Study on the High-Dimension Non-linear Index Model GWS(n,m) for Soil Erosion Simulation and Its Parameters Identification Method

Aiming at the problems of great demand of samples, applicability and simulation precision of the routine empirical statistical soil erosion models, based on the reasonable extension of the high-order grey dispersed array dynamic model GWS(n,1) a multivariable high-dimension non-linear index model GWS(n,m)was established. Through the optimization by MATLAB, the searching and identification of the parameters in the non-linear model was come true, and the actual problems of limited sample, non-linear and high-dimension etc. were solved by the model. In the case of artificial rainfall. The simulation results of soil erosion in sloping farmland indicated that the valid coefficient of the model E_f > 0.91, the average error rate delta < 6.2%, the stimulation precision was high.