Promoting Diversity in Particle Swarm Optimization to Solve Multimodal Problems

Promoting diversity is an effective way to prevent premature converge in solving multimodal problems using Particle Swarm Optimization (PSO). Based on the idea of increasing possibility of particles "jump out" of local optima, while keeping the ability of algorithm finding "good enough" solution, two methods are utilized to promote PSO's diversity in this paper. PSO population diversity measurements, which include position diversity, velocity diversity and cognitive diversity on standard PSO and PSO with diversity promotion, are discussed and compared. Through this measurement, useful information of search in exploration or exploitation state can be obtained.