Computer Science · Computer Vision and Pattern Recognition
An Improved Inertia Weight Mutation Particle Swarm Optimization
See on the mapAbstract
Proposes an improved inertia weight mutation particle swarm optimization to solve the premature convergence problem,and to avoid the slowconvergence in the later convergence phase.When running the algorithm,different inertia weight values are given to particles according to their fitness.Thus the algorithm is engaged with both good exploration ability and good exploitation ability.When the optimum information of the swarm is stagnant,mutation operator is introduced to change the location and speed of the particles which are close to the local optimum position,and thus to reduce the possibility of trapping at the local optimum.According to the experimental results using four typical functions,the global searching ability and the speed of convergence of the new algorithm are both improved,and the premature convergence problem is effectively avoided.