Grey Discrete time-varying Model and Its application

Abstract

Although the Grey Discrete Model and its improved models have been successfully employed in some fields and have promising results, the prediction results may be inaccurate sometime. We bring a grey discrete parameters model by introducing quadratic time-varying terms, which is called as quadratic time-varying parameters discrete grey model (referred to as QDGM (1, 1)). The paper investigates the properties of the new model, and concludes that QDGM (1, 1) possesses white exponential law coincidence, linear law coincidence, quadratic law coincidence. Then, we optimize the iterative starting value of the new model. Finally, the new model is compared with three discrete grey models using the same data instances. It is proved that the new model greatly improves the simulation and prediction precision.