A Flexible Tracking PSO Algorithm for Non-Stationary Optimal Solutions

In this paper, a flexible tracking particle swarm optimization (FTPSO) for non-stationary optimal solutions is presented. The improved algorithm involves accurately detecting the changes all of the search space and reliably updating obsolete particle memories, which has been shown to be effective in locating a changing extrema. To simulate the dynamic environment, three different types of goal movement implemented by Angeline are investigated, linear, circular and random. The effectiveness of the new strategy has been examined and tested using the parabolic De Jong benchmark function.