A modified particle swarm optimization algorithm with applications

摘要

In this paper, firstly a modified particle swarm optimization algorithm (MPSO) is developed, in which the mean value of past optimal positions for each particle and the mutation operation are considered for avoiding premature. In the optimization test, MPSO performs better than particle swarm optimization algorithm (PSO). Then MPSO is applied to solve four portfolio optimization models with the real data from the Hong Kong Stock Market, and optimal values are obtained when the number of swarm n=80,160, respectively. Finally, actual return rates of these models are calculated in numerical experiments, and it is illustrated from these graphs of actual return rates that when considering higher return, Cai’s model performs better in short-term investment.