The Optimization of the Non-equigap DGM (2, 1) Model

Abstract

Based on the principle of GM (1, 1) model, firstly, this article advances the basic form of non-equigap DGM (2, 1) model. Secondly, on the assumption of getting non-equigap series' 1-AGO series by accumulating, let the prediction series obey the form of nonhomogeneous exponent, this article optimizes the grey derivative and background value of non-equigap DGM (2, 1) model by calculating the definite integral of the whitened differential equation, and then, establishes a new non-equigap DGM (2, 1) model. The new model breaks through the limitations of the non-equigap series' prediction, which only obeys homogeneous exponential law, and it improves the fitting precision and prediction precision. Furthermore, it has enlarged the application of GM (1, 1).