Fig. 1: A modeling framework for hybrid mechanistic modeling and deep transfer learning. | Nature Communications

Fig. 1: A modeling framework for hybrid mechanistic modeling and deep transfer learning.

From: Scale-up of complex molecular reaction system by hybrid mechanistic modeling and deep transfer learning

Fig. 1

This figure outlines the process for developing a hybrid model. Initially, a molecular-level kinetic model for naphtha FCC is developed as the mechanistic model. The model is then used to generate molecular conversion data under various process conditions and feedstock types, which are employed to train a residual multi-layer perceptron (ResMLP) suitable for process scale-up of complex reaction systems. On this basis, small-sample pilot or industrial data are used to fine-tune partial neural network parameters, enabling accurate prediction of product distribution for a pilot or industrial-scale plant.

Back to article page