Inverse approach to find an optimum α for grey prediction model

An optimization approach combined with a grey model to improve the modeling error of grey prediction is presented in the paper. In this study, each background value α at a discrete point is considered as an independent parameter. By an optimization approach, an objective function of the square modeling error summed at a discrete point is minimized to find the optimal solution of parameters. Then, the set of these optimal α can be used as parameters to build up another GM(1,1) to predict the next optimal α. Based on the procedure expressed above, the optimal GM(1,1) model is then developed based on the use of an optimal set of background values instead of the use of an average background value