ACPSO: a new cooperative particle swarm optimization algorithm with adaptive updating parameters

This paper presents a particle swarm optimization algorithm named ACPSO that adaptively updates inertia weight and acceleration coefficients of each dimension for every particle in each cycle. ACPSO simultaneously uses the advantages of adaptive structure of APSO and cooperative structure of CPSO by combining them. Adaptive structure of ACPSO allows updating velocity equation with most appropriate value of parameters in each cycle so that response will produce in less iteration. Cooperative structure of ACPSO has caused to 1) be useful to solve high dimensional problems; 2) prevents trapping in local optimum using increases the population diversity; 3) improves all dimensions of problem in each cycle unlike other methods that may be in some of the components are better and in others worse; application of the new ACPSO algorithm on several benchmark optimization problems shows a marked improvement of performance rather than other methods that have non cooperative structure with constant, time varying or adaptive parameters initializing. Compare of convergence speed shows that convergence speed of ACPSO is faster than other methods even non adaptive cooperative method.