Particle swarm optimization (PSO) which is a bionic evolutionary algorithm is first introduced by J Kenndey and R C Eberhart in 1995. It is intelligence algorithm. Its concept is simple, easy in implementation, and has fewer parameter. As a result, it has attached great importance and has widely used in many areas since it is introduced.The early convergence speed of the elementary particle swarm optimization is particularly fast, but it has a shortcoming that the capability of search is bad and the solution precision reduces. On the other hand, for different problems, there should be different balances between the local search ability and global search ability.A inertia weightωis brought into particle swarm optimization by Shi Y. It solves the shortcoming of particle swarm optimization. At present, there are many methods to establish inertia weight. Among them, the strategy of linear decreasing inertia weight is widely applied which is proposed by Shi Y and R C Eberhart. That is called a particle swarm optimization with the strategy of linear decreasing inertia weight. But the LDWPSO algorithm also has the deficiency. Firstly, the convergence rate of the algorithm is low. The inertia weight has nothing to do with the particle. Thus, the intelligence factor of the particle search process decrease. Secondly, LDWPSO algorithm needs to forecast the maximum numbers of iterations. For different problems, the maximum numbers of iterations is different, and difficult to forecast. That affects the adjustment function of the algorithm.This article proposes a particle swarm optimization algorithm with nonlinear dynamically adjusting (NDWPSO). Which is related with in the position adaptation value, and has the capability of adjusting the size automatically according to adaptation value. with inertia weight, the intelligent factor of the particle search process is enough, the convergence rate of algorithm is enhanced, the forcast of the maximum iterations numbers can be canceled and the algorithm is stronger.The algorithms of LDWPSO and NDWPSO are tested using four well-known benchmark function, the result indicate that the accuracy of convergence of LDWPSO is better than NDWPSO and the convergence speed of NDWPSO is significantly superior to LDWPSO. |