Novel Particle Swarm Optimization for unconstrained problems

Estimation of Distribution Algorithm (EDA) is a class of evolutionary algorithms which construct the probabilistic model of the search space and generate new solutions according to the probabilistic model. Particle Swarm Optimization (PSO) is an algorithm that simulates the behavior of birds flocks and has good local search ability. This paper proposes a combination (EDAPSO) of EDA with PSO for the global optimization problems. The EDAPSO proposed in this paper combines the exploration of EDA with the exploitation of PSO. EDAPSO can perform a global search over the entire search space with faster convergence speed. EDAPSO has two main steps. First, the algorithm generates new solutions according to the probabilistic model. Then, EDAPSO updates the whole population according to improved velocity updating equation. EDAPSO has been evaluated on a series of benchmark functions. The results of experiments show that EDAPSO can produce a significant improvement in terms of convergence speed, solution accuracy and reliability.