Neural Networks and First Principle Models for Bioprocesses

摘要

This paper analyzes the combination of prior knowledge in the form of first principle models (parametric models) and neural networks. These models are called hybrid models. Neural networks and hybrid models were used to identify a fedbatchfermentation. Different neural networks were integrated into the hybrid model structure. The performance of these hybrid models is compared with "traditional" neural networks.