Swarm intelligence(SI) is a kind of method to realize artificial intelligence by simulating biological colony behaviors in nature.It emphasizes the simplicity of individual behavior

group stigmergy and down-up research strategy.The advantages of swarm intelligence are analyzed.According to social information share

the group migration strategy is designed and refreshing rule of information density is presented

which realizes groups' cooperation

avoids close competition and accelerates convergent velocity.The strategy achieves the best balance between better exploiting and global exploring

enhances the searching velocity and attains global convergence.Optimum experiments by four testing functions show that the improved MEA is feasible and efficient.The improved MEA are used to on-line adjust PID parameters and better control performances are obtained.