An efficient model development strategy for bioprocesses based on neural networks in macroscopic balances: Part II

There is a need for efficient modeling strategies which quickly lead to rel iable mathematical models that can be applied for design and optimization o f (bio)chemical processes. The serial gray box modeling strategy is potenti ally very efficient because no detailed knowledge is needed to construct th e white box part of the model and because covenient black box modeling tech niques like neural networks can be used for the black box part of the model . This paper shows for a typical biochemical conversion how the serial gray box modeling strategy can be applied efficiently to obtain a model with go od frequency extrapolation properties. Models with good frequency extrapola tion properties can be applied under dynamic conditions that were not prese nt during the identification experiments. For a given application domain of a model, this property can be used to considerably reduce the number of id entification experiments. The serial gray box modeling strategy is demonstr ated to be successful for the modeling of the enzymatic conversion of penic illin G In the concentration range of 10-100 mM and temperature range of 29 8-335 K. Frequency extrapolation is shown by using only constant temperatur es in the (batch) identification experiments, while the model can be used r eliable with varying temperatures during the (batch) validation experiments . No reliable frequency extrapolation properties could be obtained for a bl ack box model, and for a more knowledge-driven white box model reliable fre quency extrapolation properties could only be obtained by incorporating mor e knowledge in the model. (C) 1999 John Wiley & Sons, Inc.