Two improved method are proposed for the difficulty and defects of the traditional GM(1,2) model. Firstly, the
correlation matching algorithm is adopted to search the reference sequence which has the maximum grey relation grade in the
history database. Secondly, the particle swarm optimization algorithm is used to identify the model parameters by taking the
evaluation function of the forecasting performance as the objective function. The simulation results show the effectiveness
and the applicability of the presented method.