Optimization of PSO Algorithm Based on Adaptive Inertia Weight and Escape Strategy

Abstract

For Particle Swarm Algorithm uses fixed inertia weights, weaken too slow and extremely easy to fall into local optimum so I propose an adaptive inertia weight method, this inertia weight has a larger weight in the early stage, can increase the global search capability and avoid algorithm falling into local optimum, in late period, this inertia weight has a smaller weight and can increase algorithm local search ability and overall increase convergence rate. But during the optimization process, There are some more complicated functions that will still fall into the local minimum even using this adaptive inertia weight. In order to avoid the occurrence of premature phenomenon so use reverse escape strategy, apply new optimization methods to the basic particle swarm algorithm, the application of several basic functions proves that the proposed optimization method has higher precision and convergence speed.