The Research Of Particle Swarm Algorithm Based On Net Optimization

Particle swarm optimization algorithm is a new kind of intelligent optimization algorithm proposed in1995, and it has the characteristics of iterative evolution and swarm intelligence at the same time. The optimization principle of the particle swarm algorithm is different from that of genetic algorithm which is dependent on the individual evolution, but it is an intelligent optimization algorithm optimizing through simulating the social behavior of the biological group.The disadvantage of the particle swarm algorithm is that it has poor adaptability for high-dimension function, poor ability to jump out of the local optimization, unstable optimization for complicated function. And aiming at these shortcomings, this paper puts forward a brand new particle swarm algorithm based on net optimization. The new algorithm integrates the thought of net into the basic particle swarm algorithm, divides the definition domain of the ready-for-optimized function into several subintervals, finds out the subinterval of the optimal solution through comparing the fitness value, and at last reduces the definition domain according to the position of this subinterval. The updating formula of the particle swarm algorithm is also improved. Experimental results show that the particle swarm algorithm based on net optimization has strong adaptability for high-dimension function, improved ability of global optimization and better optimization speed.This paper also does some improvement on linear decreasing inertia weight particle swarm algorithm and particle swarm algorithm with contraction factor, which are two most used algorithm, and the improvement effect is satisfactory.