Optimization of fluidized bed spray granulation process based on a multiphase hybrid model

摘要

An optimization strategy is presented to improve the granule quality of fluidized bed spray granulation in this paper. An accurate description of the relationship between the granule quality and the manipulated variables is first established as a prerequisite for optimization. Considering the characteristics of the fluidized bed spray granulation process and its multiphase features, a multiphase hybrid model is built with the population balance equations as the basic model, in combination with the use of black box models to obtain the unknown parameters. RBF neural network is adopted as the black modeling method in this work to describe the relationship between the parameters and the manipulated variables. Batch experiments of fluidized bed spray granulation are then used for hybrid model validation. The results show that the model has high accuracy and good generalization ability, and it can effectively reflect the characteristics of each phase of the granulation process. Based on the established multiphase hybrid model, an optimization problem and its corresponding online optimization strategy are proposed for fluidized bed spray granulation process. An improved differential evolution algorithm is used to solve the problem for online optimization and adjustment of the granulation process. Experimental results illustrate that both the hybrid model and the improved DE algorithm are effective when used in optimization of batch fluidized bed spray granulation process.