Searching ARMA Model Parameters Based on MLE-Based by Applying PSO Algorithm

On the condition of small samples,it is difficult to achieve ARMA model parameters by only using moment estimation (ME).In this paper,a particle swarm optimization (PSO) algorithm is used to obtain optimal numeric solutions of small sample ARMA model in the sense of maximum likelihood estimation (MIE). The principle and evaluation index of PSO are discussed in detail,which is based on analyzing likelihood function of ARMA model.Actual example shows that the joint PSO optimization method used for estimating AMAR model parameters has better characteristics in comparison with other methods.At the same time,some advantages and disadvantages are expatiated and analyzed from perspectives of algorithm and likelihood function.