A hybrid PSO-BP-PSO method for DEM parameter calibration of moist granular materials

To explore an efficient method to calibrate the discrete element model (DEM) parameters of piglet feed, a typical moist granular material, an experimental study under different water-to-feed ratios was conducted using the stacking angle as the evaluation index. The relation between the stacking angle and the water-to-feed ratio was obtained by the physical tests. With the Hertz-Mindlin with Johnson-Kendall-Roberts (JKR), the DEM of piglet feed was established. The Plackett-Burman and steepest ascent method was utilised to select significant parameters and their ranges. Finally, the particle swarm optimisation (PSO) - back propagation (BP) - PSO algorithm was used to implement the parameter calibration of the piglet feed with arbitrary water-to-feed ratios. The results show: the stacking angles of the DEM of piglet feed are highly consistent with those from physical tests and the relative error is less than 1.01%. The proposed method shows better accuracy than response surface methodology (RSM).