves the performance of PSO significantly.