Abstract:

Aiming to overcome the disadvantage for the common GM (1,1) model of badly forecasting results to those nonsmooth variation sequences, a new GM (1,1) model with parameters identification method for the intension expression is proposed. The intension expression, describing the nonlinear relations between developing coefficient, the grey input, the background weight parameter, and the boundaryvalue and forecasting value, are deduced. Then the particle swarm optimization (PSO) algorithm is adopted to identify internal parameters of the intension expression, thus the PSOGM (1,1) model is founded. The typical numerical examples demonstrate that the PSOGM (1,1) model can provide fast convergence rate, and has betterpredicted precision than common GM (1,1) model. Moreover, the proposed model is comfortable not only for smooth variation sequences, but also for nonsmooth variation sequences.