Abstract: Research into setting the values of the acceleration coefficients C₁ and C₂ in particle swarm optimization (PSO) is one of the most significant areas for performance improvement.On the basis of the investigation of C₁ and C₂, this paper present the parameters investigation conclusion and proposed a novel PSO which C₁ and C₂ are generated as a couple of random numbers within a certain range in each iteration process.The experimental results show that the new method performs better than the methods with fixed values of C₁ and C₂ for most benchmark problems tested.