Convergence Analysis in Swarm Intelligence for City Tour Optimization

Abstract

Particle swarm optimization (PSO) algorithm has been widely used to solve many problems. However, PSO has limitation in dealing with premature convergence when each particle unable to move to find the global optimum solution. This research has investigated the various conditions for the PSO to determine when a premature convergence happened. We used city parks in Kediri City, Indonesia as an object for a city tour optimization. Furthermore, PSO by adding mutation operator belongs to Genetic Algorithm and dividing the swarm group into sub-swarm are used to investigate the convergence condition because they have been proven can successfully avoid a premature convergence. The result shows that the solutions produced by the addition of these operators can find better solutions compared to the simple PSO.