CALIBRATION OF BONDING MODEL PARAMETERS FOR COATED FERTILIZERS BASED ON PSO-BP NEURAL NETWORK

Abstract

In this paper, the ultimate crushing displacement and ultimate crushing load of the coated fertilizer granules were obtained by uniaxial compression test as 0.450 mm and 58.668 N, respectively. Then the DEM model of the encapsulated fertilizer was established, and the Plackett-Burman and Steepest ascent tests were taken to determine the factors that had significant effects on the results and their ranges of values, respectively. Finally, the PSO_BP neural network was trained using full-factor test data, and the correlation coefficients of training process, validation process, testing process and overall performance were obtained as 0.98057, 0.95781, 0.96724 and 0.97459, respectively, indicating that the trained PSO_BP neural network fits well and can predict the ultimate crushing displacement and ultimate crushing load. The ultimate crushing displacement Y1 and ultimate crushing load Y2 are 0.450 mm and 58.703 N, with a minimum relative error of 0.06% from the actual value. This study can provide new methods and ideas for the calibration of discrete element simulation parameters.