Quality prediction for flotation column based on DEPSO and RBF

The cyclonic static micro-bubble column flotation (FCSMC) is a new type of mineral flotation device with complex internal mechanism. The existing empirical mechanism model, just applicable for the description of the micro behavior of flotation process, is inapplicable for prediction of the quality of flotation. Based on massive actual process data of flotation, RBF networks are adopted to describe the relationship between production conditions and flotation quality of FCSMC. The DEPSO hybrid algorithm combining of different evolution (DE) and particle swarm optimization (PSO) is proposed to optimize the parameters and architecture of RBF for optimal prediction model. The predict model is tested by practical data from a mineral processing plant, and simulation results show that the model converges fast with better prediction accuracy and generalization capacity.