Integration of first-principle models and machine learning in a modeling framework: An application to flocculation

In this paper, an integrated hybrid modeling approach with first-principles is implemented to model flocculation processes. The application of the framework is demonstrated through a laboratory-scale flocculation case of silica particles in water. In this modeling framework, it is demonstrated that the integration of first-principles models and machine-learning approaches accurately predicts the dynamics of the system. The first-principles model used in this study incorporates population balance and mass balance models combined with the kinetic expressions of the agglomeration and breakage phenomena. The predictive abilities of such modeling framework is compared with a fully first-principles model, and moreover with a hybrid model that was developed in a prior work, which used a population balance model as the first principles model and a deep learning algorithm for the determination of the flocculation kinetic parameters.