A novel GM(1, 2) forecasting model with parameters identified recursively
参数由递推辨识的新型GM(1,2)预测模型

To improve the prediction performance, we propose a novel GM(1,2) model for prediction. The recursive prediction equations are derived directly from the definition of the model. The parameters of prediction equations are identified by using the particle swarm optimization algorithm (PSO). Typical numerical examples are given to demonstrate that the novel GM(1,2) model provides faster convergence rate and higher prediction precision than conventional GM(1,2) models and other improved GM(1,2) models mentioned in references.