Improving solution characteristics of particle swarm optimization using digital pheromones

摘要: In this paper, a new approach to particle swarm optimization (PSO) using digital pheromones coordinate swarms within an n-dimensional design space is presented. basic PSO, initial randomly generated population propagates toward the global optimum over series of iterations. The direction movement in based on individual particle’s best position its history trail (pBest), and entire (gBest). This information used generate velocity vector indicating search promising location space. premise research presented paper fact that for each member dictated by only two candidates—pBest gBest, which are not efficient locate optimum, particularly multi-modal problems. addition, poor move sets specified pBest stages can trap local minimum or cause slow convergence. presents use aiding communication improve efficiency reliability, resulting improved solution quality, accuracy, efficiency. With empirical proximity analysis, pheromone strength region determined. then reacts accordingly probability may contain optimum. additional from causes particles explore thoroughly more efficiently accurately than PSO. development method results several test cases.