A New Method for Improving Prediction Precision of GM(1,1) Model

摘要

An abnormal GM(1,1) by deforming raw one combining approach of optimization of background value and optimization of the initial item in time response function is proposed in this paper. Through the optimization of background value the improved model can simulate and predict sequences with purely exponential distribution perfectly. Optimization of the initial item in time response function may express the principle of new information priority fully and reserve the initial item in original GM(1,1) model. The numerical example indicates that the model improves the precision of simulation and prediction remarkably.